Could the Stock Indexes, Country Risk Proxies and ADRs have Anticipated the Crises in Argentina, Brazil and Mexico in the 90's

International Research Journal of Finance and Economics
ISSN 1450-2887 Issue 15 (2008)
© EuroJournals Publishing, Inc. 2008
http://www.eurojournals.com/finance.htm
Could the Stock Indexes, Country
Risk Proxies and ADRs have Anticipated the Crises in
Argentina, Brazil and Mexico in the 90´s
Franco Parisi
Associate Dean, School of Finance and Management Department
Universidad de Chile, EuroAmerica Chair in Finance
Diagonal Paraguay 257, ofice1102, Santiago, Chile
E-mail: fparisi@neogcios.uchile.cl
Tel: 56-02-678-3366; Fax: 56-02-222-0639
Roberto Friedmann
Terry College of Business, University of Georgia, March, 2007
Abstract
The crises in Argentina, Brazil and Mexico have been widely studied, indicating
that their causes were, to a great extent, unanticipated. In this paper we analyzed if the
stock indices, country risk proxies and ADRs anticipated the critical financial events lived
by those economies during the 90´s. Our results are mixed; in that neither one of these
variables could anticipate the crises in all three economies. In particular, the local stock
market index anticipated the crisis only in the Brazilian case within 120 days previous to
the event. The country risk proxy, as measured by the Brady Bonds appears to be a good
indicator that anticipated the crisis in Argentina and Mexico, but not for the Brazilian case.
However, using the EMBIG as an alternative country risk proxy, the crisis in Brazil could
be anticipated, but this did not hold true for Argentina and Mexico. The ADRs portfolios
from Argentina and Brazil did not present negative CARs in statistical terms in any
window, implying the ADR portfolio did not anticipate the crisis. The combination of
macroeconomic and international economics aspects, together with the selected financial
factors would then be the key to anticipate crisis in Latin American. Yet, as pointed out
below, the mixed results suggest a need for further analyzing these variables and their
ability to predict crises.
Keywords: Latin America, Stock index, ADRs, Country risk proxies, crisis, Argentina,
Brazil, Mexico.
JEL Classification Codes:
I. Introduction
Given their remarkable effects and subsequent hardships in the entire region, the financial crises that
Latin America experienced during the last decade have been the focus of several studies. The general
aim has been to try to identify the causes and factors that lead the collapse of these economies. In this
paper, the financial crises in Argentina, Brazil and Mexico are analyzed to see if the local stock market,
the country risk proxies, and the ADR portfolios could have been utilized to anticipate the crises, in

