|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
BRAZIL' S ELEVATED INTEREST RATES: A CASE OF IRRATIONAL PESSIMISM
OR GUARDED OPTIMISM?
A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS
UNIVERSITY OF FLORIDA
THERE was a child went forth every day;
And the first object he look'd upon, that object he became;
And that object became part of him for the day, or a certain
part of the day, or for many years, or stretching cycles of years.
Walt Whitman, Leaves of Grass (1900)
This work is the product of my interactions with a great number of people over the
course of the last five years, and would not have been possible without their influence,
which was oftentimes unbeknownst to them. My time as a research assistant at the
Federal Reserve Board helped to me ground my undergraduate training in
macroeconomics and my research with David Lebow and Jeremy Rudd trained me to see
the truths apparent in large, obscure datasets. It was there that I was first exposed to
Phillips Curves, Taylor Rules, and the theory of rational expectations, all of which are
utilized throughout this paper.
While at Florida, Dr. Terry McCoy and Alex Sevilla have served as my advocates
in my desire to combine my disparate disciplines, Latin American studies and finance,
into a harmonious curriculum that has brought the two worlds much closer (at least for
me). The faculty and staff of the Center for Latin American Studies and the Florida
MBA Program have all greatly facilitated the process as well. Professionally, Eric Allred
and Kathy Fitzpatrick at Johnson and Johnson in Jacksonville helped me to contextualize
the Brazilian business environment better than any in-class or personal travel ever could
have and my experience working with them has completely shaped my perspective in
writing this paper.
Working with my thesis committee has been wonderful. Dr. Terry McCoy has
continued to work with me through the peaks and valleys of my research, more often than
not exhibiting more confidence in me than I had in myself. He also helped me flesh out
the topic after listening to one of my many anecdotes about my time living in Brazil. Dr.
Andy Naranj o has provided exceptional guidance in my time here at Florida, and has
made me question not just Brazil's macroeconomy, but also some questions more related
to my personal professional development. Time will tell if he is right. Dr. Libby Ginway
has graciously offered to serve on my committee as well, despite the topic being as
esoteric to her as Brazilian science fiction is to me; obrigado, Doutora.
No set of acknowledgements could ever be complete without extending gratitude to
those friends and family members who offer support, both financial and emotional,
through any undertaking. I am especially grateful to my mother and father in those
regards, as they have always supported me in my decisions, always confident that I was
taking the right, if less-traveled, path.
TABLE OF CONTENTS
ACKNOWLEDGMENT S ................. ................. iii........ ....
LIST OF TABLES ................ ..............vii .......... ....
LI ST OF FIGURE S ................. ................. viii............
AB STRAC T ................ .............. ix
1 INTRODUCTION .............. ...............1.....
2 AN ABBREVIATED HISTORY OF INTEREST RATES, INFLATION, AND
MONETARY POLICY IN BRAZIL ................. ...............9................
Theoretical Bases for Inflation ................. ...............14................
Post-W ar Promises .................. .... ........ .. ...............17.....
Military Rule and an (Overly) Managed Economy .............. ....... .............1
Civilian Rule and Attempted Stabilizations............... .............2
O Plano Real: Finally, a Plan that Works! ............. ...............25.....
3 BRAZIL' S CURRENT INTEREST RATE DYNAMICS ................... ...............2
Testing for rationality in price forecasts ......____ ..... ... .__ .....__ ..........3
Real Interest Rates: Comparisons and Determinants................ .............4
Apologists for Elevated Real Interest Rates ............... ...... ........ ..........4
Historical Factors and Current Realities of High Real Interest Rates ......................47
Existence and Causes of High Interest Rates ......____ ..... ... .__ ........__......5
4 O JEITO JURIDICO BRASILEIRO: COPING WITH THE PROBLEMS
RESULTING FROM HIGHLY ELEVATED INTEREST RATES ................... .......56
Firm-Level Financing Decisions .............. ...............56....
Impacts on Large Firms .............. ...............58....
Impacts on Smaller Firms .................. ....... ..............6
Implications of Interest Rates for Individuals ................ .............. ...............66
Mortgage Finance ............... ....... ...............67
Everyday Financing for Individuals............... ..............6
The Disparate Effects of High Interest Rates ................. ..............................72
5 CONCLUSIONS, RECENT OB SERVATIONS, AND FINAL REMARKS ............75
A Review of the Causes of High Interest Rates.................................. .............. .76
The Show Must Go On: Evolutions of Very "Brazilian" Credit Mechanisms ...........78
Epilogue: Recent Data and Trends .............. ...............80....
LIST OF REFERENCES ............_ ..... ..__ ...............82...
BIOGRAPHICAL SKETCH .............. ...............88....
LIST OF TABLES
2. 1 Annual Inflation ........._._. ._......_.. ...............11....
3.1 Rationality Tests in Inflation Expectations ........._.._.. ....._.. .......__. .......3
3.2 Comparative Nominal and Real Rates ...._.._.._ ........._.._......_ ..........4
4. 1 Source of Financing Constraints and Firm Capital ........._.._.. ....._.._ ................64
LIST OF FIGURES
1.1 Ex-Ante Real Interest Rates in Brazil ................. ...............6............ .
2. 1 Deficits and Public Consumption of Fiscal Debt ................. ................. ........ 12
2.2 Inflation Rates, Before, During, and After Hyperinflation and the Real Plan.............14
2.3 Federal Debt/GDP Ratio ........._ ....... __ ...............22..
3.1 Inflation's Expected and Realized Values ......___ ..... ..__. .. ...._ .......3
3.2 Year-over-year Inflation Rates and Rolling Volatility of Inflation ..........................37
3.3 Contribution of Variables to Expected Inflation ................. ................. ..........39
3.4 Comparable Nominal Short Term Interest Rates ................. ................. ..........41
3.5 Stylized Depiction of a "Bad Equillibrium" ............. ...............45.....
3.6 Recent Trends in SELIC and Inflation............... ...............48
3.7 Brazilian Balance of Payments ............ ......_ ...............52
4.1 Aracruz' s Debt Structure .............. ...............60....
5.1 Flowchart of Interest Rate Determinants ...._.._.._ ..... .._._. .... .._._.........7
Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Arts
BRAZIL' S ELEVATED INTEREST RATES: A CASE OF IRRATIONAL PESSIMISM
OR GUARDED OPTIMISM?
Chair: Terry L. McCoy
Cochair: Andy Naranjo
Major Department: Latin American Studies
This thesis examines interest rates in Brazil, with a focus on their historically
volatile nature, the theoretical and practical reasons for their elevated levels, and the ways
in which firms and individuals mitigate the negative effects of a high interest rate
environment. Brazil's macroeconomic history has been extremely volatile, and the
country experienced triple digit inflation rates as recently as 1993. The instability of
prices, as well as the existence of what has been called a "weak macro regime," has
necessitated interest rates which exceed those of peer nations. A cycle emerges, in which
exogenous shocks are propagated into the country, making it a more risky environment
for investments and raising the intrinsic risk associated with government debt, as default
is seen as a more likely possibility. The nature of this cycle, as well as other determining
factors of interest rates, is examined in this work.
In spite of these interest rates, Brazil has experienced some periods of very
impressive economic growth. The buffers between the deleterious effects of high interest
rates and the rates of growth exhibited in the country are also examined in this work.
Specifically, this thesis will answer how firms raise capital in this environment, and show
the effects of firm size and nationality on the methods of Einancing used. The effects of
high interest rates also trickle down to individuals, and decisions such as home Einance
and durable goods purchases, as well as more day-to-day Einancing of household
consumption. This thesis also explores the network of informal, and oftentimes
subsidized, credit that has emerged in the country.
In the light of the recent economic success of the country, investors from Brazil to
New York are calling for a lowering of the reference interest rates in the country. This
thesis blends previous theory, survey-based research, and anecdotal evidence from the
most recent data to illustrate the causal mechanism that would allow that lowering to take
place, and argues that interest rates should, in fact, be lower than they actually are.
For most, the word "Brazil" conjures up images of the Amazon, Samba, and
Calrnaval. The commotion and volatility that all of these images imply is not dissimilar
to the thoughts of economists on the country: rampant inflation, sky-high interest rates,
precarious commodity prices, and unpredictable economic growth. Brazil suffers greatly
from these preconceptions. In a reciprocally causal relationship, interest rates are high
because of the risk that doing business in the country implies; those high interest rates act
as a continual impediment to growth and (more importantly) to economic equality in the
region. Those impediments to equitable growth further fuel the level of risk that the
market now assesses to investments in Brazil. Without access to credit, the middle class
is stifled, inequality grows, and balanced growth is further out of reach. This paper will
focus on the cost of credit in Brazil, with a specific focus on the history of volatility that
has necessitated high interest rates, the theoretical and practical factors which influence
their determination, and the mechanisms which have evolved to help the country function
despite the inability to effectively utilize traditional debt Einance.
There exists no market for long-term local debt, not even for the most securitized
lending such as mortgages, and only the wealthy with access to overseas funding can
effectively leverage their Eixed assets. In addition, as interest rates remain high, a larger
percentage of government revenues continue to service the national debt, in essence
paying continually for the country's dismal economic past, instead of funding initiatives
aimed at preparing Brazil for the future. In fact, the government pays the greatest cost for
these high interest rates, as the government debt/GDP ratio remains very high, crowding
out private investments. This diversion of funds cannot be stopped: any talks of a
moratorium or of a restructuring of debt payments to allow for more social programs or
productive investments sends shocks through financial and currency markets, raising the
"Brazil risk" and making the debt grow instantaneously, in terms of reais, as a large
portion of the debt is denominated in dollars. This was evidenced most recently in
October of 2002, in the run-up to now President Lula's election: with talks of profligate
public spending and disregard of Brazil's obligations to the international financial
community, the Real hit an all-time low against the dollar of 4.03 on October 12, 2002.
Since that time, the government of Lula has pursued orthodox economic policies
not unlike those of his predecessor, Fernando Henrique Cardoso. The currency market
has adjusted accordingly, with the Real experiencing a 15% appreciation in 2005 and a
cumulative 51% appreciation since year-end 2002. Notwithstanding this impressive
adjustment in currency values, the overnight bank lending rate (hereafter the SELIC rate)
in Brazil remains at 18%, implying an ex-ante real interest rate of over 13%.2 At such
high levels of interest in the least risky of investments (overnight bank lending), it is easy
to see why no local long-term debt market has matured. Adding factors for default,
maturity, and liquidity risks, private lending rates quickly approach prohibitive levels.
Despite solid growth prospects and performance, interest rates seem "sticky" and
unbalanced growth continues, j eopardizing the future of the country.
Reais is the Portuguese-language plural of the Brazilian currency, the Real.
SAt year-end 2005 the expected change in the price level in the following twelve months was 4.76%, as
measured by the Banco Central do Brasil and published on the Bank's website. The real interest rate is
simply the nominal rate less expected inflation.
This vicious cycle is extremely unfortunate, as the Brazilian economy has been
performing very well in recent years. Growth in real GDP was 4.9% in 2004 and is
expected to be 2.44% and 3.45% in 2005 and 2006, respectively (Fundagdo Getuilio
Vargas and Latin-Focus magazine's consensus estimates). These rates of growth are all
above the average growth of 2.0% for the twenty year period ending in 2003. Inflation
has remained essentially in check since the institution of the Real in 1994 (more on this
below) and productivity growth continues to accelerate as synergies resulting from
investment in the real sector and international economic liberalization are being realized.3
In short, it appears that the "pais do future" has finally arrived into its economic
"promised land" and that the country is performing as the economic powerhouse that
many have believed it could be for so many years.
In order for this transition to economic powerhouse to occur, many things must
take place simultaneously. Unemployment must be low to foster economic opportunity.
In Brazil the official statistics for unemployment hover around 10%, with levels of
underemployment higher and the employment index for the informal economy mitigating
the poverty effect from this official unemployment (according to data from various
editions of the monthly employment survey from Brazil's national statistical agency,
IBGE4). Wealth must be fairly distributed to kindle a middle class capable of
entrepreneurship (the true engine of growth); in Brazil income inequality is worse than
anywhere else in Latin America and on par with some countries in Africa (Bulmer-
3 Productivity is difficult to measure in Brazil, and is not officially published by the government. I rely on
the work of Ferreira and Rossi (2003) for this generalization.
4 The survey is the Pesquisa Mensal do Emprego and the agency is the Instituto Brasileiro de Estatistica e
Thomas, 2003, p. 11). Macroeconomic fundamentals such as inflation must stay low and
relatively nonvolatile in order to provide appropriate signals to markets. Another key
component of economic growth is the interest rate that affects the decisions made by
governments, banks, and non-financial firms.
Standard economic models would suggest that the interest rates would be declining
over time in Brazil. In the market based approach to interest rates, this works through
two channels. First, expectations of future inflation directly influence the nominal
interest rate: i, r, + E(Fr, ,); that is the nominal rate today is approximately equal to
today's real rate plus the expectation of inflation in the following period (more formally,
the expected annual inflation over the term of the interest rate).5 Another channel
through which good economic performance should lower interest rates is through the
country risk premium. Stylistically this is the rate at which the market assesses a
premium to transactions in a country (and in a countries currency) vis-a-vis a base
country of reference.
These two mechanisms should be acting to lower interest rates in Brazil.
Specifically, the expectations of future inflation should be diminishing as recent inflation
performance has been below the historical average and much below the hyperinflationary
levels which ended with the Real Plan in 1994. The government now operates in a new
regime of inflation targeting. This regime has performed well, is transparent, politically
independent, and operates with a legal mandate to control inflation (though it does not
operate in an apolitical vacuum). As agents update their frameworks for evaluating
inflationary expectations, the nominal interest rates should decline. Furthermore, solid
5 The same is true when monetary policy is viewed through the Taylor Rule, where expectations of future
inflation are positively correlated with interest rate moves by the central bank.
economic fundamentals will themselves drive non-inflationary growth. In Brazil, some
decline in expectations has occurred and later in the paper the formation of expectations
and "inflationary memory" will be discussed. Specifically the prospects of an
asymmetric response function to inflationary news will be evaluated: does the market
take longer to believe good news than it does to incorporate (and penalize) bad news?
Most important, however, to the formation of inflationary expectations is the credibility
given to the monetary policy of the government. Brazil now operates in a completely
new regime from that of just a few years ago. Transparent inflation targeting in the
confines of a politically independent central bank such as that done in Brazil is the "state
of the art" in monetary policy, indeed more theoretically advanced than the regime of
even the U.S. Fed, which cloaks its policy decisions in obfuscating language, and
represents a clear break from the political nature of a loosely defined monetary policy in
Brazil's past. If this new regime is believed to be the regime going forward, expectations
should not factor in the policy mistakes of previous regimes. Later this paper will
address this concern as well.
The so-called country risk premium should also be falling. Brazil has returned to a
peaceful democracy since the end of the military dictatorships, capital markets are fully
developed and fairly liquid (compared to historical norms), and capital moves freely
across the border. Even the leftist politicians of years ago are pursuing orthodox
economic models and are allowing for increased autonomy of the nation' s central bank.
In short, conditions are such that inflationary expectations should not be anchored in the
hyperinflationary or even high-inflation period, and the country risk premium should be
converging to a more reasonable level. Furthermore, the ex-post change in the currency
has implied a currency premium, and not a discount, for investing in reais. Despite the
seemingly convincing case for lower real-interest rates, the precipitous decline has not
occurred. As seen in the chart below, the real rate has been in a band between 10-18%
since 2001, and is in the mid-point of that band now.6 This paper will examine reasons
for interest rate persistence, Figure 1.1 Ex-Ante Real Interest Rates in Brazil
Ex-Ante Real Interest Rates
examine the path necessary to dnive i
the rates lower, and to examine
mechanisms which are already in 12
place to mitigate the impact that s
interest rates currently have on 8 8 8 8 8 8 B B B B B
consumers and firms.
There exists a robust literature on the subj ect of Brazil's macroeconomic
fundamentals, and much work has previously been performed on the question of the
country's interest rates. Much of the work examines one component of the interest rate,
focusing on inflationary expectations or on one of the drivers of the real interest rate.
Several papers seek to justify Brazil's interest rate policy, and estimate Taylor rules for
Brazil. Most notably, Muinhos and Nakane (2005) emply various techniques to show
that Brazil's interest rates are higher than those of other economies, while Bogdankski et
al. (2001) estimate monetary policy rules for Brazil consistent with the level of interest
rates currently present in the country. In an important contribution to the literature,
Favero and Giavazzi (2002) estimate a full Taylor rule for Brazil that shows that the
6 Defined as the SELIC rate less expected change in the IPC index over the subsequent twelve months
(BCB data, author's calculation and graphic).
interest rate associated with non-accelerating inflation is around 17.5%. The implications
of their methods and results are discussed fully in chapter three.
Building on the theoretical work of Keane and Runkle (1990, 1998), Carvalho
(2004) builds a complete set of tests of the rationality of the forecasters that are surveyed
by the Central Bank on the future movements of inflation. Her work largely rej ects
rationality in the data, and similar tests (with more recent data and slightly different
specifications) are reproduced in the third chapter of this work. Carvalho stops short of
identifying the driving forces of expectations, focusing on their rationality. A recent
contribution from Cerisola and Gelos (2005) discusses how those expectations are
formed, and suggests that the relative importance of the various determinants changes
over time. This work is expanded upon in this paper, in light of the most recent
The question of real interest rates has also been addressed by several authors.