---

International Research Journal of Finance and Economics - Issue 15 (2008)
114
line with the thinking and empirical evidence one can derive from works such as Esquivel and Larrain
(1998), García and Valpassos (1998), Ganapolsky and Schmukler (2001), Bustelo, García and Olivié
(1999), Buscaglia (2003) and Bustelos (2004).
According to several authors, the 1994-1995 Mexican crisis, like the Asian one, was not related
to fiscal problems, but rather, to consumption problems due to the "boom" of private credit which was
present from the beginning of the term. Marginal factors that have also been suggested as facilitating
this particular crisis were the oil price decline pressures, and a weak banking sector.
In addition to the Asian crisis, another economic event that very seriously influenced Latin
America was the Russian crisis of August 1998, which, as indicated by Malki (1999), resulted from
fundamental domestic weaknesses known a-priori. In particular, Goldfajn and Baig (2000) suggest the
Russian crisis hurt the country substantially, and its negative effect reached other countries, especially
Brazil, triggering a financial crisis that culminated with the flotation of the Brazilian real in 1999 and
the subsequent significant devaluation of its currency. In addition, portfolio managers in Russia were
forced to drop positions in Brazilian bonds; so, while the devaluation in Russia took place, the
Brazilian Brady Bond prices fell along with a subsequent loss in reserves, which in turn, added to the
panic on foreign and local investors. Investigating the possible roots of the Brazilian crisis, Carneiro
and García (1994), Barcinnski and García (1998) and García and Valpassos (1998), stated that capital
flows were an important factor, combined with a very restrictive monetary policy-- since a great
interest rate differential generated a massive short-term inflow to Brazil, and the size of this inflow
magnified two macroeconomic problems: 1) an increased fiscal deficit due to interest payments on the
debt, used to sterilize the incoming flows; and 2) it worsened the valuation of the currency.
Argentina, was not exempt of the consequences of these episodes, becoming a very special case
and therefore addressed by a significant number of studies such as: Ades and Buscaglia (1999); De la
Torre, Levy and Schmukler (2002); Bambari, Saront and Tommasi (2002); Feldstein (2002); Krueger
(2002) and Buscaglia (2003). This body of work comments on the repercussions that other financial
crises had on this country, as well as its internal political problems, and the resulting great vulnerability
of the economy. The Argentinean crisis was unleashed due to its dependence on the Brazilian market,
along with problems anticipated by the Asian crisis, and capped by a combination of a weaker regional
demand and a loss of competitiveness, which subsequently deteriorated its external accounts. The
periods of impressive Argentinean growth and financial stability were more than clouded by what
would be its greatest economic depression. Buscaglia (2003), for example, explains the causes of the
Argentinean crisis based on the fiscal deficit, the fixed exchange rate, and the operations of
international financial markets. His argument, is that the absence of complementary reforms prevented
the possibility of adjustments within the convertibility regime, and that the Argentinean political forces
were incapable of reaching inter-temporary agreements to give reliable commitments to the pegged
currency. Bustelo (2004) though, suggests that the Argentinean crisis was the result of an
incompatibility between a very expansive fiscal policy and the fixed exchange rate regime.
A variety of studies have compared the Argentinean and Asian crises a (e.g., Palma (1998),
Chang and the Velascos (2001), Bustelos (2002), and Nishijima (2002), and Bustelos (2004))
suggesting significant differences between them. In the case of the Argentinean crisis, the thinking is
that it advanced in slow-motion, was prolonged, and perfectly expected since at least 1999; which is
certainly not the case with the Asian crisis.
Chang and Velasco (1998), Bustelo (1998), Bald (1996), Millar and Weber (1999), Chowdury
(1999), Corbett and Vines (1999), and Bustelos, García and Olivié (1999) all use models that explain
the Argentinean crisis as the result of: 1) incompatibility between economy and politics, 2)
unpredictable contingencies and/or 3) a spillover effect derived from country to country events. These
three arguments, are used by the Bustelos, García and Olivié’s (1999) model for the Argentinean,
Brazilian and Mexican crisis.
Building from this prior research, this paper addresses whether or not the local stock market
indices, the country risk proxies and the ADR portfolios of these countries would be able to anticipate
these crises, reflecting prices corrections. Why is this worth considering? Given our goal to enhance