Favero and Giavazzi (2001) point to term premia, currency aversion and the lack of
independence of the central bank. The last point is also identified as a cause of inflation
by Tullio and Ronci (1996), who estimate a Phillips curve with a proxy for the perceived
independence of the bank. This estimation clearly shows a correlation between weak
political independence and higher inflation. Palley (2004) and others have written on the
weak fiscal situation of the federal government and how high interest rates are both a
cause and an effect of that deficit Einancing. Their work is examined and critiqued in the
second half of chapter three. Carneiro and Wu (2004) point to Brazil's insertion into the
global economy as a proximate cause of its high interest rates, a point also examined by
Garcia and Didier (2000). Finally, Arida and Lara-Resende (2005) also point to the lack
of local long-term debt as a reason for (and not an effect of) high interest rates.
Less work has been done on the effect of high interest rates in the country.
Bonomo et al.(2003) empirically verifies credit constraints at the firm level, and Terra
(2003) shows that the size and nationality of the firm is a determining factor of the
severity of that constraint. This paper will examine the capital structure of firms,
principally using published annual reports (in the case of large firms) and survey data (in
the case of smaller firms) from the World Bank (2003). Some research has been done
with respect to credit constraints at the individual level, and this paper will make use of a
recent survey (Brusky and Fortuna, 2002) to discuss the unmet needs of the masses with
respect to everyday access to credit and capital. Fagnani (1996) and Borba (2004)
discuss the lack of mortgage finance and suggest ways in which that gap could be
eliminated. Finally, Pinheiro and Cabral (1999) show the major aspects of Brazil's
"gray" market for credit which allow it to flourish.
This paper will act as a compendium of all of this research, bringing the previous
authors' disparate viewpoints together to tell, in one voice, the story of Brazil's interest
rates. The most recent data will be considered as well, updating some results that were
written several years ago, while conj ecturing either that other results are no longer valid
or that the credit markets have not yet responded to a shift in the fundamentals of the
Brazilian economy. In doing so, the paper seeks to answer whether Brazil's credit
markets are suffering from irrational pessimism, routed in the past may not repeat itself,
or if they are subj ect to a guarded form of optimism that will require more economic
austerity and good performance to result in lower interest rates.
AN ABBREVIATED HISTORY OF INTEREST RATES, INFLATION, AND
MONETARY POLICY IN BRAZIL
Brazil's history of inflation, interest, and monetary policy will be extremely
different for the reader solely familiar with the stable regime of the United States or the
other more industrialized nations. It does not, however, differ drastically from other
Latin American nations. The government' s use of the inflationary taxes to finance the
public deficit are not uncommon to the region, nor are rates of inflation that are
meaningless in annual terms and are calculated and discussed in monthly rates, rates
which, in turn, are factors of 10 above the average annual inflation rate of the most stable
post-industrialized economies. To help the reader to better understand this process and
its causes, this paper will review this history in order to gain insight into the causes of this
interest rate dilemma. Indeed, one cannot study the history of interest rates per se in an
economy such as Brazil's. The series is highly discontinuous, variable, and ill-defined
(even by the Central Bank). This is due almost entirely to the unpredictable nature of
inflation, along with the regime changes experienced in the country over the last fifty
The study and history of inflation is integral, then, to the study of interest rates in
any country. Individual investors demand rates of return which exceed the increase in the
price level; if those rates of return cannot be met, agents have strong incentives to
consume income as it is earned, thereby lowering the savings rate.l Empirical evidence
shows that the Taylor rule also holds in emerging economies as well as industrialized
economies; as expectations of future inflation become worrisome, central banks use their
policy instruments, namely interest rates, to quell inflationary pressures.2 It is because of
these linkages that this study begins with an examination of the history of inflation in
Brazil. Special attention will be paid to the role of government in seeking to tame
inflation's course so as to steer the economy towards low-inflation economic growth and
also to the various monetary policy regimes which have sought to control the rate of price
In the post-war period, inflation began at a fairly stable pace, fluctuating in a
range below 20% annually until late 1959 (Bulmer-Thomas, 1993).3 Even the low
inflation of this period, however, was sparked by government policies and manipulation.
By 1945 the Vargas administration had increased the tax share of GDP to 26.5%, an
astonishing three-fold increase from the level just six years prior ibidd). A dual exchange
rate regime also existed in this period which gave preference to exporters. This regime
may have promoted export growth but had inflationary pass-through consequences
because of the high exchange rate granted to importers. This sort of indirect taxing of
consumer goods began a long standing tradition of government manipulation, albeit
indirect, of the price level ibidd). However, beginning in the 1960s, inflation quickly got
SWhen given the choice to spend R$1 today on an asset, or invest at 10% to purchase the asset in one year
at R$1.15, a rational actor would spend the R$1 today (even if he/she were to hold the good and sell it in a
2 See, for instance, John B. Taylor, "Using Monetary Policy Rules in Emerging Market Economies," a
paper presented at the 75" Anniversary Conference of the Bank of Mexico, November 14 15, 2000
(available on-line at httpl u\ \\ l\banxico.gob.mx).
3 As in the United States, reliable, official economic data does not exist for the pre-World War II period.
out of control and became rooted in the Brazilian psyche, having long-lasting
implications for the conduct of economic policy.
In that decade, inflation exceeded 40% per year and Table 2.1 Annual Inflation
grew to exceed 100% annually by 1980. By the Inflation (percent)
introduction of the most recent stabilization plan, the Real 1970-74 20.9
Plan, inflation was running at unfathomable levels (table 1980-84 148.0
at right, author's calculations). 1990-93 1,764.9
The acceleration of inflation in the period beginning in 1960 was caused in part
by the lack of a market for government debt. In fact, outside of the Banco Nacional de
Desenvolvimento Econdmico e Social (BNDES), there were few resources available for
long-term Einancing. Housing Einance was not available (it remains largely inaccessible
to this day) and any long-term credit was restricted to a few commercial banks and
Einance companies. At the time, Brazil was governed by a military government (more on
this below), and its response was fairly orthodox: remove the pressures on the deficit that
was causing the monetization of the deficit. Prices for public services were increased
(though this has some pass-through to inflation in other sectors) and expenditures were
cut (Coes, 1994, p. 87). Inflation fell to below 40% a year by the late 1960s, all things
considered a great accomplishment but not low enough to engender the sort of long-term
lending the country needed for development purposes.
To that end, the government instituted its second, much less orthodox, response: the
correCdo monetaria literally "monetary correction" and now known as inflation
indexation. The BNDES had long been indexing its long-term lending to offset the
inflation-induced real erosion of loan repayment. Then Minister of the Economy Roberto
Campos was a former president of the BNDES and suggested that indexation be used in a
broader sense in the economy.4 With the passage of the now-famous law 4.3 57 in 1964,
financial contracts could be written to allow for ex-post revisions in their notional
amounts to reflect inflation. The legislation also introduced a new type of government
bond, the Oly1 igaz Isti Reajusta~vel de Tesouro Nacional (ORTN), which increased the
government' s ability to finance its deficit through selling new government bonds to the
public, bonds which were indexed to a pre-determined index of the cost of living.'
Figure 2. 1 Deficits and Public Consumption of Fiscal Debt
According to Coes (1995),
S2,400 Fiscal Delicits and Public Consumption of Government Debt
(millions of current cruzeiros)
public absorption of this 1,900
Public Purchases of
new debt was one of the 1,400 Gort. Debt 197
key factors in bringing M~scal Defieit
down inflation after 1964
(seegrah).1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 19
Indexation became a central feature of the economy, and following the ORTNs, the
government began to track Unidadesdddddddd~~~~~~~~ Patries de Capital, or UPCs, which allowed for the
indexation of all debt, private and public, to the government-set levels for inflation
adjustment. In addition to the ORTNs, the government also began to issue Letra~s de
Tesouro Nacional, or LTNs, which were offered at various maturities and at market-
determined interest rates. All of these new assets helped to create a more liquid financial
market and made the more narrow monetary aggregates less important, as Brazilian
SBNDES, 50 Tears ofDevelopment. On the web at http://www.bndes.gov.br/english/book50year~s
SThis was a relatively progressive concept. As a point of reference, the U.S. Treasury first introduced its
inflation-protected securities, the TIPS series, in 1997.
agents could move between inflation-adjusted and nominal currency with ease, thus
blunting the monetary erosion to accumulated wealth from the inflation experienced in
However, outside of the financial sector notional correction was much less
advanced. Salary contracts were often negotiated on annual bases and even those
contracts which were indexed were corrected at a staggered pace (every six or twelve
months). The government itself contributed to this lag: it wasn't until 1979 (fifteen years
after the beginning of indexation) that the minimum wage was moved from an annual to a
semi-annual cost of living adjustment--with the variability that inflation exhibited in this
period, even this semi-annual adjustment left much to be desired (Coes, 1995, p. 89)!
Throughout the history of indexation, the government continued to manipulate the
economy through its policy choices. Take for instance the choice of price index used to
correct for inflation: a mining company would like its workers' wages (or the price of any
of its inputs, for that matter) to be corrected by the price change in the factory's output
(say, iron ore), while workers would like their incomes to be adjusted vis-a-vis changes in
the price of the basket of goods they consume. The government is essentially torn
between these competing pressures and any choice of index merely manages the
distortion in real prices that occurs without any indexation; it does not, however,
completely solve the problem. When oil prices sky-rocketed in the 1970s, the
government introduced a new price index that could be manually adjusted to control for
exogenous shocks beyond the control of the national economy. The result of this
manipulation was a real erosion of all assets corrected by the ORTNs, and reduced
purchasing power of the adjusted cruzeiro.6 Inflation continued to be high (despite
Figure 2.2 Inflation Rates, Before, During, and After Hyperinflation and the Real Plan
Annual Inflation 0%Annual Intlation
120 (12-month percent change in Imphicit price deflator) (1-month percent change in implicit price deflator)
100 1\. 70% i Real Plan
July 1, 1994
20, 10% \nr
Feb- Feb- Feb- Feb- Feb- Feb- Feb- Feb- Feb- Feb- Jn8 a-7 Jn9 a-5 Jn9
45 49 53 57 61 65 69 73 77 81
indexation) throughout the 1970s and 1980s and then accelerated to astronomical levels
in the late 1980s and early 1990s. Only with the introduction of the Real Plan was it
slowed to manageable levels. The charts above tell the story better than words possibly
can (note the scales): inflation so high, volatile, and endemic that by the mid-1980s it no
longer made sense to refer to inflation in the year-over-year or annualized sense, but
rather by 1-month percent changes (author' s exhibits from data collected from the BCB).
Despite various stabilization plans (as discussed below), it wasn't until the introduction
of the Real Plan in 1994 that the problem of inflation was finally tackled.
Theoretical Bases for Inflation
Before examining in more detail each of the relevant periods of recent Brazilian
economic history, it is useful to examine the theory behind the inflation that has plagued
the country and clouded the ability of policymakers to set a clear path for the future.
Specifically, what does theory tell us about these rampant (and variable) accelerations in
inflation? According to the classic model, the root cause of inflation all across Latin
6 Essentially the government would decide how much of the inflation was e \ogen oui" to the country and
only allow the indexation to account for the remainder.
America during this period was the same: monetization of Eiscal deficits. Despite the fact
that external Einancing was important to Latin America's growth and economic planning
in this period, these high rates of inflation also suggest that much of the debt was in fact
monetized. The Central Bank prints money in order to finance the balance between its
deficits and new debt issuance (extemal and internal) in order to pay the government' s
bills. The so-called inflation tax arises when the government prints too much money, its
real value declines, and the new, devalued currency is used to pay nominal balances owed
to the government' s lenders. In this monetary model, increases in the money supply are
essentially treated as exogenous, meaning they are predetermined and not affected by
other variables such as the change in the price level. If this were true, growth in the
money supply the infamous printing of money that people oft blame for hyperinflation
in Latin America would lead to a devalued currency. The policy prescription for
stopping inflation would be clear: turn off the printing presses!
Treating the money-printing policy decisions of the central bank as exogenous,
however, does not entirely apply to the Brazilian case. If budget deficits cannot be fully
financed through external funding, the bank has no choice but to increase the money
supply. As Fisher and Easterly argue:
Milton Friedman's famous statement that inflation is always and everywhere a
monetary phenomenon is correct. However, governments do not print money at a
rapid rate out of the clear blue sky. They generally print money to cover their
budget deficit. Rapid money growth is conceivable without an underlying fiscal
imbalance, but it is unlikely. This rapid inflation is almost always a fiscal
phenomenon. (Fischer and Easterly, 1993, pp. 138-39.)
The true variables of interest then, the ones driving inflation, would be the size of
the budget deficit and the appetite of the investing community for govemnment-issued
debt. Furthermore, critics of the monetarist theory in Brazil note that inflation sometimes
spiked at times when there was no substantial increase in the monetary aggregate, and
have pointed out instability in the estimates of monetary demand in Brazil. They cite
these two facts as evidence that the monetarist model is not appropriate for the Brazilian
case.' Arguing for a structuralist approach to inflation in Brazil, opponents of
monetarism say that the money supply is in fact endogenous, determined by other
variables, not the least of which is the price level. According to this theory, supply
constraints or excess demand causes inflation in specific prices. Seeing their purchasing
power eroded, other agents in the economy also raise their prices. The central bank then
is faced with a dilemma: restrict the monetary supply and foster deflation, or
accommodate the price increases with increased money so as to not engender
unemployment and/or recession. In short, str-ucturalists argue money growth is the effect,
and not the cause, of inflation.
Which of these theories is right? It is possible that there exists a feedback
mechanism wherein initial price increases prompt monetary accommodation, which in
turn brings about more inflation. Comes (1995) performs Granger causality tests on the
general price index (IGP-DI) and various monetary aggregates (the monetary base, Ml,
M2, and M3). The results do not indicate that inflation causes increases in the broad
monetary aggregates. For the more narrowly defined monetary aggregates, the tests are
on the border line of significance (in probabilistic terms) and suggest some relationship
may be present. In the case of the reverse relationship, the test results vary. For broader
monetary aggregates (M2 and M3, which include highly liquid monetary substitutes), the
SSee, for instance, Cardoso (1983), Rossi (1989), and Cardoso and Fishlow (1990).
SThe test used is actually a variant of the Granger causality test which eliminates potential bias from serial
correlation in the data.
tests indicate that changes in money do induce inflation. In short, for broadly defined
monetary aggregates, the monetarists' argument appears to hold water. Comes asserts this
distinction is due to the fact that during inflationary times (and especially after indexation
began), the increasing liquidity of near-money assets implies that it is these measures
which should be studied. Recent work (Feliz and Welch, 1992) indicates a neo-
monetarist model is most appropriate for Brazil: prices respond to both contemporaneous
and expected future changes in the money supply. Just the same, the failure to
overwhelmingly rej ect the structuralist argument in the Brazilian data may indicate that
there is some feedback mechanism which says that inflation may further fuel
accommodative monetary policy.
Regardless of the causes or the exact transmission mechanism, inflation remained
high through various administrations and economic plans. Without knowing the exact
transmission mechanism, but knowing the correlation between money supply, inflation,
and growth, one should by now be asking a very relevant question: how did so many
governments fail to manage inflation and control the country's economic situation more
effectively? Government policies had a direct impact on accelerating prices from the
1950s to the implementation of the Real Plan. Simultaneously, stabilization programs
were put in place to mitigate the negative consequences of the government' s aggressive
(and oft unrealistic) development goals. Were those policies simply ineffective, ill-
conceived, or were they subj ect to extraordinary exogeneities that rendered them futile?
Brazil emerged from World War II in a better position than many of the
developed economies. War had generated a demand for the country's exports, and the
United States had declared Brazil's coffee an importing priority during the war. As a
result, at the end of the war the national coffers were full: more than half a billion dollars
lay in the national reserves, and a positive trade surplus contributed much to that boon.
This good fortune quickly changed course, however, as that surplus was consumed by
aggressive development plans by the new administrations. Promising "fifty years of
progress in five," Juscelino Kubitschek began massive development proj ects, including
the initiative to develop Brazil's interior and to create a new capital, Brasilia. There were
some success stories which arose from his development plans: during his administration
real GDP grew at an average of almost 7 percent per year, leading to a per capital
increase of about 4 percent (FGV and BCB). The decrepitude of these plans, however,
was the government's inability to finance them: while Vargas had increased the tax share
to around 27% of GDP, this alone was insufficient to finance these aggressive plans.
Furthermore, export revenues were stagnating, only partially offset by foreign capital
flows (Coes, 1995). The government was left with almost no choice: they monetized a
great portion of this deficit. Not surprisingly, inflation ensued, with the first bout spiking
in 1959 and the annual inflation rate exceeding 30 percent. The successive
administrations of Quadros and Goulart, from 1961 to 1964 imposed some conservatism,
but it was too little, too late. Opponents of government manipulation favored greater
reliance on market prices and feared Goulart' s seemingly dictatorial aspirations.
Furthermore, inflation had taken root and the fiscal deficit proved harder to eradicate than
one would have hoped; when the military ousted Goulart in 1964, the inflation
accumulated in the twelve months prior exceeded 90 percent (as measured in the IGP;
author' s calculations). This paper does not posit that the principal reason for Goulart' s
removal was economic. His attempts at stabilization ranged from highly orthodox to
completely contrarian with respect to economic theory, embracing IMF lending
originally, then rej ecting it for a more "radical nationalist" strategy (Skidmore, p. 13).
This move to a more nationalist strategy wasn't just economic it sought both a political
and economic removal from the world scene, simultaneous with a move to the left of the
political spectrum. Goulart asserted a causal relationship between the economy's woes
and its insertion in the global economy, as well as domestic policies. It was this tumn to
the left that alienated many factions within Brazilian society and prompted the formation
of a shadow government consisting of various segments of society (Ibid., p. 14). The
inability to control inflation, then, was symptomatic of serious division in the government
and in society that was crippling with respect to any unified policy; its rampant nature
helped to usher in a defining moment in Brazilian history: military rule that would last for
Military Rule and an (Overly) Managed Economy
The military imposed stringent measures to reign in inflation. Fiscal strategy
included the raising of taxes, changes in the pricing strategy of public utilities and
services, control of wages and credit, and announced targets for monetary expansion.
The government' s profligate spending was not corrected, however, as government' s share
of GDP remained at historical levels. The idea was simple: cover the existing spending
with more tax and service revenue, but don't decrease spending! Government spending
as a share of GDP remained essentially constant throughout the period (at around 10%,
according to statistics from IBGE, 1991). However, the revenue base of government did
expand and covered a greater share of these expenditures, allowing inflation to cool in the
late-1960s. However, excessive government management of the economy became a
staple: policies allocated resources instead of prices, and those resources were allocated
based on the government' s agenda. Indexation further complicated pricing, and sped the
transmission of shocks between sectors of the economy (since the broadest inflation
measure, the IGP, was used as the basis of indexation).
A new administration, headed by Costa e Silva, took the reins in 1968 and
exacerbated these difficulties by allowing for more inflation, as it viewed inflation as a
trade-off with real GDP growth (Coes, 1993, p. 15). Price level management was also
institutionalized, in bureaucratic Brazilian fashion, with the creation of the Conselho
Interministerial de Prepos (CIP), and wage adjustment was changed to reflect actual
inflation in an ex-post fashion. During this administration, real output grew at over 10
percent per annum, while inflation remained around 20 percent and the public deficit was
brought to below 1% of GDP for obvious reasons this period in Brazil's economic
history is known as the milagre.
Economic sensation quickly turned into cessation with the first oil shock in 1973.
At the time, Brazil was highly dependent on foreign oil (80% of the crude consumed was
imported), and was especially impacted by the shock (Coes, 1995). The new
government' s management of the crisis was commendable, however: with Brazil's
economic management institutions, only about one-half of the price increase was passed
on to consumers, while subsidies covered the rest. Through both monetary and fiscal
restraint, the inflationary threat posed by the oil price increases was quelled. The shock
to consumers was mitigated, but the effects on government remained deleterious: some
increased borrowing to cover the subsidies, a negative impact on the trade balance, and a
reduction in Brazil's international reserves.
Brazil's government decided to launch a second national development plan in the
period following the shock, and greatly encouraged importation of intermediate and
capital goods to that end. This was prompted by the level of world interest rates: real
rates on dollar-denominated loans were zero or negative throughout much of the 1970s
(Coes, 1995, p. 19). Expansive monetary policy led to soaring inflation, however, and at
the time the world economy was experiencing its first bout with stagflation (the
phenomenon of simultaneously high inflation and low output growth), which limited the
ability of Brazilians to export abroad. This led to a sharp increase in real interest rates
over the period from 1977-1978 that aimed at cooling the overheating economy. By the
time of the second oil crisis, inflation was still excessively high and the current account
remained in significant deficit.
The response of the government to the second oil shock was typically
bureaucratic: strict controls on nominal interest rates, coupled with subsidized credit for
certain sectors of the economy. Promulgations abounded, from pre-announced
devaluations, to targets for correCdo monetaria and expansion of the monetary base. The
intent was clear: stop the inertia in inflationary expectations. The policies lacked
credibility however, and after a year of record-setting inflation (1 10% in 1980), the
government switched its mechanism of management to that of demand control, including
more aggressive management of the monetary base. The government at this time also
began the sale of exchange-rate indexed bonds to the Brazilian public. Due to the
inordinately high levels of expected devaluations, the availability of these bonds led to an
increase in real interest rates: there was no longer a transactions cost for hedging
domestic currency exposure because the government would adjust returns vis-a-vis
With domestic interest rates so high, foreign borrowing skyrocketed and by 1982,
interest payments accounted for over 70% of the current account deficit and 4.2% of
GDP (IBGE). Public enterprises also took part in the foreign debt markets, borrowing
abroad to Einance the external debt of the central bank, thereby increasing reliance of the
government on foreign funding. While an increase in borrowing abroad should force
down the domestic interest rate, in Brazil that did not occur. The reason for this
theoretical result is that Brazil was backed into the constraint imposed by high foreign
debt. With the stock of external Figure 2.3 Federal Debt/GDP Ratio
35 Federal DeMI/ GDP Ratio
federal debt at the time in excess of (p percent)
Eifteen percent of GDP (see graph), 25
any marginal increases in the debt 1s
level further decreased the country's t
credt sandngincrase th drin 1964 1967 1970 1973 1976 1979 1982 1985
from debt servicing, and further exposed the country to the whims of international
sentiment and exogeneities. These factors served to increase the country risk element of
the real interest rate (as measured and detailed below) and counteracted any monetary
expansion from the foreign borrowing.
As if the situation wasn't bad enough already, in late 1982 another exogenous
shock rattled the now-fragile Brazilian economy, as Einancial contagion from the
Mexican debt crisis affected the country's cost of capital and, for that matter, the
availability of foreign lending. Though Baer and Coes (1990) argue that the lending
policies had begun to disfavor Latin America before this date, this crisis increased the
perceived "country risk" even further. Political (election-related) pressures prevented the
country from borrowing from the IlVF (exacerbating its cost of capital).9 When lending
finally came from the IlVF (with the usual strings attached), its success, in economic
terms, was so painful that the government lost much political support (per capital real
income had fallen by more than 10 percent cumulatively in the years 1980-1983). This
loss of support sped the transition to civilian rule and helped to usher in the government
Civilian Rule and Attempted Stabilizations
Sarney's attempts to stabilize the economy were various and categorically
unsuccessful. His first plan, the Plano Cruzado, began in 1986 and included price and
wage freezes, along with the prohibition of indexation and a new currency, the cruzado,
worth 1,000 cruzeiros. The basic aim was to eliminate inertial inflationary expectations
by ending the indexation and taking the cruzeiro out of public memory. The plan worked
initially, but after a few months some allocational effects became obvious. The price
freezes had caught some sectors off-guard, and relative prices became skewed, resulting
in shortages and rationing. In an attempt to salvage the plan, the Plano Cruzado II was
announced, adjusting the prices of a gamut of goods, including automobiles, government
services and agricultural products. A floating peg of the cruzado was announced, as was
a return to indexation (though this time the nominal anchor was to be the return of BCB
notes). In the international community, the risk level associated with the country reached
9 Bacha (1983) estimated the effect of this decision at around 320 million dollars in extra interest to be
"' Tancredo Neves was elected president but died before assuming office. Upon his death, his Vice-
Presidential candidate, Jos6 Sarney, became President-elect.
new highs when the government announced it would temporarily suspend interest
payments to foreign creditors in February, 1987. These measures eroded any and all
confidence the government had gained; inflation returned to levels near 1,000 percent per
year just eighteen months after the introduction of the plan (IBGE).
A new plan followed, named after the Economy Minister who devised it, Luis
Bresser. Indexation remained, the currency was allowed to devalue, but wages and prices
were frozen for a period of three months (only after allowing government and agricultural
prices to "catch up"). Real interest rates were allowed to rise, and the government
announced that its deficit would not exceed 3.5% of GDP. This particular part of the
plan, however, was derailed when the President authorized large wage increases for
public sector employees after the initial three-month freeze. After that initial three month
period, the inflation rate again approached 15% monthly and Bresser resigned.
A new plan, O Planzo Verdo, or the Summer Plan, was announced on January 14,
1989. Real interest rates were elevated and cuts in public spending were announced.
Prices were again frozen and the economy was de-indexed with a new currency unit, the
novo cruzado, becoming the unit of measure (again, dropping three zeroes from the
original cruzado). Two factors led to this plans demise: a nominal peg of 1:1 with the
dollar was announced and public spending also spiked ahead of the presidential election,
wearing away the credibility in the plan's announced tenants. Inflation hit four and six
percent in February and March (despite the price controls), and the exchange rate could
not be maintained at 1:1. By the end of the election, which chose Collor de Mello as its
leader, inflation surged to over 80% monthly (Coes, 1995).
The Collor Plan was immediately announced by the new administration, and was
the most drastic plan yet. It included a sharp fiscal tightening and a sudden freeze in
nearly 70% of the country's financial assets. However, there were exceptions to this
freeze and those types of assets (principally short term bonds) allowed liquidity to again
expand and inflation again ensued."l A new package, Plano Collor II, sought to re-
administer the price level through a "reference interest rate" that served to re-index the
economy and allow for more of the inertial inflation previously experienced. These plans
were ultimately unsuccessful as well, as they failed to attack the principal causes of
inflation: large federal deficits, deficits which were, in turn, fueled by rampant corr-uption
which later forced Collor de Mello from office in 1992.
O Plano Real: Finally, a Plan that Works!
In the two year administration that followed, Franco appointed two politicians and
an academic economist to the post of Finance Minister. Unsatisfied with their
performance, he moved sociologist Fernando Henrique Cardoso to the Finance Ministry
in May of 1993. Cardoso and three academic economists authored the stabilization plan
that is still in place as of the date of this writing, O Plano Real. This plan incorporated
the lessons learned from all the previous stabilization attempts and improved upon them.
One of these authors described it as a three-pronged strategy to stabilize the economy: a
fiscal strategy centered around a constitutional amendment, monetary reform taking place
with the slow adoption of a new currency, and an opening of Brazil's economy, both in
the form of continued privatization, a new foreign exchange policy, and a liberalization
The quantity theory would suggest this result: it is the growth rate of the money supply, and not the
stock, which matters for inflation. According to the theory, a sudden, one period drop in the stock of
money, followed by growth would generate one-period deflation, followed by re-inflation of the economy.
of the capital account (Franco, 1995). The fiscal deficit was addressed immediately with
the introduction of new taxes of Einancial transactions, cuts in federal government
spending, and the creation of a new Fundo de EvaN/~ii.4l tio Fiscal. As a result, by mid-
1994 there was a primary fiscal surplus of 2.6% of GDP. The currency adjustment and
deindexation of the economy came in two pieces: first the creation of new thridadeddddddddd~~~~~~~~ Real
de Valor (URV), which was a composite of several price indices and which emerged as
the new index rate. However, with cuts in federal spending and a credible economic
agenda, the URV found (as it was designed to do) a point at which the economy's prices
stabilized. Once that stabilization had occurred, the Real was introduced as the new
currency. In terms of the foreign exchange component to the plan, the value of the Real
was allowed to fluctuate in what Cinquetti (1998) has called a "upper-bounded floating
exchange rate regime"--the cost of the Real was to never exceed one dollar, but was
allowed to fluctuate downward. The success of the plan has been credited to many
things, including some exogeneities. However, it was the first of the plans to allow
market mechanisms to settle within a new economic framework (no price/wage freezes)
and slowly removed indexation instead of a sudden, incredible regime change.
Moreover, it also addressed the fundamental structural issue, the federal deficit. With the
government constrained to a balanced budget and new sources of revenue, the Central
Bank was no longer forced to monetize the deficit, which would have had the usual effect
of fueling inflation.
As mentioned above, the Real Plan continues in place today and has guided the
Brazilian economy through a period of relative prosperity. It hasn't gone completely
unchanged, though, and currently we are in a period which could be known as the "Planzo
Realll"' if one were to follow the semantics of previous stabilization attempts. In early
1999 Brazil adopted inflation targeting, after floating the currency completely and
managing a 50% devaluation. This development changed one of the fundamentals of the
Real Plan: the exchange rate was no longer the nominal anchor, but rather the economy
was anchored by the inflation rate. This shift has an important implication for monetary
policy: the Central Bank now has an explicit target to meet in order to keep its credibility
and must unabashedly use its policy instrument (interest rates) to control inflation else
risk the public's confidence in the Bank. This was evidenced during the crisis of 2002,
when the Bank had to decide not aI hehrlrll, but by how much to raise interest rates
(Giavazzi, et. al., 2005). This crisis was handled well, with inflation barely exceeding its
(adjusted) target, and real GDP contracting very slightly (-0.2%) in 2003, a clear success
when compared to what could have happened if the contagion from Argentina would
have been more fierce.
The pages above convey the variability and volatility in Brazil's recent economic
history. Under the guise of various monetary policy regimes, political organizations, and
attempts at economic stabilization, Brazil has seen inflation so high that it cannot be
shown on a chart, deep business cycles, and extreme ebbs and flows of international
capital. Surely this economic rollercoaster is in the minds of economic actors as they
evaluate their risk tolerance with respect to the country and is implicit in the relatively
high spread vis-a-vis U.S. Treasuries from which the country suffers.
In the last ten years, however, the new monetary policy regime of the Real Plan has
endured the stress test of an international liquidity crisis, capricious sentiments of what
Thomas Friedman has called the "electronic herd" of international investors, and the
continuation of orthodox economic policies by one of Brazil's most left-leaning
politicians, Lula da Silva. How deeply discounted must the recent performance of the
economy be in light of the more distant past to justify the real interest rates experienced
by the country? For how long must the regime of inflation targeting work in order to
mend expectations and drive down the inflation premium in nominal interest rates? How
much of a fiscal surplus must be maintained for investors to believe the debt is
sustainable, in order to reduce the default risk premium? Will the past performance of
the Real reduce investor aversion to the currency and eliminate the devaluationary bias
implicit in the domestic interest rate? In short, what must Brazil do to lower both
nominal and real interest rates, in order to achieve more balanced growth and productive
investment? These are the questions that the next section of this paper will seek to
BRAZIL'S CURRENT INTEREST RATE DYNAMICS
Implicit in any monetary policy action or
inaction is an expectation of how1 the future
Trill unfold, that is, a forecast.
The macroeconomic history of Brazil has been anything but stable in recent
hi story. Specifically, only the most recent decade has been a time of relative stability in
terms of the usual macro-indicators. Despite this stability, however, interest rates remain
at exceedingly high levels. The basic interest rate equation, as illustrated in the
introduction, will now be examined to determine the sources of the persistent gap
between the yields on U.S. Treasury bonds and similar debt instruments issued by the
Brazilian national government.
The approximate Fischer equation, i, = r, + E(Fr,,, z), implies that there are at least
two components to the nominal interest rate: the real interest rate and the premium
demanded for investors to compensate for their expectations of future price inflation.l
There are many factors which affect the real rate of interest, r: investors views of the
currency, the possibility of sovereign default, the international economic environment,
the equilibrium growth rate of the economy, the national financial situation, and the
possible convertibility/liquidity risk for operating in a given currency (regardless of the
notional exchange rate)-to name a few. Many factors also influence the expected future
inflation: inflationary inertia, autonomy and credibility of the central bank (or lack
SThe exact Fischer equation has an interaction term between the real rate and expected inflation. It is
significant in high inflation countries such as Brazil but is ignored here because the equation is not
estimated, but rather used as the basis for the organization of the paper.
thereof), state of fiscal balances, and sentiment vis-a-vis the international economic
environment. Both parts of the model deal with markets' perceptions and forecasts of
probabilistic scenarios and future events. All standard economic models presuppose that
markets act efficiently and rationally, which means (among other things) that these
expectations are based on all available information and are the true "best guess" of the
outcome of these future events. In some cases, however, rationality may not hold. It is
with that possibility that this section of the paper begins, later returning to more standard
components of interest rates.
Testing for rationality in price forecasts
The so-called New-Keynesian Phillips Curve posits that it is the deviation
between expected and actual inflation that causes the short-run trade-off between
inflation and employment. Consequently, empirical models (such as those used by the
Brazilian Central Bank) require the use of observable phenomena (lagged inflation, actual
unemployment) as well as expectations of the future course of those phenomena. If a
systematic bias were to exist in the markets' expectations of inflation, that bias would
clearly have an effect on nominal interest rates. If, for instance, the market consistently
predicted inflation of 5% higher than that which actually was to occur 12 months hence,
the nominal interest rate would be 5% higher than it would be if the expectations did not
have this bias. In Brazil, one hypothesis would be that though the Real Plan has
stabilized inflation, the markets still "remember" the hyper-inflationary times of the early
1990s and do not believe that inflation will be as contained as it actually has been. An
alternative hypothesis would be that markets impetuously believe the Central Bank' s
announced target and set their expectations accordingly, not foreseeing the danger that
the Bank may not be able to manage the economy (down) to the target (this would
amount to accepting the target without considering evidence which may exist to the
contrary of the announced policy rate of inflation).
The inflation-focused rational expectations literature, pioneered by John Muth,
hypothesizes that expectations about future inflation are made in such a way that
incorporate all available information to the forecaster, and that errors in forecasting are
nonsystematic. This implies that agents use all available information to (on average)
perfectly predict inflation. In its inflation targeting regime, the Brazilian Central Bank
should then pay close attention to these private inflation forecasts, reacting quickly to
changes therein as these indicators truly point towards future changes in inflation. If a
systematic bias is suspected, the Bank should measure that systematic bias in order to
make some inference about the true private beliefs and expectations about future
inflation. Incentives exist for the Bank to make inferences about true future inflation
from the public's expectations, as any marginal improvement in its ability to foresee (and
control) future inflation will only further bolster its credibility.