---

115
International Research Journal of Finance and Economics - Issue 15 (2008)
our understanding of these severe crises, the merits of addressing them via three financial measures
(stock indices, country risk measures, and ADRs) is based on the fundamental hypothesis that financial
markets are characterized by being deep and well-informed. One could thus reasonably expect, that
they should have been able to anticipate this type of crisis, or at the very least, they could have
suggested them to some extent.
Our results are mixed; in that not one variable could anticipate the crises in all three economies.
In particular, the local stock market index anticipated the crisis only in the Brazilian case within 120
days previous to the event. The country risk proxy, measured by the Brady Bonds, appears as a good
indicator that anticipated the crises in Argentina and Mexico, but not for the Brazilian case. However,
using the EMBIG as an alternative country risk proxy, the crisis in Brazil could be anticipated, though
not for Argentina or Mexico. The ADRs portfolios from Argentina and Brazil did not present negative
CARs in statistical terms in any window, implying the ADR portfolio did not anticipate the crises.
II. Data and Methodology
The data used in our paper was extracted from five sources. The first was the daily closing prices for
the three stock indexes, MERVAL, BOVESPA and IPC, belonging to the financial markets of
Argentina, Brazil and Mexico respectively. The second source of data corresponded to the daily
closing prices of the S&P500 stock index, including the same periods as the ones for each market index
mentioned. The third source was the “last” weekly prices for the Brady Bonds belonging to the
countries studied, using an average of the different types of Brady Bonds for each country. The fourth
data set, was the daily EMBIG index for Argentina, Brazil and Mexico. The final data set are the daily
returns from the closing prices of the Argentinean and Brazilian ADRs traded in the NYSE.
To use series of complete data for the variables and for the S&P500, two methods were used.
The first method consists of the "omission" of the data that does not appear in both of the indexes that
are going to be compared (in other words, those dates that don’t appear in the local index or in the S&P
500 where eliminated); the second method consists of the "repetition" of the last data available for the
missing data. The cases in which data for a certain day was missing from both indexes, that date was
eliminated from the sample. After collecting the information relative to the market indexes of each
country and the S&P500, the daily return rates were calculated.
To analyze each financial market before the crises, the appropriate relevant periods were
considered. Specifically, in the case of the Argentinean data: from January 1999 to December 2002; the
Brazilian data was analyzed from January 1998 to December 2001; and the Mexican data, from
January 1992 to December 1995. The key events that we will study are two for Argentina: 1) the
"Corralito" which occurred on December 1st 2001 which was a Saturday -- (the following work day,
which compromised the Argentinean ADRs, was Monday December 3rd 2001); and 2) the devaluation
on Sunday January 6th of 2002, affecting the market at the following work day. For Brazil, the relevant
day of the event is the devaluation of the Real, which was Friday January 15th 1999. In Mexico the
event day was in November, 1994.
To calculate the expected return of the local stock return index, country risk and ADRs in the
days near the event, the Adjusted Market Model fixed by discontinuous transactions (Dimson, 1979)
was used, calculated as indicated in equation (1). The excess daily return, ARit, for each portfolio
studied in day t, is calculated as ARit, = E(Rit) - Rit.
_
_
_
E(R )
*

(1)
it
it
Rmt
it 1* Rmt 1
it 1* Rmt 1
Where Rit is the return i in day t; Rmt is the portfolio market rate of return1 in day t. i and i are
parameters of the Adjusted Market Model, calculated by OLS, adjusting the errors by autocorrelation
and heteroscedasticity according to Newey and West (1987), calculated for the sample periods for each
1
Using the S&P500 return as the market return in the cases of the local index and the ADRs. In the case of the country risk proxies, the market
portfolio was the average return of the proxy from days –241 to 121 in the EMBIG case and from –17 to –31 weeks in the Brady Bonds case.

---

International Research Journal of Finance and Economics - Issue 15 (2008)
116
event, where day 0 is the event day, and where beta is equal to the sum of three betas in t-1, t and t+1.
Later, the daily abnormal returns, ARit, are added up to calculate the CARi, as shown in equation (2).
CARt = ARit
(2)
The cumulative abnormal returns, CARi, are calculated for different windows, as shown in
equation (3).
CAR t, t+k = CARt/n
(3)
The statistical significance test used for each CAR from the day of event t to the day of the
event t+k, CARt,t+k is the one in equation (4), where CAR is the standard deviation of the CAR for the
considered2 sample.
T = CARt,t+k/ k CAR
(4)
The methodology used to determine the presence of abnormal returns in Brady Bond prices
from Argentina, Brazil and Mexico was based on Event Study method, which considers day 0 as the
event analyzed, where the objective is to detect the existence of some meaningful change in the Bonds
prices before the event and measuring it. Since the data collected for these samples is weekly3, the
event 0 will be the week of the announcement of the event, and these windows will include
approximately the same number of days as the event study done for the market indexes of those
countries.
Little it is known about the distribution of returns around an event such as devaluation. Thus,
we needed to use a nonparametric test like the bootstrap methodology developed by Efron (1982).
From the CAR of each sample, we re-sampled 1000 times with replacement. We repeated tests in all
the windows and counted the number of times the results were over the critical value, (95% and 99%
level of confidence). We can thus assess -- in a nonparametric way -- the probability of the event being
relevant for a given period of time, as shown in equation (5).
Qi > critical value/ number of tests
(5)
Qi is the return for the critical event. So, in order to be valid, it must be greater than 90%.
III. Results
We calculated the cumulative abnormal returns, CARit, for the stock markets indexes from Argentina,
Brazil and Mexico, according to the described methodology and different windows. The stock market
indexes results analyzed differed by country. They did not change significantly depending on the
correction method used, with exception of the "Devaluation" in Argentina, where the windows
displayed important differences in their statistical significance between the “Omission” and
“Repetition” methods4. Also, the event "Devaluation" in Argentina displayed different "day 0"
depending on the method used, due to the absence, in both cases, of market information for "the exact
day 0" corresponding to the day of the event announcement.
2
In the case of bonds, the standard deviation used in the significance test is calculated for the period between the weeks–31 and –17 regarding event
day 0.
3
Given the low liquidity of these instruments it was not possible to find data of greater frequency.
4
This may be due to the fact that the sample has a bigger absence of data, which could explain why we find significant differences in the results
depending on the method used.