There are two conditions for rationality. Unbiasedness means that the forecasts are
on average centered on actual, observed inflation. Suppose Ent+k is the actual inflation
observed in period t+k and Et(Ent+k) is the aggregate survey response to the future inflation
at period t+k, using all information available up to period t. The literature then claims
that if 0 = (0,1) in the following regression, the aggregate forecast is unbiased:
4+k t( EirkI ji, Or that the k-peariod aheadforecasto nltoseult
inflation k-periods ahead plus an error, which (in the case of rationality) is centered on
zero and distributed normally. Further performing a joint test for efficiency implies that
no available information could have improved the forecast. Allowing 9, to be the vector
of information available at the time of the forecast, efficiency implies P = (0 1 0 ... 0) in
the regression: Frtk E,1t+k t+k .
Expected inflation is difficult to measure. In fact, in many cases empirical
analysts assume an autoregressive process of actual inflation actually governs inflationary
expectations.2 In Some other cases, actual expectations can be observed. For example,
the term structure of interest rates can give some information about the market' s
expectations for future short term interest rates.
Also, expectations surveys exist in many countries for the most basic
macroeconomic variables, among them prices. The difficulty with surveys is two-fold: it
is difficult to know if they are truly the same expectations which affect policymakers
decisions/markets' assessments and, depending on the source of the survey, it is not clear
that the respondent has economic incentives to use all possible available information in
making his/her forecast. Imagine a phone survey to 10,000 Brazilian homes, for instance.
In many cases, the homeowners would simply answer the survey without doing the
proper research and certainly without consulting their own macroeconomic models. The
Brazilian Central Bank publishes the consensus forecast of professional forecasters for
growth in GDP, fiscal performance, and inflation. This series mitigates the
aforementioned downsides to survey-based measures. This particular measure of
inflationary expectations is a part of the Relatorio do M~ercado that is utilized by the
Bank (and made available to the public), and the series is the average of the responses of
professional forecasters who have clear economic incentives (bonuses) riding on their
2 The judgmental forecasts used in the Federal Reserve's classified Greenbook are no exception: the larger
FRB/US model does have an explicit role for expectations, however. See Federal Reserve (1997).
accuracy. Moreover, at the Central Bank of Brazil, inflation models do in fact treat these
survey based measures as rational expectations (Carvalho 2004). In fact, Fraga (2003)
shows that the point estimate of these expectations in a monetary policy reaction function
is over two, implying a one percent-point change in expectations translates to an over two
percentage-point move in interest rates. It is obvious, then, that the question of
rationality in this data should be examined.
Near the beginning of the inflation targeting regime, the Central Bank began to
survey private forecasters on their views of inflation. Those forecasts were initially very
unreliable, relying on faxes from respondents which were later compiled in Excel
spreadsheets: they suffered from low reliability, high variability, unpredictable response
rates and (worst of all for researchers) the data was never made available to the public
(Carvalho, 2004). However, beginning in late 2001 the Bank began to utilize a web-
based interface for the collection and dissemination of this information. The survey is
conducted daily with the reporting institutions submissions and the means, standard
deviations, and medians made available on a weekly basis.
A criticism of survey based measures is that when they are analyzed in the
aggregate, it is not necessarily true that each forecaster is using the same information set,
as it is possible some private information exists. In the case at hand, imagine an
investment bank that conducted its own point-of-sale surveys during the month to aid in
its forecasting. To alleviate this problem, this paper uses the methodology put forth by
Carvalho (2004); one day is chosen of each month to be representative of the entire
month and that day is chosen to minimize the chance of private information. Carvalho
suggests using the last day of the prior month, indicating that there is little chance that
private information such as a survey could help an institutional forecaster foresee price
changes in the following month. To make this point clear, consider this example: a
forecaster reports his 12-month price forecast on 2/28/XX ; the aggregate survey response
on this date will be treated as the March data point.
The data refute the null hypothesis (of rationality) in many cases. An interesting,
explicable result holds true: forecasters can predict short term movements in prices much
better than they can longer term movements. Specifically, tests were run on the means of
foreseen monthly inflation for periods 1, 3, 6, and 12 months into the future and also for
inflation accumulated in the next 1, 3, 6, and 12 months (the two 1 month regressions are
identical). It was only for forecasts in the 1- and 3-month horizons for which the null
hypothesis of rationality Table 3.1 Rationality Tests in Inflation Expectations
Rationality Tests for Expectations in INPC Prices
cannot be rej ected at the (standard errors in parentheses)
Slope Coefficient on
90% confidence level; this Dependent Variable Constant Expected Inflation!
One-month Inflation, t+1 -0.22 1.6
holds for both foreseen (0.15) (0.2.5)
One-month Inflation, t+3 0.7 0.03
one-month price changes, One-month Inflation, t+6 1.27 -1
as well as cumulative One-month Inflation, t+12 1.08 -0.89
inflation measures (see Accumulated Inflation, t+3 0.26 1/ 1.15
table). This rejection of Accumulated Inflation, t+6 4.46 0
Accumulated Inflation, t+12 14.77 -0.79
rationality may be the
result of a new monetary policy regime that is still building credibility and is also highly
subj ect to the adverse inflation shocks that the country experienced, especially during late
Keane and Runkle (1990) argue that testing the rationality of survey data is not
valid when conducted on survey means because each individual forecast that contributes
to the mean "will differ because of differences in information sets" (p. 4). While this
problem is partly alleviated by the method of selecting a representative day to reflect the
months' observations, some aggregation bias may still exist, especially in times of
(relatively more) uncertainty. Another way to compensate for the fact that this data
comes from a survey is to utilize a weighted least squares estimates, employing the
standard deviations of the forecast means to place less emphasis on data points about
which there is less consensus. Doing so tightens the estimation of the one- and three-
month ahead forecasts, making the case for rationality even more statistically sounds but
does not allow for accepting the null of rationality in forecasts of a longer horizon,
consistent with an OLS estimation. Because these estimations do not qualitatively alter
the findings, they are not reported here. The main finding, using data through the end of
2005, remains unchanged: forecasts in the very short-term are rational and reasonable
predictors of future inflation but forecasts for time horizons exceeding three months
exhibit notable flaws.
This paper does not seek to be is an exhaustive model of inflationary expectations
in Brazil. Recent work from Carvalho (2004) utilizes internal Bank data to build a panel
data estimation, based on the submissions of the individual surveyors which make up the
means and standard errors studies here. While theoretically superior, eliminating all of
the biases exposed by Keane and Runkle (1990, 1998), the results do not drastically vary
from those generated by the more simple estimations put forth in this paper. Most
importantly, unbiasedness is refuted for the 12-month case even when the panel is
estimated accounting for the non-spherical nature of forecast errors due to aggregate
shocks which effectively correlate errors across all forecasters (events like September 11,
2001 and the ensuing run-up in oil prices and the election of Lula da Silva in October
2002 which caused a significant depreciation and inflation through the typical pass-
The implication for policy-makers is clear: account for the systematic bias in
inflation expectations. The Fisher equation suggests that nominal rates should reflect the
real rate (to be discussed below) plus a premium for future inflation. If expectations of
inflation at time t for period t+i are systematically lower than inflation' s realized values at
time t+i, and the Central Bank knows this bias exists, there should be an additional
premium for the bias in inflationary expectations; more directly, interest rates should be
even higher than the casual observer of inflationary expectations would think. Indeed,
this seems to be the case in Brazil. Expectations for 12-month inflation were on average
96 basis points below realized values for the period September 2003 December 2005
(standard error = 21; see chart below). Moreover, for the whole sample, 12-month-ahead
Figure 3.1 Inflation's Expected and Realized Values
Expected Inflation (t) -Realized value (t~i)
-8 12 6 *3
expectations were 289 basis points below the realized value (standard error = 82).
Assuming the Central Bank knows this bias exists, it has a strong case for increasing the
wedge between nominal and real interest rates by the amount of this bias, making up part
of the difference which would appear irrational to the casual observer.
Recently it appears expectations have been tracking closer to inflation's realized
values. Rolling regressions show point estimates in the rationality tests trending closer to
one, implying rationality in the more recent forecasts, but with standard errors too large
to make conclusive statistical remarks. While there is not enough evidence to make a
statistical case for this argument, graphically one can see smaller errors that appear to be
centered on zero in the most recent year for expectations made three and six months
ahead. This may be due to the apparent success of the inflation targeting regime to
control the variability of price movements. Since mid-2003 the variability in 1-month
inflation has fallen precipitously, as measured by a rolling standard deviation measure in
the graph below (though it has crept up in the most recent six months). At the same time,
12-month expectations of inflation have also fallen and have, on average, been closer
Figure 3.2 Year-over-year Inflation Rates and Rolling Volatility of Inflation
12-month ahead inflation
0.6 expectations (right axis) 8
standard deviation In
0.2 1 -month Inflation (left-
612912001 612812002 613012003 613012004 613012005
to realized values as discussed above. It is the expectation of 12-month inflation which
should matter the most to the Bank, as the transmission between monetary policy and
inflation is estimate to have a lag of about three quarters in Brazil (Bogdanksi, et. al.
2001). If this trend continues, then, and expectations continue to accurately gauge future
inflation, the Bank should clearly remove some of its wedge between actual expectations
and the difference between the nominal and real rates.
This study of expectations of future inflation has been instructive. By
illuminating the obvious problems with expectations in the past, an excuse can be made
for the high level of nominal policy-set interest rates in Brazil: policymakers knew this
bias existed and set nominal rates systematically higher than the casual observer would
have deemed necessary. If, however, that bias has been eliminated, as the recent data
analyzed in this paper seem to suggest, a clear case can be made for eliminating the
wedge between the straight theory (the Fischer equation) and the nominal interest rate set
by the Central Bank.
If these expectations are in fact trending down and becoming better indicators of
future inflation, it is crucial for the Central Bank to understand what factors influence the
expectations in order to begin to systematically lower inflation (and interest rates).
Cerisola and Gelos (2005) estimate a model of inflationary expectations, estimating from
mid-2000 through the third quarter of 2004. The model estimated is:
ri, ao + Pl <-1 + Pz rtT + P3pb,_3 + P4r,_3 + p,reer,_3 + p06g, t- t,, Where the
variables are (left to right) expected inflation, a constant term, lagged inflation, the target
for inflation, the primary fiscal balance, the real SELIC rate, and deviations from trend of
the real exchange and wage rates. The most important determinants of this model, after
estimation, are the announced inflation target and the stance of fiscal policy, as proxied
by the primary surplus, which serves to capture expectations of future monetization. The
point estimate of the coefficient on fiscal policy, in fact, indicates that a one percentage
point increase in the primary surplus coincides with a one percentage point decrease in
inflationary expectations. Consistent with expectations, a positive inflationary pass-
through is found with a depreciated exchange rate (as prices of imported consumer goods
and also intermediate goods rises); the wage variable also exhibits the expected sign
(positive), also indicating an increase in inflationary expectations due to demand
pressures (the lagged nature of the variable may also indicate a labor cost of goods to
consumer price pass-through as well). An important finding, uncovered through
recursive estimation, is that "since early 2003, and particularly during 2004, the
demonstrated commitment to inflation targeting ...and monetary policy has begun to pay
off, as expectations are [more anchored in the] target and less dependent on past
inflation" (Cerisola and Gelos, p. 12). In fact, as shown in the graph below, which
shows the coefficients Figure 3.3 Contribution of Variables to Expected Inflation
from a recursive estimation laoo
of the expectations c, n
equation, the inflation
target and primary surplus
now account for more than
60% of the variation in 20 iaonrre
(explainable) expected ass
2000Mill 2001MO04 2001M09 2002MO2 2002Mi07 ZW2002M2 2003MC05 2003M10 2004MO3 2004MO8
Source: Cerisola and Gelos (2005), p. 15.
inflation. This suggests
that, along with the SELIC rate itself, the government's policies nearly completely shape
inflationary expectations and that through diligent stewarding of these variables the
government can lower the wedge between real and nominal interest rates.
This paper has now made one case for lowering interest rates: lower expectations
of future inflation coupled with improved predictive power of those expectations vis-a-
vis inflation's realized values. If this improvement is sustainable, the Bank should pay it
credence, lowering rates in lockstep with the improved (predicted) inflation outlook.
Indeed, as the epithet to this chapter suggests, a central bank' s actions are influenced
heavily by forecasts. As those forecasts project a more benign inflationary environment
and as they become more reliable, the bank should feel comfortable taking a more
accommodative stance in its monetary policy. This does not address the question of real
interest rates, however. The remainder of this chapter will seek to a) discover whether
real interest rates are high in Brazil when compared to comparable countries and b)
decompose the real interest rate half of the Fischer equation into its relevant subparts,
looking both at historical levels of these variables and their drivers, as well as expected
future levels to estimate the effect of each going forward on the Bank' s policy decisions.
Real Interest Rates: Comparisons and Determinants
Interest rates in Brazil are high by any standard. Despite the country's recent
economic success, rates exhibit stickiness with respect to historical levels. Comparing
the country to economic comparable sheds some light on the subject. The graph below
shows short term interest rates in five countries, including Brazil.3 In the case of Chile
the rate presented in this exhibit is the policy rate set by the Central Bank of Chile (the
TPM), while in the case of Argentina, Israel and Turkey the rates presented are the one-
day BAIBOR rate (for Argentina) and the interest rate on deposits held within the Banks
of Israel and Turkey, respectively. These countries are all good comparable for Brazil,
3 The countries in this analysis have not been selected at random. Others have compared this group due to
their analogous episodes with hyperinflation and stabilization. See, for instance, Beckerman (1992).
and even if we were to aggregate the rates, the average interest rate lies nearly
monotonically below that of Brazil in the post-Real Plan era.
Figure 3.4 Comparable Nominal Short Term Interest Rates
90 percent Comparable Nominal Short Term Interest Rates
80 Argentina Memo: 2005 Inflation and Debt/GDP Ratios
Inflation Debt /GDP Rati
70 IArgentina 12.0 700
60 Israel 2.7 10100
Chile 3.6 80
50 .Israel Turkey 7.71 680
Turkey Brazil 6.2 5000
Jan-02 May-02 Sep-02 Jan-03 May-03 Sep-03 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05
The persistence of Brazil's high interest rates is striking in this graph, especially
when considered side by side with the hyperinflationary experience and economic crisis
of 2002 in Argentina (the contagion from which didn't affect Brazil nearly as badly as the
initial crisis affected Argentina), and the relatively recent conquest of hyperinflation in
Israel and Turkey (early 1980s and 1999, respectively). The countries (even Turkey)
have returned to nominal rates well below Brazil's and are relatively indistinguishable at
present from industrialized economies (with the exception of Turkey). The experiences
of these countries seems to have completely been forgiven by investors, as rates were
high for a short-period of time and have since returned to the levels one would expect in
the absence of dire financial crises: except in the Brazilian case, where one could look at
these relatively risk-free rates (this is the overnight lending rate used between banks!)
and suspect that the country was in the midst of a financial meltdown. An oft-cited
reason for Brazil's high interest rates is its level of public debt. In this peer group, Brazil
fairs very well in that category, with a lower public debt / GDP ratio than all but Chile,
and half of that of Israel (CIA World Factbook data). Indeed, this common explanation
does not hold water when attempting to justify Brazil's high interest rates. Nor do
differences in inflation explain the difference in these interest rates. Furthermore, while
not theoretically correct, if one were to use past inflation as a rough proxy for
expectations, it is easy to back-out a real rate for each of these countries (table below).
Without a doubt, the real rate in Brazil at Table 3.1 Comparative Nominal and Real Rates
Nominal Rate Inflation "Real" Rate
year-end 2005 was exorbitant when Argentina 5.8 12.0 -6.2
Israel 1.2 2.71 -1.5
Chile 4.3 3.6 0.7
compared to its peers, two of which exhibit
Turkey 10.01 7.71 2.3
Brazil 18.0 6.2 11.8
negative real interest rates!i Note: Inflation rates as reported in The Economist,
December 24 2005, and are year-over-year changes
to Nov. 2005 with the exception of Turkey (Dec.).
Muinhos and Nakane (2005) give some statistical footing to this assertion about
the high level of real interest rates. They employ various measures, including (but not
limited to) a HP-filter smoothing, and I/S curve extraction, and an extraction from
estimating the marginal product of capital to find significant differences between Brazil's
equilibrium real interest rate and that of other economies. Their work does not seek to
explain the difference in rates, only to validate the assertion that the wedge does in fact
exist statistically. In every specification of their models, the rate for Brazil is
significantly higher than that of its peers.
Standard international economics textbooks posit that the real rate of interest is
intertemporally equalized across countries.4 The evidence at hand suggests this is not the
case, as the striking difference of Brazil's nominal interest rate in the graph above is not
accounted for solely by differences in inflation. This point would be even more clear if
the above chart were to include fully industrialized economies, which exhibit nominal
rates categorically lower than the countries in the above panel.
All of these standard economic models make some simplifying assumptions
which do not hold in the Brazilian case. Specifically, they assume that long-run
purchasing power parity holds, meaning that inflation rates between nations either
converge or are equalized through exchange rate movements. They assume that default
risk is the same across countries, that perfect convertibility between currencies exists, and
that there is no liquidity constraint on government securities. A sustainable fiscal
position is assumed in standard models, and these models also assume perfect
jurisdictional certainty, meaning that disputes over contract values will be adjudicated in
a predictable fashion and that there is no relevant adjudication risk in the country; this
hypothesis in particular builds on the "original sin" work developed by Hausman, et. al.
(2003), which says that once countries begin to dollarize their financial liabilities it is
difficult to de-dollarize them, in part because of the aversion of local currency
denominated debt due to this jurisdictional risk. The relevant data on Brazil indicates
that none of these simplifying assumptions has held in the past and that while some of
them do in fact hold today, there is still a wedge between theory and practice due to what
for Brazilian Central Bank head Arminio Fraga has called "...weak macro regimes ..
and a long history of misbehavior" (Fraga, 2005, p. 297).
4 MOst include a footnote or caveat to this statement, but it is seldom emphasized. See, for instance, Husted
and Melvin (1998).
Apologists for Elevated Real Interest Rates
Three basic hypotheses explain the persistence of high real interest rates in Brazil:
the "bad equilibrium," insufficient fiscal adjustment," and "sequence of negative shocks."
In the "bad equilibrium" argument, monetary policy is seen to have been necessarily
conservative in the pegged exchange rate regime (1994-1998) in order to compensate for
the high (and accurate) risk of devaluation. In this case the economy was using a foreign-
exchange anchor that lacked credibility and perceived sustainability. However, the
nominal target was changed to the inflation rate with the maxi-devaluation in 1999 and
the move to inflation targeting. In the post-float, inflation-targeting regime (1999-
present), the appropriate adjustment was not made in interest rates at the time of the float
and now policy is not able to suddenly shift them down. The basic tenant is that very
high real interest rates raise the level of risk on government debt, and so a lower real
interest rate brings down the possibility of default and thereby justifies a lower country
risk premium and a lower interest rate. This theory deserves some attention because if
applied properly, it describes the Brazilian case fairly well. It is easiest to portray this
multiple equilibrium model graphically (and not algebraically), so consider the graph
below: the horizontal axis is the domestic interest rate, and vertical axis is the risk-
adjusted return on foreign assets in local currency terms, (1 + i* + z). The non-linear
function (that appears to be a sine wave) is the expected return on domestic financial
instruments, including an adjustment for possible default. This function is non-linear
because initially higher interest rates imply higher expected returns but if they raise too
much, increased bankruptcy risks due to higher debt service burdens increase the
bankruptcy risk. However, as the interest rate gets too high the possible default risk may
go down, as the government acts to alter perceived default risk (through IFI lending or
through the creation of a budget surplus). At certain prohibitively high levels of interest,
however, unstable debt dynamics dominate and the expected return plummets.
Regardless of where on the curve interests rates are, in equilibrium, this function must
intersect the exogenously-set international rate of interest. In this model, point A is the
stable, low interest rate equilibrium, B has a low interest rate but is unstable, C is the high
interest rate equilibrium, and D is an unstable high interest rate locus. Imagine the
economy was operating at point A when it was hit with an external interest rate shock,
that temporarily moved the foreign interest rate relationship from the solid to the dotted
line. Once this shock was removed, the economy would move to point C, not to point A.
This is the argument of the bad equilibrium hypothesists, some of whom use the 1997
Asian Financial Crisis as Figure 3.5 Stylized Depiction of a "Bad Equillibrium"
this exogenous shock
E(R, (1 + i + z)
(Palley, 2003). The
challenge in their minds iS B
to move Brazil from point C
to point A. A E(R)
This theory amounts to
Domestic interest rate
a sort of chicken-and-egg
argument, and in their oft-cited work on Brazilian interest rates, Favero and Giavezzi
(2002) argue that Brazil "should not rely on a transition happening automatically" (p. 6).
Instead, Brazil should follow the example of other countries that have made a similar
transition: endure a large primary surplus for a protracted time in order to justify the
lowering of the SELIC rate.
This prescription follows logically into the second hypothesis, which says that
fiscal adjustment is not yet complete. Despite a sizeable primary surplus, the debt/GDP
ratio remains high enough to continue to crowd out private sector funding needs (as that
debt is rolled over before it is retired), which raises the market interest rate. The policy
prescription here is the same: continue to run a fiscal surplus in order to bring down rates
by reducing the amount of government debt outstanding.
The "adverse shock" hypothesis claims that shocks to the economy such as the
U. S. stock market, 9/11, Lula, and the Argentinean meltdown have forced the Central
Bank to use interest rates to smooth volatility that otherwise would have crippled the
economy (notwithstanding the crippling effect of the high interest rates). Proponents of
this theory would then argue that in the absence of further shocks, interest rates will soon
begin to move to lower levels. In essence, the argument is that ex post rates are high
(because inflation has not picked up, nor did the risks to the economy really materialize),
but ex ante they are in line with theory and the prevailing risk assessment at the
beginning of the terms of the rates. While theoretically sound, the analysis of the data
presented earlier reveals a non-sequitur in this argument: the economies that actually had
the fall-outs have already seen their rates recover and Brazil's rates (ex-ante and ex-post
real rates as well as nominal rates) are persistently high (see introduction and this chapter,
above). Apologists for the theory counter that Brazil's case was peculiar: monetary
policy had to be excessively tight to make sure credibility was maintained for the
incipient inflation targeting regime, as well as to defect the exchange rate. Evidence to
that point is that the real money stock from Sept 2000 Mar 2001 fell by 3.8%, though it
had increased 8.6% the previous year (Vieira de Faria, 2003). Furthermore, an exchange
rate-based response to these crisis may not have worked because of the dollarized nature
of the country's public and private liabilities: the effect of a depreciation is indeterminant
as exports are increased but net financial liabilities increase as well. Because of these
points, "Central Bank autonomy in the process of setting its key interest rate is more
theoretical than real" (Vieira de Faria, 2003, p. 46). Even if Brazil's interest rates should
have recovered from these shocks three years hence, a dedollarization of financial
liabilities would help the countercyclical nature of exchange rate adjustments become
more effective in the future, thereby putting less pressure on monetary authorities to alter
the basic cost of capital.
Historical Factors and Current Realities of High Real Interest Rates
The challenge, then, is to identify what measurable factors are influencing the real
interest rate and to suggest policies which will influence those mechanisms in order to
justify a lower real interest rate. This section of the paper lends heavily on the vast
literature, pioneered by Favero and Giavazzi (2001) and mentioned in the introduction
which seeks to explain various components of the real interest rate. The goal will be to
act as a compendium of that research and to look at the current performance of the
Brazilian economy and how it fits into the models estimated by these economists.
In their work, Favero and Giavazzi find that there are two maj or factors which
influence the country's interest rates: monetary policy and term premia. They estimate a
Taylor rule for Brazil, noting this simple rule explains the actions of the bank very well.
Furthermore, they estimate that the equilibrium policy rate (the rate consistent with zero
accelerating inflation) is around 17.6 percent. This rate, called the "neutral rate" by some
economists is extremely high (for reference, the Federal Reserve estimates the neutral
rate in the United States to be between three and five percent'). A first read of this paper
SWhile not an official estimate, numerous Fed Governors and staff have put the estimate in tlus range.
See, for instance, money.cnn's interview with St. Louis Fed President William Poole (11/10/2004).
may leave the impression that this is an open-and-close case: interest rates are high out of
necessity to stabilize inflation. Indeed, the average SELIC rate in 2004-2005 was 17.7%
and inflation has essentially been stabilized in those two years, as indicated in the chart at
right (through Dec. 05). What this ex- post modeling misses, however, is why the rates
have to be so high in the first place. Indeed, the oscillations in the SELIC rate in the last
two years, indeed in the six years of inflation targeting, have accurately shaped
expectations of future inflation
Figure 3.6 Recent Trends in SELIC and Inflation
and have helped to control
inflation during this time. Could Slert
the same oscillations not occur in
a range of 8-10%, or 5-7%? Why
must the level of interest rates be
so high if it is the change in rates
Jan-04 Apr-04 Jul-04 Oct-04 Jan-05 Apr-05 Jul-05 Oct-05
that controls inflation by
marginally curtailing or encouraging activity?
Favero and Giavazzi argue that one important factor pushing up the high neutral
rate is fact that the BCB enj oys none of the institutional guarantees that other central
banks have in their mandates; "we know of no other central bank that successfully
operates an inflation targeting regime and is not formally independent" (p. 5). This point
cannot be stressed enough; in their work on Brazilian inflation in the pre-Real Plan era,
Tullio and Ronci (1996) estimate that a Phillips Curve that includes a proxy for the lack
of political independence of the Central Bank6 that shows that it is strongly correlated
6 Their proxy is determined by the turnover in Central Bank governors and chairmen.
with inflation and accounts for a portion of the rampant inflation this period.' This lack
of independence, then, directly influences the real rate, as argued by Favero and
Giavazzi, as well as pushes up the nominal rate through increased inflationary
expectations. The importance of independence has a long history in the economics
literature and its conclusions are summed up very well by Alan Blinder, who writes:
Monetary policy requires long time horizons because the lags through which
interest rate changes affect output and inflation are typically long. If monetary
policy decisions were left to politicians, on a day-by-day basis, the temptation to
reach for short term gains at the expense of the future (that is, to inflate too much)
would be hard to resist, especially in the run-up to an election. Knowing this, many
governments wisely try to depoliticize monetary policy by, e.g., putting it in the
hands of unelected technocrats with long terms of office and insulation from the
hurly-burly of politics. (Blinder, 1998, pp.56-57.)
As discussed, Brazil does not belong to this "wise" group of nations with depoliticized
central banks. Recent changes and stirring in Brasilia towards increased Central Bank
independence should be applauded, then, and will be discussed in more detail in the
conclusion of this work.
In terms of longer policy rates, they employ a system of equations to simulate
future SELIC rates in order to give estimates of theoretical rates of longer maturities that
are based solely on monetary policy rates.8 Their conclusion is that almost all (95%) of
the spread between observable long-term rates and this theoretical rate is explained by
term premia and not by uncertainty about monetary policy. Their conclusion is valid
based on their assumptions, but incorporating the work of Cerisola and Gelos (2005,
discussed above) on inflationary expectations shows that those same fundamentals and
SOne could make the reverse argument: high inflation causes turnover (i.e. the firing) of Central Bank
officials. Tullio and Ronic report Granger causality tests which refute this alternative interpretation.
STheir dynamic simulation is based on a simple Taylor rule and is closed by an equally uncomplicated
model of inflationary expectations, which is based solely on the target rate of inflation and its recent
performance. Their model is overly simple and ignores the factors which affect expectations, as discussed
in the previous part of this chapter. However, this simple specification does fulfill their goal, which is a
simulation of ceteris paribus future policy rates.
debt dynamics are also an important part of agents' expectations of inflation and should
not be ignored when analyzing short-term interest rates. Because of that important
transmission channel, Favero and Giavazzi's findings on long term rates can be extended
to considerations of the Selic rate as well.
Since these term premia are not affected by devaluation expectations, Favero and
Giavazzi conclude that the spread is caused by macroeconomic fundamentals and debt
dynamics. Specifically they point to the level of government debt and the volatility of
international financial markets as influencing the perceived risk of government default.
When internal financing rates are widely determined by international markets, and the
government debt is large, the default risk is highly affected by the correlation of the
growth rate of output (which in turn affects the level of the primary surplus) with
Financial spreads. This additional consideration in the Brazilian case implies that a
larger-than-otherwise necessary surplus is required to make the debt sustainable and to
bring down the cost of capital. Favero and Giavazzi point to the experience of Italy in the
mid-1990s when a one-year primary surplus in excess of 6% of GDP was interpreted as a
commitment to a more responsible fiscal policy and which began a virtuous cycle that
helped move the country towards a permanently lower cost of debt.
The driving forces identified thus far are a large and unmanageable (not to mention
dollar-denominated) national debt and the lack of institutional independence granted to
the Central Bank. While the work of Favero and Giavazzi and others in this area is
accurate, this particular policy mix alone will not bring about the panacea of lower real
interest rates that the country so desperately needs. More recent work points to other
factors which need to be considered as well in order to bring about the level shift in the
country's cost of capital.
In fact, some of this work refutes the analysis of Favero and Giavazzi to a limited
extent. Writing in 2003, Thomas Palley notes Brazil's current debt situation implies
dynamics that cannot avoid default at existing real interest rates, the surplus needed to
stabilize the debt would need to be 5.1% of GDP in order to stabilize the debt,
alternatively noting that in the growth of national debt as a percentage of GDP is given by
the following equation: gD, T v I-I-g At the time of his writing, the domestic
interest rate, i=18%, the deficit/GDP ratio, d/Y = -3.75%, the GDP/Debt ratio Y/D =
1.67% and GDP growth g(Y) = 3.5%, implying that interest rates, i, needed to fall to
9.75% just to stabilize the debt ratio (Palley, 2003). At the time of this writing, the
relevant indicators were i=17.25%, the deficit/GDP ratio, d/Y = -5.6%, the GDP/Debt
ratio Y/D = 2.59 and GDP growth g(Y) = 2.5%, implying that interest rates needed to fall
to around 12% to make the debt sustainable. While this is slightly more palatable than
the estimate above, it is still clear that the current 17.25% interest rate prohibits debt
sustainability. This same doomsday scenario was put forth in a 2002 critique of new IlVF
lending from the Washington-based Center for Economic and Policy Research, which
argues that "[assuming the most optimistic of scenarios] .. stabilizing the debt-to-GDP
ratio through primary surpluses would not be potentially achievable" (Weisbrot and
Baker, 2002, p. 3). These scenarios have not, as of yet, played out because of the
aggressive debt management policies of Lula, which will be considered in the conclusion
of this work.
Fiscal concerns are not the only ones that affect the real interest rate, however.
Arguing for a broader "Monetary Conditions Index," Carneiro and Wu point out that
small countries in the global economy often times have a peculiar combination: high real
interest rates and deficits in the current account. Using panel data, they show that (on
average) "every 1% increase in the current account deficit (as a percentage of GDP) is
associated with an increase in real interest rates of 20 to 30 basis points" (2004, p. 304).9
Furthermore, they report a strong statistical relationship between the size of the current
account deficit and the real interest rate (measured ex post). The theory here is relatively
straightforward: increases in the current account deficit create an external funding need
which can only be satisfied in the risk-adjusted rates of return for new (marginal)
borrowing are sufficient. To the extent that Brazil ran significant trade in the period
through 2003, this explanation is valid. However, the argument looses validity when
considering the persistent trade surpluses generated by the country since mid-2003 (see
chart below). To their credit, Cameiro and Wu only consider data through 2000 (though
Figure 3.7 Brazilian Balance of Payments
Foreign direct investments (FDI), Current transactions (CT) &
External financing requirements (EFR)
US$ billion Accumulated in 12 months
-32 FDI -CT Deficit - EFR f
1999 2000 2001 2002 2003 2004 2005
their work was published in 2004). Is this the break-down of a previously persistent
statistical (and theoretically sound) relationship, or did the work suffer from
misspecification? It seems unlikely that this misspecification would hold true across a
panel of 20 countries, as Cameiro and Wu estimate. More likely than not, as Nobel prize
winning economist Joseph Stiglitz writes, "Brazil's financial markets today are probably
a good example of excessive pessimism" (Esterl, 2002). As argued in the inflation
9 My translation of the original text, written in Portuguese.
expectations section, it is the position of this paper that the markets are persistently slow
to adapt to the evolving insertion of the Brazilian economy into the global marketplace.
This basic idea is supported further by the decomposition of what are called "the
cousin risks," the country risk and the currency risk, so-called because of their highly
cointegrated nature (Garcia and Didier, 2001). This work has shown that one of the
components of Brazil risk, that of convertibility risk, was very important through the time
of the international financial crisis of the late-1990s but no longer has statistical
significance. In fact, this decomposition has shown than in the post-inflation targeting
period, the currency risk can be estimated at above 5% per year, while the explicit Brazil
risk is less than 3% (Garcia and Dider, 2001, p. 32).10 The irrationality hypothesis, then,
is supported in light of the 18-month rally of the Real vis-a-vis the dollar though
December 2005, along with the strongly export-driven current account surplus,
juxtaposed with the persistence of high real interest rates. If the market had foreseen this
performance, interest rates would have dropped significantly, but they didn't.
Another persistent aspect of the Brazilian economy which contributes to high
interest rates is the so-called jurisdictional uncertainty. No long-term market for credit
exists in the country, and the only long-term credit available outside of the country is that
which is denominated in dollars. Arida et. al. (2005) argue that this inability has
produced an important new risk element in the country, that of uncertainty over the
adjudication of contracts, as well as enhanced currency risk in light of the dollarization of
financial liabilities. The authors argue that a policy aimed at correcting this uncertainty
would lower interest rates through an increase in convertibility (or liquidity) of the
'O Admittedly, this estimate came before the explosion in the country risk premium associated with the
Lula election, but as levels have returned to "normal" since that time, the author feels the statement is still
currency, as well as a reduction in the effort costs of the Central Bank to artificially
lengthen the maturity of that portion of the national debt that is denominated in local
currency. The perception of the riskiness of Brazil's currency and the inability of agents
to enter into long-term financial contracts in their home jurisdiction adds to the interest
rate dilemma, then, by increasing demand for short-term credit and making the future
even less easy to predict.
Existence and Causes of High Interest Rates
In short, interest rates are high in Brazil because of perceptions and expectations.
Inflationary expectations are the maj or driver of interest rates, and Brazil's volatile past
with respect to inflation rates have left a permanent scar on economic agents' ability to
foresee the future. Furthermore, it may be in light of this inability that monetary policy has
been excessively punitive, in order to curb inflation in a worse-case scenario sense (i.e.
when realized inflation is even greater than the prevailing expectation). Real interest rates
are also high. When compared to other countries that have similar macroeconomic
fundamentals and problems and have weathered, (or even caused) some of the shocks that
flustered the Brazilian economy, the prevailing interest rate in Brazil is persistently higher.
The economy's debt position, including its level, its indexation to the dollar, and its
growth, all contribute to the inherent risk of lending in the country. A net importing
position has also added to those external financing requirements, further altering
perceptions of the sustainability of the economic model the country is pursuing and raising
the cost of capital. It is the assessment of markets that necessitate these high interest rates
and no sudden change in the rate would be accepted instead it would cause capital flight
and currency depreciation, as the country would be deemed even more risky. But
expectations take time to evolve. The conclusion of this paper will examine the most
recent Brazilian economic data and argue that a fundamental change has occurred, one that
is still awaiting the approbation of the international financial community.
O JEITO JURIDICO BRASILEIRO: COPING WITH THE PROBLEMS RESULTING
FROM HIGHLY ELEVATED INTEREST RATES
Interest rates have been extremely high in Brazil even given the highly
unpredictable economic environment and to attempt to control inflation, which has
rocked the country's stability since the late 1950s. As argued by da Cunha and Ferreira
(2004), the cost to individuals of this punitive monetary policy is equally extremely
detrimental to individuals who do not have access to long-term credit. These same costs
are experienced by firms, which must forecast their performance (and that of the overall
macroeconomy) when deciding their optimal capital structure and in making financing
decisions. The consistently high interest rates (from which ex post real rates are difficult
to accurately forecast) implies a business environment that inhibits balanced growth.
However, during that time Brazil has had impressive periods of growth and has seen
industrialization and wealth creation for many of its citizens. The mechanisms that have
allowed this growth to occur despite cripplingly high real interest rates are the subj ect of
this chapter. First this paper will examine the peculiarities ofBrazilian finance at the
firm level, later turning to peculiarities more personal in nature.
Firm-Level Financing Decisions
The large overhang of government debt (a true hang-over from times of large
budget deficits) has a deleterious impact on Brazilian firms. Government issuance of
debt needs to be purchased by investors: this pushes out the demand curve for capital
and, in doing so, increases the equilibrium interest rate in the economy (as discussed in
the previous chapter). The result is known as "crowding out," in which government debt,
by definition less risky than debt of Brazilian national firms without access to
international funding, takes available, investable savings "out of the system." The result
is a credit crunch for Brazilian firms, that is to say a scarcity of capital and the de facto
imposition of capital constraints for firms. These capital constraints have been verified
empirically by many scholars, most recently Bonomo et al.(2003) and Terra (2003). Both
papers examine sales accelerator investment models of investment. In this model, if all
the factors that influence investment are accounted for, cash flow should not affect
investment decisions. Stylistically, if a firm has no credit constraints and subsequently a
proj ect exhibits an expected return above the marginal cost of capital, it should be
pursued regardless of the firm's cash flow position at that time. However, if the cash
flow variable explains some part of investment decisions, then there is evidence that the
firm faces a credit constraint and can only make those decisions in economic "good
times" of high cash flow. In both papers, significant levels of credit constraint are shown
for all Brazilian firms.
Terra (2003) bifurcates the analysis by using panel data and aggregating the
regression results by firm type and also by the level of dependency each firm shows with
respect to external financing. The tested hypothesis is that firms that are more
international in nature will exhibit less credit constraint. A second hypothesis is that
firms that are more dependent on external finance will invest more when capital inflows
are high (and consequently firms' access to credit is less constrained). A perhaps obvious
result follows: firms that are multinational and/or larger in nature show less credit
constraint than more domestic/smaller firms, particularly in periods of high capital in-
The implication of this result is made obvious when comparing the composition of
private, corporate sector debt: firms that are large and multinational comprise more than
half of private sector debt and that share increases in times of rising or elevated inflation
(Tullio and Ronci, 1996). Large firms, generally less risky, are more able to handle the
disclosure costs of debt finance, and can marginally raise debt finance at less of an impact
to the cost of equity.' Moreover, these firms can more easily tap international debt
markets, sometimes borrowing at rates that are preferential to those that the government
receives (more of this below).
Impacts on Large Firms
Size is also an interesting component to the equity financing of these firms:2 in the
post-Real Plan era, over 70 Brazilian firms have directly listed their stocks on the New
York Stock Exchange through the American Depository Receipt program, or ADRs
(State Department, 2005). ADRs allow the firms to sell their equity on U.S. markets in
shares that are denominated in dollars, both improving liquidity and removing investor-
level uncertainty over foreign exchange effects. This cross-listing has made their stocks
more liquid and has forced upon them US GAAP reporting standards, both measures
presumably serving to lower the cost of capital of these firms by increasing the
desirability of their equity issues. This increase in ADRs coincides with the opposite
phenomenon in the Sho Paulo stock exchange, the BOVESPA, as the number of firms
SConsider a firm with R$ 1 million in equity and a firm with R$ 100 million in equity. A R$1 million debt
issuance makes the equity in the first case much more risky, all else equal.
2 This is, of course, inextricably linked to the interest rates through the CAPM relationship.
listed on the BOVESPA fell 15% from 2001 to 2003 (BOVESPA data). In bypassing the
Brazilian equity markets and raising money abroad, this subset of Brazilian corporations
hopes to be able to raise money at rates that are effectively lower than the risk-free rate in
Brazil. Raising this money in foreign markets should lower the aversion of the typical
international investor, as ADRs are, by their very nature, free of foreign exchange rate
risk. This hope has not materialized, however. The cost of capital for firms that list as
ADRs does not materially differ from those that only list in Sho Paulo (Hail and Luez,
2005), 20.0% vs. 21.4%. However, this result does not hold across companies from all
countries (the panel is of over 30 countries), but is largely consistent for all countries in
Latin America.3 This result implies, then, that the cost of increasing a firm's dollar based
equity (though revenues are principally in reais) is not always justified by a lower cost of
capital. However, these firms do exhibit common characteristics which explain their
propensity to cross-list on the NYSE. De Medeiros and Tiberio (2005) employ a logit
model which proves that the propensity to cross-list is governed by some important
principles. Firms that are larger, as noted, tend to cross list, as do firms that are highly
active issues on the BOVESPA. Firms with a high market-to-book ratio also experience
an increased propensity to cross-list, perhaps because the benefit to doing so is greater
(because their stocks are worth more). Also, firms with better corporate governance
standards exhibit greater probabilities of being cross-listed. All of these relationships are
as expected, and may relieve some of the credit constraints of these firms, but do not
remove the fundamental result of Hail and Luez (2005), which is that these firms do not
enj oy significantly lower cost of capital, despite their ability to raise equity abroad.
SIndeed, WACC estimates for Brazil vary by more than the other Latin America nations, which exhibit the
same similarities in sample means (Hail and Luez, 2004).
Indeed, something systematic (and, as noted, outside of currency considerations) is
raising the cost of capital for these firms; they seem to suffer greatly from the same Brazil
risk that haunts the government' s key interest rates.
These ADRs are just one piece of a complex financing strategy that helps large
firms try to evade the problems associated with costs of constrained capital. Firms often
employ multi-tiered Einancing strategies. In addition to listing both on the NYSE and the
BOVESPA, paper-marker Aracruz diversifies its debt structure across many interest rate
mechanisms and currency denominations (chart at right, below).4 The firm has interest
rates that are tied not only to the SELIC Figure 4.1 Aracruz's Debt Structure
rate (those contracts which are indexed to Foreign Currency Denomination and Interest
Rate Determination of Aracruz's Debt Structure
the Taxa de Juros de Lon~goPrazo, Brazil's Incal, TJLP_ Incal, Fixed
government-set long term interest rate), but
Foreign, Incal, Folvign
also to the LIBOR rate and to the LIBOR Ib currency
26% B basket
movements of a basket of currencies.
Importantly for the firm as well is the fact \ ,
that it issues debt in both reais and dollars, .ore 05SCI ECA aa~s
creating a natural hedge for its dollar-denominated revenues.'
One impressive facet of this debt finance is the low interest rates associated with
the various trenches of the firm' s long term debt. The local currency denominated
trenches are financed mainly through bonds linked to the TJLP and the spread over the
TJLP that Aracruz pays implies yields on these bonds of less than 10%, markedly better
SThis section relies solely on the author's analysis of the firm's 2005 20-F Filing with the SEC, available
on-line from the SEC EDGAR database.
SIn 2005 the firm reported that 98% of its revenues were denominated in dollars (company form 20-F).
than the overnight Selic rate discussed above. Even more impressive, though, is that the
yield on the foreign currency denominated bonds is less than 6%, similar to average risk
North American firms. How does Aracruz borrow at rates preferential to those of the
Brazilian national government? One answer to this question is that a large portion of
these loans come from the National Economic and Social Development Bank (the
BNDES) and is effectively subsidized: the Brazilian government lends the money at rates
that are below its own cost of capital. This subsidy is somewhat recovered by the Bank
through the large equity stake that the BNDES holds in Aracruz. The firm also benefits
from below market-rate debt financing from the World Bank and the International
Finance Corporation, both at rates below 6%. Thirdly, the scope and dominance of
Aracruz in its sector may imply that indeed its operations justify a smaller risk premium
than that assessed to the national government.
Aracruz is not alone with respect to Brazilian firms that seek to lower their cost of
capital through international debt issuance and other financing activities. An analysis of
other Brazilian mega-firms will reveal similar patterns: financing subsidized by the
BNDES (who is often a large equity holder in the firms), ADRs that serve to increase
equity liquidity, and loans from multi-lateral financial institutions that offer loans at
highly preferential rates. State oil firm Petrobras issues long term debt at much better
rates than the federal government, with coupon rates on its bond ranging from 3.7 9.7%6
In the notes to its financial statements, the firm remarks that its long term borrowings
"consist primarily of the issuance of securities in the international markets, debentures in
the domestic capital markets, financing from the BNDES, and ...facilities guaranteed by
6 As of this writing, the bonds were trading at values immaterially different from their book values, so this
comparison is valid.
other multilateral agencies" (Petrobras Form 20-F, 2005, p. 105). It adds that short term
credit is "derived mainly from commercial banks including trade lines of credit and
commercial paper," and that interest rates on its short term instruments range from 2.86-
6.01% (Ibid., p. F-20). This firm, then, shows no capital constraint and can easily tap
into international markets, perhaps even more easily than the Brazilian government itself.
The interest rates that it pays are comparatively low (though the reader should be
reminded that they are essentially free of foreign exchange risk, since they are
denominated in dollars).
These two mega-firms can effectively bypass the problems associated with high
domestic interest rates by raising capital internationally (though the question remains to
be answered with respect to the overall cost of capital, implying highly irregular equity
risk premiums this question is beyond the scope of this work, though).' The cases of
Aracruz and Petrobras are not unique, however. An examination of any of the Brazilian
megafirms will reveal similar characteristics (low interest rates, liability dollarization)
and perplexities (why does the government subsidize so much of these firms' capital?).
Impacts on Smaller Firms
A more interesting question, however, is the manner in which firms that are
smaller can raise capital for their operations. Terra (2003) and others have shown that
firms that do not fall into this category show significant credit constraint and oftentimes
cannot undertake profitable proj ects because of the scarcity of capital. Yet these firms
SThe cost of capital for a firm is determined by: we"ke + Wd*kd Or the sum of the products of the portions of
capital and equity in the firm' s capital structure with their respective after-tax costs. This analysis is
mainly concerned with the term kd, while the observation made here refers to a high cost of equity, ke
continue to exist and, in many cases, to grow. What is the source of their financial
constraint, and how do they finance themselves in order to grow?
In 2003 the World Bank and the International Finance Corporation charted the
World Business Environment Survey. Its mission is to assess the business environment
of member nations with respect to opportunities and limitations for growth, government
intervention, corr-uption, financing decisions, and expectations for the future. The dataset
is rich, striating firms across countries and by size.8 The table on the next page was
generated from this survey and speaks to two important questions: what is the source of
financial constraint and how are firms financed? It reveals differences across firms' sizes
that may, by now, seem obvious to the reader. A striking difference is the problems
posed by high interest rates. 90% of small firms identify interest rates as a problem,
while only 73% of large firms do so (which admittedly is still a high proportion). This
difference does not even fully capture the impact of these rates, however, since a firm of
500 employees counts equally as a "large" firm as does a firm of 15,000 employees.
Within the large firms there is a negative correlation between the value of a firm's fixed
assets (as a proxy for size within the group) and the extent to which high interest rates are
Interestingly enough, high interest rates are not the only variable for which there is
a significantly different effect across firm size. For small firms, paperwork related to
financing is a problem for over 55% of firms, while only 30% of large firms feel this
constraint. This may, in turn, be related to the higher reported lack of non-bank equity,
foreign bank lending, and export and lease financing available to smaller firms. If a
SThis dimension is a categorical variable: "small" is 1-49 employees, "medium" is 50-499 employees,
while "large" is 500 or more employees.
# of Interest Special Foreign Non-Bank Export Lease
Firms Collateral Paperwork Rates Connections Monev Banks Equitv Finance Finance Credit
Small 30 43.3% 56.7% 90.0%1 26.7% 16.7% 10.0% 23.3% 10.0%1 6.7% 10.0%
Medium 141 44.0% 41.8% 83.7%1 27.0% 9.2% 9.2% 14.9% 9.2%1 3.5% 10.6%
Large 30 43.3% 30.0% 73.3%1 30.0% 13.3% 6.7% 10.0% 6.7%1 3.3% 10.0%
Table 4. 1 Source of Financing Constraints and Firm Capital
Financial Constraints Imposed by (a lack of):
Source of Firm Capital
# of Retained Foreign Family Money Supplier Public
Firms Earnings Eqiy Local Bank Deveomn Bank Banks Members Lenders Credit (A/P) Leasing Sector Other
Small 25 58.4% 0.0% 11.6% 0.0% 0.4% 14.4% 4.0% 6.2%/ 1.0% 0.0% 4.0%
Medium 123 38.1% 6.8% 23.0% 2.1% 3.5% 3.9% 0.3% 12.2%1 6.1% 2.1%1 1.9% o
Large 24 37.9% 4.3% 21.8% 1.3% 14.0% 0.8% 0.0% 7.2%1 8.3% 2.1%1 2.4%
Note: Not all firms answered questions regardmg the source of firm capital
Source: Author's Calculations from World Business Environment Survey, from the World Bank
copious amount of compliance and disclosure is required, in addition to the bureaucratic
costs for loan processing, it is easy to understand why smaller firms are not more
externally financed. The same is true of subsidized credit from the BNDES, which would
be included here as non-bank equity or as credit, depending on the arrangement. The
lesson, then, would be to reduce the bureaucracy as much as possible in order to open up
these preferential sources of capital to smaller and more medium sized firms. It is these
smaller firms that truly need capital, but if the costs are too high and only a large firm can
bear them (due to economies of scale), development initiatives are thwarted and credit is
not optimally distributed. The survey also illuminates some differences in the source of
firm capital, all of which naturally follow from the implications of the sources of
financial constraint. A notable gap is in the financing which occurs through retained
earnings. This internal financing implies a firm can only grow once it "gets going,"
financing projects through profits. Nearly 60% of the capital of small firms is generated
through retained earnings, while only 37.9% of the capital of large firms comes from this
internal source. Local banks are also notably less important to small firms, accounting
for twice as much of the capital of larger firms. There is a reversal between the
categories of foreign banks and family members, with the former disproportionately
financing large firms, and the latter being important nearly exclusively in the case of
This capital structure has important implications. In the case of large firms, they
can expand through external financing, even when they are not growing organically.
This option is not as available to smaller firms, who must wait for positive cash flows
(and net income) to finance themselves through profits. Financing through family
members is an option, but it comes with an important theoretical concern: markets are not
allocating these resources to the firms, rather it is familial allegiance that act as
guarantors of the borrowed assets. While this is arrangement commonly provides capital
to firms, there is no mechanism to filter out bad investment ideas: presumably the funds
are granted in a less stringent fashion than, say, a loan officer or private equity investor
would undertake. A more competitive marketplace for smaller firms to bid for capital,
then, would lower the cost of that capital for profitable proj ects and assess (and account
for) the risk implicit in other, not as lucrative, endeavors. With more Einancing available
in a competitive marketplace, entrepreneurs without the access to the "special
connections" that makes Einancing easier for some will also be able to finance their ideas.
Access to capital is one of the most important facets of successful new business
development and a more market-oriented allocation would help Brazil to more efficiently
allocate its relatively scarce capital. It is in this direction that the evolution of capital
markets in Brazil should move.
Implications of Interest Rates for Individuals
If small firms are ill-prepared to pay high interest rates to finance their endeavors,
individuals are even more poorly equipped to bear the cost of financing. The lack of long
term markets denominated in reais discussed by Arida et. al. (2005) not only affects the
level of interest rates through the jurisdictional uncertainty argument, but also affects the
day-to-day financing decisions of Brazilian individuals. The implication is that finance is
not available for things that citizens of other countries take for granted: mortgages have
not achieved the penetration of economies with lower interest rates, automobiles and
other consumer durables are largely purchased "d ,irisia" (or for cash), and purchases on
crediario, or in stores financed in "atd 6x sem juros," or in equally divided monthly
installments (usually up to six months) to credit cards without any difference in the
purchase price. All of these alternatives to traditional credit have evolved to fi11 a gap
created by the unattractive nature of typical interest-laden installment contracts.
In the case of mortgage Einance, there are two systems which have evolved to help
allocate capital to potential homeowners. An official government entity known as the
Sistema Financeira da Habitagdo, the Housing Finance System or SFH, is the main
system in the mortgage market. This system sets strict limits on the value of houses
available for Einance, and on the value and conditions of the loans. The Einancing for this
system comes largely from the Fundo de Garantia, a compulsory savings account akin to
Social Security in the United States. The original idea for this system was to offer below-
market level interest rates for housing Einance for lower income Brazilians. The reality,
as documented by Fagnani (1996), is that a regressive allocation of housing Einance
resulted, with upper-middle class and upper class Brazilians receiving most of the
The fund also experienced cash-flow cyclicalities: with contributions for fresh
capital stemming from the worker' s fund and performance of loans linked to the
unemployment rate, during the volatile periods of the 1980s and 1990s the fund was not
able to provide necessary Einancing for homes (Borba, 2004). Because capital in the
system became more scarce, interest rates remained high, despite an explicit government
guarantee of the fund, and are now set at the reference rate plus twelve percent, which
STo the best knowledge of the author, this arrangement does not exist in the English-speaking world;
accordingly, the Portuguese word will be used throughout the text when discussing this form of borrowing.
currently amounts to over thirty percent annually (Pinheiro and Cabral, 1999, and current
The j oint effects of the cronyism with which funds were distributed and the cyclical
nature of fund availability has left an unmet demand equivalent to 15% of the existing
housing capacity, 84% of which comes from families which earn less than three
minimum salaries a month and 81% of which consists of urban households, who seek
funding to no avail (Borba, 2004). In short, the reality of this system has been to crowd
out and/or raise the cost of home financing even more for that segment of society least
able to bear its costs.
The second system, known as the Calrteira Hipotecdria, or Mortgage Portfolio,
allows Brazilian banks to set the conditions of loans. The costs are even higher for these
mortgages, as they are funded through popular savings accounts in the commercial banks
that issue them and mortgage rates have historically averaged 14 to 16 percent above the
reference rate, with the difference in rates between the Carteira Hipotecdria and the SFH
stemming from typical commercial banking spreads and the absence of the effective
government subsidy in the administration of the loans (Pinheiro and Cabral, 1999). The
problems with the Carteira Hipotecdria are even more pronounced with respect to
availability, as one must first have an account with a bank to receive a mortgage. In
Brazil commercial banking penetration is below 50% and more than 1/3rd of individuals
who would like a bank account do not have one (World Bank, 2003).
In response to these difficulties, the state of Sho Paulo has initiated its own
program. Financed through a 1% supplementary tax, the Sao Paulo Housing and Urban
Development Corporation has a statutory obligation to provide housing programs and
Einance to families that earn between one and three minimum salaries per month, with
preference to those earning three or fewer. In the years following the implementation of
the policy, new production of houses in Sho Paulo state approached 300,000 units, with
75% of that production benefiting the lowest tier of income earners (Borba, 2004). This
subsidized finance was subj ect to manipulation however, driving applicants to report
lower household incomes in order to receive more subsidies (Borba, 2004). While the
government is actively seeking to correct these distortions at the time of this writing,
programs such as this should be replicated and improved upon in the rest of Brazil to
mitigate the effects of high interest rates on the housing sector.
Everyday Financing for Individuals
In addition to the impact seen in mortgage Einance, high interest rates and credit
inaccessibility have impacts on the more day-to-day Einancing activities of individuals.
To better understand these effects, the BNDES charted a survey in late 2001 in the cities
of Sho Paulo and Recife to capture the demand for financing instruments and the methods
people in the lower classes use to satisfy those demands (Brusky and Fortuna, 2002).
This section leans heavily on their work, and extends it by looking at more recent data
sources. They find that the credit most used by individuals can be categorized along the
lines of formal, semi-formal, and informal, with a second difference existing between
monetary credit and "parceled" credit, or credit that is extended for the explicit purchase
of an asset.
The most formal credit offerings include loans from banks, personal lines of bank
credit, cash advances on credit cards, and loans from microfinance organizations (in the
case of monetary credit), and credit card, post-dated checks, and crediario (equally
spaced charges to a credit card for a fraction of the purchase price) in stores. These
formal lines of credit are governed (as one would expect) by the rules of the Brazilian
Einancing system, and to some extent, by the same rules as corporate lending. Individuals
must be registered with the federal government, show proof of residency, show proof of
income in the formal sector, and have good credit standing (as measured by the credit
protection services SPC and Usecheque) ibidd., p. 17). Loans from this formal
mechanism vary in duration from 1 to 36 months, with interest rates varying between 0
and 1 1.5% per month. Because of their more stringent conditions, and high interest rates,
this method of Einancing is the least used by the lower classes of society.2
The semi-formal sector is much more utilized, especially in the case of parceled
credit. Very few documents are required, and a clean credit history is not required, as the
credit agencies are rarely consulted. What is much more important in this type of
arrangement is the personal relationship between lender and borrower, with the types of
instruments including loans from employers, a~diota~s (or loan sharks), loans from
employee funds, supplier credit (in the case of small firms), and also crediario in
lojinhas, or company stores. The terms are generally shorter (one to four months in
duration), with interest rates running from 0 to 45% per month. While the rates are at
times higher, this form of credit is more utilized because of the less stringent conditions
under which it is granted.
The most commonly used method of Einancing for the masses is that of informal
credit, as classified by Brusky and Fortuna. These instruments include loans from
parents and friends, as well as take-now, pay-later schemes (fiado, or on faith, in
2 An important exception is the case of crediairio, which the authors note is "so important that even those
who do not meet the conditions of use still manage to utilize it, often purchasing products in the names of
the friends or parents." (My translation, ibid., p. 21).
Portuguese), the use of the credidrio of others, and the use of sorteio, or group savings
plans in which members pool their funds for a determined amount of time, then take turns
spending the accumulated funds. A reader less familiar with the Brazilian economy may
ask why these funds are administered by groups, instead of being deposited in personal
savings accounts that can be managed by each of the individuals. The reason is that these
funds exist predominantly in the subsection of society that cannot legally have accounts
at financial institutions. The terms of all of these arrangements are favorable, with
interest rates between 0 and 5% per month, for one month, and no required
documentation. The downside is that the limits of borrowing are generally lower, with
notional amounts rarely exceeding R$200. Despite these small amounts, it is much more
common for individuals to borrow in this form than in more formalized mechanisms.
Despite the high level of interest rates in Brazil, the Brazilian economy has largely
overcome the crippling nature of high interest rates with respect to the credit offerings.
There exist instruments to mitigate this effect and to spread its cost around the economy
(in the case of credidrio, for instance, the real value of the purchase is eroded by interest-
free installments). Pinheiro and Cabral (1999) identify three main institutions that allow
these irregular credit offerings to flourish: a well-developed informational infrastructure
(which essentially allows for the use of credidrio and post-dated checks), the presence of
public banks (which subsidize the cost of capital), and the use of group mechanisms to
ensure repayment (which lower risk of lending to unsecured individuals). These
mechanisms should continue to be utilized, as their existence will allow for even lower
levels of interest rates once the central bank begins to inj ect more monetary stimulus into
the economy. And until interest rates do decline, it should be the priority of all parties to
ensure that these mechanisms only become stronger to ensure some form of credit is
available to those excluded from the formal sector.
The Disparate Effects of High Interest Rates
Brazil's high interest rates have effects on all segments of society and on all sectors
of the economy. Multinational firms must be much more creative in their Einancing
decisions in order to lower their cost of debt, issuing securities in multiple currencies.
These liabilities create a mismatch in revenues and expenses (unless natural hedges exist)
that must either be insured or factored in to the cost of debt. Because of their scope and
size, however, these firms are best equipped to deal with the challenges of prohibitive
domestic credit markets.
Small firms do not enj oy the same fortune. Oftentimes their lending is constrained,
and expansions are limited to those which can be undertaken out of the firm's retained
earnings or from additional paid-in capital from principal investors (most notably family
members). Access to foreign banks is limited for this group, and family members and
other allegiance-based arrangements oftentimes fill the Einancing gap in a less-than-
High interest rates also have a strong impact on individuals. The mortgage
market is largely unfinanced (due to the risk-laden nature of fixed-rate, long-term
contracts) and undesirable (mortgages exceeding 25% annually need incredible land
appreciation rates to be justifiable!). Even the purchases of domestic appliances and
clothing are affected by these high interest rates, which (through the reference rate of the
economy) affect the APRs on credit cards and other consumer debt instruments. Due to
this otherwise activity-stopping effect, the economy has evolved with respect to
consumers. Certain debt instruments rely on family or personal allegiance to lower the
perceived risk for the creditor, while other arrangements use group collateral or savings
plans to spread the risk across individuals. Out of the necessity to sell their merchandise,
department stores also become financing organizations, selling merchandise on crediciio,
allowing consumers to pay in installments, while transferring possession to (though not
title of) the merchandise on the day of the sale. The burgeoning nature of these
arrangements should be supported and will help to foster an even more diverse financing
environment once interest rates move to more tolerable levels.3
The Brazilian economy has learned to cope with the reality of high interest rates.
Without a doubt, however, these rates still have significantly negative effects on the
economy. It is a testament to the spirit of the Brazilian people that economic growth has
occurred in spite of the high cost of capital. One can only imagine what the ingenious
Brazilian spirit could do in a more fertile economic environment.
3 Some of these instruments will likely disappear, as well, as there will be little or no need for certain of
these instruments at lower levels of interest.
CONCLUSIONS, RECENT OB SERVATIONS, AND FINAL REMARKS
This paper has shown that Brazil's economic history and structure have historically
demanded the persistence of high nominal and real interest rates. Those rates have
principally been necessary to combat inflation, though in periods of fiscal expansion they
have, at times, fallen pro-cyclically, with expansions in the money supply merely
accommodating the aggressive fiscal stance of government. The economy has been
rocked by maj or shocks, not the least of which were the oil crises of the 1970s and 80s
and, more recently, turmoil in emerging markets that all threatened contagion. These
shocks have propagated through the real sector of the economy, as well as adding
The political economy has been a veritable laboratory for economic theory and
practice. Incredible attempts at expansion and development led to overheating because of
lack of external Einancing. The internal Einancing necessitated by this otherwise
unfinanced policy resulted in hyper-inflation that was combated both with orthodox
(austerity) and heterodox (manipulative) measures. Various attempts at stabilization
failed by not addressing the most endemic problems of the economy, which could all be
traced to theoretically apparent relationships: budget deficits, lack of regime credibility,
self-fulfilling expectations and inflationary inertia. The plan under which the economy
operates at the time of this writing has succeeded in stabilizing the economy in the last
eleven years, bringing low-inflation growth, coupled with productivity advancements.
A review of the causes of high interest rates
Interest rates have not fallen from their historical levels. This paper has shown that
the rates, whether measured in nominal or real terms, exhibit stickiness with respect to
economic fundamentals and when measured vis-a-vis other economies with similar
economic histories. In fact, countries whose crises have influenced the level of interest
rates in Brazil seem to recover to neutral policy rates following those crises more quickly
than does Brazil (which is only affected through contagion). When measured in nominal
terms, they highly reflect the public' s expectations of future inflation. Those
expectations have historically been overly optimistic, though recently there is evidence
that they are more centered on inflation's true, ex-post realized values. These
expectations are driven by past inflation (the inflationary inertia effect) and by other
observable economic phenomena. More recently, however, they have been dominantly
determined by government policies such as the fiscal stance, the announced inflation
target, and the government's inter-bank reference rate, a taxa Selic. This suggests that
the factors determining inflation are well within the government' s control, which in turn
suggests that the level of nominal interest rates could be lowered following the proper
macroeconomic recipe at the very least by lowering the public's expectation of inflation.
Several factors influence the real interest rate as well. Some have argued that
monetary policy needs to be highly contractionary in order to quell inflation. These
theorists point to an equilibrium, non-accelerating inflation rate of interest of nearly 18%.
These statements are derived from an ex-post analysis of the data, and do not give us any
insights into why interest rates have to be so high in the first place. Along the same lines,
some have argued that the weak constitutional independence of the Brazilian Central
Bank requires that interest rates be maintained at levels higher that those which would be
otherwise required, in order to send clear signals to the market of the Bank' s commitment
to the often politically unpopular Eight against inflation. This effect has been quantified
both on its influence on interest rates and indirectly on its influence on inflation. Brasilia
would be wise, then, to grant the bank constitutional independence, giving governors and
the President of the Bank irrevocable terms of service that do not coincide with the
More analysts of this question have pointed to the fiscal stance of government and
the debt dynamics of the nation when seeking to explain high interest rates. Because
public indebtedness is high, a high default probability is assessed, pushing up the
endemic country risk. Furthermore, as a large portion of that debt is denominated in
dollars, and the real has not been stable, the implied revenue/liability mismatch further
adds to perceived risk. There is no doubt that the level of government debt is high in the
country, and the government should continue to maintain primary surpluses in order to
pay down that debt.
The insertion of the Brazilian economy into the global economic scene has long
been cited as a cause of high interest rates. Maintaining a current account deficit puts
significant downward pressure on the exchange rate, endangering price stability through
exchange rate pass-through and adding to the equilibrium level of external Einancing
requirements of the economy. The low-liquidity of Brazilian currency internationally has
also been cited historically as a cause of high interest rates, but work using more recent
data has rej ected that assertion.
Finally, there is significant evidence from Brazil and other emerging economies
that there is an "original sin" from issuing government debt denominated in dollars.
Once that dollar debt is available, aversion to local currency issuance lowers the liquidity
of the existing debt, and raises the market-clearing interest rate for that debt as well. In
the Brazilian case, it is argued that this effect has been so strong as to eliminate the
possibility of long-term, local currency-denominated debt. Existing debt, moreover, is
deemed so unattractive as to raise the risk associated with adjudication and valuation, a
risk known in the literature as jurisdictional uncertainty. In Brazil, the institutions that
would eliminate that adjudication risk are weak and should be strengthened in order to set
the stage for a robust local, long-term market for credit.
All of the relevant determinants of interest rates are brought together in the
flowchart at right. The model is an Figure 5.1 Flowchart of Interest Rate Determinants
extremely interconnected one, and
Lack of Regime Lack of Central Ifainr
Credibility / Bank
steps to delink these causal Regime Incipience Independence Mmr
mechanisms will take time. Public
Inflation / Fiscal
Building regime credibility, for Policy
instance, will only lead to lower
Pass- I Inflation 0
interest rates once the public factorS ~, hrough
that credibility into its expectations,
and once those new expectations Target /' IDemand
lead to lower inflation, which will
in allow the Central Bank to lower
The Show Must Go On: Evolutions of Very "Brazilian" Credit Mechanisms
The marketplace has evolved in Brazil to mitigate the effects of these high interest
rates. Large firms effectively bypass the problem by raising money abroad. Those with
access to these loans often times make acquisitions abroad just to diversify their asset
base and to be able collateralize new obligations in dollars. These same firms often
qualify for sub-market rate loans from development banks and organizations, most
notably the Brazilian state-run national development bank, the BNDES. These
mechanisms allow these firms to lower their cost of debt and continue to finance their
Smaller firms are somewhat more constrained. Expansion in these firms is
somewhat limited by profitable activities; investment activities in these firms are
disproportionately financed from retained earnings. Furthermore, relationship finance
(family members, friends, etc) is extremely important to these firms. The implication of
this finance is clear, however: there is no market mechanism or due diligence to filter out
bad investment ideas and the resulting allocation of credit is often times highly
For individuals, access to credit is varied. Bank penetration is low, and is often
times limited to those individuals in the formal sector of the economy. Mortgage credit is
often subsidized but ironically limited to the wealthy, or those with special connections
because of the bureaucratic nature in which the credit is issued. There is evidence,
however, that this is improving, as mortgage funding increased 55% in 2005 from the
previous year, though still not enough to cover the housing deficit (PREI, 2006).
Lending has increased an average of 40% annually since 2003, when the Central Bank
began to put the pressure on commercial banks to provide funding (Inter-American
Dialogue, February 2006). This growth cannot continue ad infinitum, however,
especially if interest rates do not fall further.
Smaller financing decisions by individuals are also affected by these interest rates.
Low bank penetration has obviated the use of extensive use of credit cards, though those
that have bank accounts often times finance through post-dated checks, credit cards, and
lines of credit. Those that do not make use of store finance options, use collective
arrangements, or use the credit of family and friends. While all of these options have
enabled the economy to continue to function, lower interest rates coupled with increased
personal banking would allow the economy to perform much more efficiently and would
lead to more equality of economic opportunity.
Epilogue: Recent Data and Trendsl
In the most recent economic data, there is reason to believe that the forces are in
play to enable a measured lowering of the key interest rate in Brazil. Inflation has been
low, well within the range of the Central Bank' s targets since early 2004, with consensus
estimates for inflation in 2006 in the mid-point of the Bank' s range. Year-over-year real
GDP growth continues to be strong, with the most recent reading of 3.1% in third quarter
of 2005 almost 75% driven by internal demand. Unemployment was categorically lower
in all of 2005 than in the previous four years, with the December reading of 8.3% the
lowest Brazil has seen in that entire period; formal job creation exceeded 1.5 million in
the 12-months to December 2005, while an expansion of investment created expanded
industrial output without inflation-building pressures on capacity, which saw utilization
of 81% in the most recent data. The announced targets for inflation are 4.5 for the next
SThis section presents data that is public and widely available, most readily from the website of the
Brazilian Central Bank (www.bcb.gov.br).
two years, and the Bank seems to have achieved credibility, as consensus IPCA data is
for inflation of 4.7 and 4.5% in 2006 and 2007, respectively; expectations have exhibited
a nearly monotonic decline since 2004 and are now approaching levels of more advanced
economies. The strengthening in the real has continued, with an additional 10%
appreciation since the beginning of 2006. This has not affected the exportability of the
Brazilian economy, however, as export prices have nearly doubled since the beginning of
2003 and export growth has accelerated markedly, creating a current account surplus for
the first time in recent history in 2003; that surplus has been sustained since 2003 and hit
a record of 45.4 billion dollars in 2005.
Public debt management is also improving. 2005 saw a primary surplus of 4.8% of
GDP, and the average for this surplus was 4.5% in the period 2004-2006. The debt/GDP
ratio closed the year at 51.6%, down from a ten-year high of over 60% in 2002. Market
expectations are for a further improvement, with forecasts of around 45% by 2009. The
portion of the debt that is linked to foreign currencies is declining, and is quickly
approaching zero: reductions of 9.4, 25.7, and 25.3 billion dollars in the most recent three
years have brought the level of foreign currency denominated debt to less than 5%, down
from over 40% in 2002. This has not entirely taken place through the retirement of the
debt, as part of this reduction has been achieved through the roll-over of that debt into
local-currency denominated bonds, the first one being issued in September of 2004. This
is a significant achievement for Brazil, which previously suffered from the "original sin"
of dollar-denominated debt (Hausman et al., 2003). This change has greatly improved
Brazil's resilience to exchange rate shocks: the Central Bank now estimates that a 1%
change in the exchange rate would have less than a 5 basis point impact of the debt/GDP
ratio, down from a high of 35% in 2002 (Bevilaqua, 2006).
There is some evidence that interest rates have begin to fall. The market-measured
"Brazil risk" is now half of what it was in early 2004 and the real interest rate (as
measured by 360-day swaps) is near 11%, down from an average of 15.2% for the same
measure in 2000-2003. The January 2006 meeting of the monetary council lowered the
Selic rate from 18% to 17.25%, followed by another cut in March 2006 to 16.5%, with
indications that further rate cuts may follow. Time will tell if these rate cuts materialize
and make rates in the country more comparable to that of its peers. Brazil's commitment
to sound policies may be rewarded in the long-run, then, and may soon be a textbook
example of slow adjustment rather than one of irrational pessimism and undue
LIST OF REFERENCES
Almeida, Caio Ibsen Rodrigues de. 2004. "Interest Rate Risk Measurement in Brazilian
Sovereign Markets." Estudos Econdmicos, 34:2 (Apr-June 2004), 321-344.
Araujo Junior, Eurilton Alves and Alexandre Barros da Cunha. 2003. "Brazilian
Inflation and GDP from 1850 to 2000: An Empirical Investigation." Estudos
Econdmicos, 33:3 (July-Sept 2003), 399-433.
Arida, Persio and Andre Lara-Resende. 1985. "Inertial Inflation and Monetary Reform in
Brazil." In John Williamson (Ed.),1Inflation andhidexation, Washington, DC:
Institute for International Economics.
Arida, Persio, Edmar Bacha and Andre Lara-Resende. 2005. "Credit, Interest, and
Jurisdictional Uncertainty: Conj ectures on the Case of Brazil." hIstituto de Estudos
de Political Econontica, Casa das Garcas, Rio de Janeiro, Brazil.
Armijo, Leslie Elliott. 1996. "Inflation and Insouciance: The Peculiar Brazilian Game."
Latin American Research Review, 3 1:3 (1996), 7-46.
Aronovich, Selmo. 1994. "Uma nota sobre os efeitos da inflaCio e do nivel de atividade
sobre o "spread" bancario." Revista Brasileira de Econontia, 48:1 (Jan-Mar 1994),
Bacha, Edmar. 1983. "Vicissitudes of Recent Stabilization Attempts in Brazil and the
IMF Alternative." In John Willliamson, ed., IM~F Conditionality. Cambridge, MA:
Baer, Werner, and Donald V. Comes. 1990. thrited States Policies and the Latin
American Economies, New York: Praeger.
Barberis, Nicholas and Richard Thaler. 2004. "A Survey of Behavioral Finance." in G.
Constantinides, M. Harris, and R. Stulz (eds.) Handbook of the Economics of
Finance. Amsterdam: JAI Press.
Bevilaqua, Alfonso. 2006. Brazil: Recent Economic Developments. February 2006.
Banco Central do Brasil website: www.bcb.gov.br. Last Accessed March, 2006.
Bogdanski, Joel, Paulo Spriger de Freitas, Ilan Goldfajn, and Alexandre Tombini. 2001.
"Inflation Targeting in Brazil: Shocks, Backward Looking Prices, and IMF
Conditionality." Central Bank of Chile, Working Paper Series No. 110 (Santiago).
Bonomo, Marco Antonio Cesar, Betina Guimaries Dodsworth Martins and Rodrigo
Ribeiro Antunes Pinto. 2004. "Debt composition and exchange rate balance sheet
effects in Brazil: A firm level analysis." Economics Working Papers (Ensaios
Economicos da EPGE) 535, Graduate School of Economics, Getulio Vargas
Borba, Maria Olyntha. 2004. "Housing Finance Policy for Low-Income Families The
Case of the State of Sho Paulo, Brazil." Speech delivered to the United Nations
Joint Congress for Housing Finance in Emerging Markets in Berlin, Germany.
Bulmer-Thomas, Victor. 2003. Thze Economic History of Latin America Since
Independence. Cambridge Latin American Studies. Cambridge, England:
Cambridge University Press.
Brusky, Bonnie and Joho Paulo Fortuna. 2002. Entendendo a demand para as
microfinancas no Brasil: Um estudo Qualitiva em Duas Cidades.dddd~~~~~~dddddd Rio de Janeiro:
Campilo, Ana Katarina and Francisco Cribari Neto. 2003. "Inflation Inertia and Inliers:
The Case of Brazil." Revista Brasileira de Economia, 57:4, part 1 (Oct-Dec 2003),
Cardoso, Eliana. 1983. "A Money Demand Equation for Brazil." Journal of
Development Economics 12 (1): 193-93.
Cardoso, Eliana A., and Albert Fishlow. 1990. "The Macroeconomics of the Brazilian
External Debt." Part 3, Vol. 2, Developing Country Debt and Economic
Performance, edited by Jeffrey Sachs. Chicago: University of Chicago Press.
Cameiro Netto, Dionisio Dias and Thomas Yen Hon Wu. 2004. "Contas externas e
political monetaria." Revista Brasileira de Economia, 58:3 (July-Sept 2004), 301-
Carvalho, Fabia A. 2004. "Testing the Rationality of Expecations on Brazilian Inflation :
Evidence from an Unbalanced Panel." University of Brasilia Department of
Economics Working Paper no. 307.
Cassettari, Ailton. 2002. "Um estudo empirico da evolugho das taxas de juros de curto
plazo no Brasil desde o "Plano Real"." Estudos Econdmicos, 32:3 (July-Sept 2002),
Cerisola, Martin and R. Gaston Gelos, 2005. "What Drives Inflation Expectations in
Brazil? An Empirical Analysis," IMF Working Papers 05/109, Intemnational
Costa, Fernando Nogueira da et al. 2001. "Meta inflacionaria, juros e pregos no varej o
brasileiro." Revista de Economia Politica (Brazil), 21:4 (Oct-Dec 2001), 93-111.
Da Cunha, Barbara Vasconcelos Boavista and Pedro Cavalcanti Ferreira. 2004. "Custo
de Ciclo Econimico no Brasil em um Model Com Restrigio a Credito." Estudos
Econdmicos, vol 34:2.
De Medeiros, Otavio and Carmem Tiberio. 2005. "Factors Influencing Brazilian Firms in
their Decision to List on Foreign Stock Exchanges," Finance, Economics Working
Esterl, Mike. 2002. "Stiglitz: 'Excessive Pessimism' Directed At Brazil." Dow Jones
International News. October 21, 2002.
Fagnani, Eduardo. 1996. "Politica Habitacional." Boletint de Conjuntura /Politica
Social, 21:1. Edigies FUNAP.
Faria, Lauro Vieira de. 2003. "Financial Markets, External Shocks, and Policy
Responses: The Case of Brazil, 2001." Revista de Econontia Politica (Brazil), 23:4
(Oct-Dec 2003), 43-57.
Faro, Cl6vis de. 1994. "A constituigio, os juros e a economic." Revista de Econontia
Political (Brazil), 14:1 (Jan-Mar 1994), 43-52.
Fasolo, Angelo Marsiglia and Marcelo Savino Portugal. 2004. "Imperfect Rationality and
Inflationary Inertia: A New Estimation of the Phillips Curve for Brazil." Estudos
Econdmicos, 34:4 (Oct-Dec 2004), 725-776.
Favero, Carlo A. and Francesco Giavazzi. 2002. "Why are Brazilian Interest Rates so
High?" Manuscript. IGIER, Universita Bocconi, Milan, Italy.
Federal Reserve Board. 1997. "The Role of Expectations in the FRB/US Model." The
Federal Reserve Bulletin, April, 1997.
Feliz ,Raul A and John Welch. 1992. "Cointegration and Tests of a Classical Model of
Inflation in Argentina, Bolivia, Brazil, Mexico, and Peru." Research Department,
Federal Reserve Bank of Dallas.
Ferriera, Pedro and Jose Rossi. 2003. "New Evidence from Brazil on Trade
Liberalization and Productivity Growth." hIternational Econontic Review, 44:4
Fraga, Arminio. 2005. "Comment on Chapter 4" in Giavazzi, Goldfajn, and Herrera,
eds., Inflation Targeting, Debt, and The Brazilian Experience, 1999 to 2003.
Cambridge, Mass: MIT Press.
Fraga, A., Goldfajn, I., and Minella, A. 2003. "Inflation Targeting in Emerging Market
Economies," Working Paper of the Central BankBBBBB~~~~~~~BBBBBB ofBrazil, Number 76.
-----. 1994. "The Formation of Inflation Expectations in Brazil: A Study of the Futures
Market for the Price Level." Revista Bra~sileira de Econontia, 48:1 (Jan-Mar 1994),
-----.2003. "Brazil in the 21st Century: How to Escape the High Interest Rate Trap."
Textos para Discussio, PUC-Rio. Number 466.
Garcia, Marcio Gomes Pinto and Tatiana Didier. 2000. "Very High Interest Rates and
the Cousin Risks." Textos para Discussio, PUC-Rio. Number 441.
-----. 2003. "Taxa de juros, risco cambial e risco Brasil." Pesquisa e Plan2eja~nento
Econdmico, 33:2 (Aug 2003), 253-297.
Hausman, Ricardo, Barry Eichengreen, and Ugo Panizza. 2003. "Currency Mismatches,
Debt Intolerance, and Original Sin." Cambridge MA: National Bureau of
Economic Research Working Paper 1003 6.
Institute Brasileira de Geografia e Estatistica. Anuario Estatistico do Bra~sil. Rio de
Janeiro. Various Issues.
Inter-American Dialogue. 2006. The Latin America Banking Advisor: February 2006.
Kanczuk, Fabio. 2002. "Juros reais e ciclos reais brasileiros." Revista Bra~sileira de
Econontia, 56:2 (Apr-June 2002), 249-267.
Keane, Michael P. and David E. Runkle. 1990. Testing the Rationality of Price
Forecasts: New Evidence from Panel Data." American Economic Review, Vol. 80,
No. 4 (Sep., 1990), 714-735.
Lima, Alexandre Maia Correia and Joho Victor Issler. 2003. "A hip6tese das expectativas
na estrutura a termo de juros no Brasil: uma aplicaCgo de models de valor
presente" Revista Bra~sileira de Econontia, 57:4, part 2 (Oct-Dec 2003), 873-898.
Lima, Elcyon Caiado Rocha. 2003. "The NAIRU, Unemployment, and the Rate of
Inflation in Brazil." Revista Bra~sileira de Econontia, 57:4, part 2 (Oct-Dec 2003),
Minella, Andre. 2003. "Monetary Policy and Inflation in Brazil, 1975-2000: A VAR
Estimation." Revista Bra~sileira de Econontia, 57:3 (July-Sept 2003), 605-635.
Moreira, Edison Marques. 2002. "Politica monetaria: a redugho cautelosa da taxa de
juros Selic no primeiro semestre de 2002." hIdicadores Econdmicos FEE, 30:2
(Sept 2002), 23-34.
-----. 2003. "Politica monetaria: os juros continuam altos no primeiro semestre de 2003."
Indicadores Econdmicos FEE, 31:2 (Aug 2003), 11-25.
-----. 2004. "Politica monetaria: a flexibilizagno no segundo semestre de 2003."
Indicadores Econdmicos FEE, 31:4 (Feb 2004), 35-52.
-----. 2004. "Politica monetaria: no primeiro semestre de 2004, Copom mantem cautela
na redugho da taxa basica de juros." Indicadores Econdmicos FEE, 32:2 (Aug
Muinhos, Marcelo and Marcio I Nakane. 2005. "Comparing Equillibrium Real Interest
Rates: Different Approaches to Measure Brazilian Interest Rates." BancoB~~~~BBBBB~~~~BBBB Central
do Bra~sil Working Paper.
Palley, Thomas I. 2004. "Escaping the Debt Constraint on Growth: A Suggested
Monetary Policy for Brazil." Revista de Economia Politica (Brazil), 24: 1 (Jan-Mar
Pastore, Affonso Celso. 1990. "InflaCio e expectativas com a political monetaria numa
regra de taxa de juros." Revista Bra~sileira de Economia, 44:4 (Oct-Dec 1990), 499-
Pinheiro, Armando Castelar and Celia Cabral. 1999. "Credit Markets in Brazil: The
Role of Judicial Enforcement and Other Institutions." Interamerican Development
Bank Working Paper R-3 68, June 1999.
Portugal, Marcelo Savino and Regina C. Madalozzo. 2000. "Um modelo de NAIRU para
o Brasil." Revista de Economia Politica (Brazil), 20:4 (Oct-Dec 2000), 26-47.
Prudential Real Estate Investors (PREI). 2006. Latin American Quarterly, January,
Rossi, Jose W. 1989. "Comportamento dos Agregados e Multiplicadores Monetarios no
Brasil." In IPEA, Perspectiva~s da Economia Bra~sileira 1989. Rio de Janeiro:
-----. 1996. "A estrutura a termo da taxa de juros: uma sintese." Pesquisa e Plan2ejamnento
Econdmico, 26:2 (Aug 1996), 521-547.
Saad-Filho, Alfredo and Lecio Morais. 2002. "Neomonetarismo tropical: a experi~ncia
brasileira nos anos noventa." Revista de Economia Politica (Brazil), 22: 1 (Jan-Mar
Skidmore, Thomas E. 1988. Thze Politics of2~ilitary Rule in Brazil, 1964-85. New York,
Oxford University Press.
Suplicy, Eduardo Matarazzo. 1997. "O real e as propostas para melhorar a distribuigio da
renda e erradicar a miseria." Cultura Vozes, 91:5 (Sept-Oct 1997), 121-128.
The State Department of the United States of America. 2005. "The 2005 Investment
Climate Summary for Brazil." http://www. state.gov/e/eb/ifd/2005/4 1988.htm.
Last accessed March, 2006
Tabata, Alicia and Benj amin Miranda Tabak. 2004. "Testes do conteutdo informacional
das decisies de political monetaria." Pesquisa e Plan2eja~nento Econdmico, 34:2
(Aug 2004), 207-249.
Tejada, Cesar Augusto O. and Marcelo Savino Portugal. 2001. "Credibilidade e inercia
inflacionaria no Brasil, 1986-1998." Estudos Econdmicos, 31:3 (July-Sept 2001),
Toledo, Joaquim E16i Cirne de. 2002. "Risco-Brasil: o efeito-Lula e os efeitos-Banco
Central." Revista de Econontia Politica (Brazil), 22:3 (July-Sept 2002), 138-145.
Tullio, Giuseppe and Marcio Valerio Ronci. 1996. "Brazilian Inflation from 1980 to
1993: Causes, Consequences, and Dynamics." Journal of Latin American Studies,
28:3 (Oct 1996), 635-666.
The World Bank. 2003. "Access to Financial Services in Brazil." A presentation given
to the Banco Central do Brasil, November 10, 2003.
http://www.bcb .gov.br/pre/SeMicro/Palestras/02_3_Suzanapt Last accessed
Ed Johnson will graduate in May, 2006 from the University of Florida with a
Master of Business Administration, with concentrations in finance and competitive
strategy, and with a Master of Arts in Latin American studies, with a concentration in the
Latin American business environment. His passion is Latin America; he has made
several trips to the region and speaks Spanish and Portuguese. Before graduate school,
he was a research assistant for the Federal Reserve Board of Governors in Washington,
DC, working on macroeconomic forecasting and analysis for the Federal Open Market
Committee. While at Florida he has worked for ExxonMobil Corporation and Johnson
and Johnson, principally on inventory and supply chain initiatives. Upon graduation he
will join Deloitte Consulting' s Atlanta-based Strategy and Operations practice as a Senior
Strategy Consultant. He holds a B.A. in economics and Spanish from the College of
William and Mary in Williamsburg, VA. A native Virginian, Ed currently resides in