---

117
International Research Journal of Finance and Economics - Issue 15 (2008)
Table 1: Local stock market Cumulative Abnormal Returns, CAR, in percentages, the t-test is in parenthesis,
for the different windows and the different observation absence correction methods. The CARs are
calculated using the S&P500 return as the market return adjusted by Dimson (1979).
Window
(0,0)
(-1,0)
(-2,0)
(-3,0)
(-4,0)
(-5,0)
(-10,0)
(-20,0)
(-30,0)
(-60,0)
(-90,0)
(-120,0)
Argentina: Corralito
6,56
7,43
11,78
0,41
-1,53
3,18
-3,05
-5,21
-14,72
-32,68
-28,68
-52,04
Omission:
(2.09)
(2.37)
(2.66)
(0.07)
(0.24)
(0.45)
(0.31)
(0.37)
(1.35)
(1.35)
(0.96)
(1.52)
6,51
7,37
11,78
0,31
-1,68
3,06
0,05
-2,76
-15,37
-30,62
-29,21
-56,77
Repetition
(2.08)
(2.36)
(2.66)
(0.06)
(0.27)
(0.44)
(0.01)
(0.20)
(0.90)
(1.26)
(0.98)
(1.66)
Argentina: Devaluation
0,49
5,28
15,19
6,44
24,41
31,78
39,50
47,91
45,14
35,48
24,86
-1,60
Omission
(0.12)
(1.30)
(2.64)
(0.91)
(3.00)
(3.49)
(3.07)
(2.63)
(2.03)
(1.13)
(0.64)
(0.04)
0,35
0,87
5,70
15,06
15,62
7,68
25,06
44,95
56,62
47,8
17,36
16,68
Repetition
(0.09)
(0.23)
(1.05)
(2.27)
(2.04)
(0.90)
(2.07)
(2.63)
(2.70)
(1.62)
(0.48)
(0.40)
Brazil: Devaluation
-12,30
-17,72
-22,29
-25,26 -30,15 -29,65
-31,80
-40,58
-52,85
-44,28
-32,67
-66,91
Omission
(3.04)
(4.38)
(3.90)
(3.60)
(3.73)
(3.28)
(2.49)
(2.24)
(2.39)
(1.41)
(0.85)
(1.51)
-12,32
-17,85
-22,47
-25,42 -30,33 -29,72
-28,1
-37,01
-57,29
-38,45
-32,68
-68,70
Repetition
(3.02)
(4.37)
(3.89)
(3.59)
(3.71)
(3.25)
(2.18)
(2.03)
(2.56)
(1.22)
(0.84)
(1.54)
Mexico: Devaluation
1,96
-2,00
-4,73
-5,72
-6,90
-9,87
-11,48
-7,90
-8,21
-17,21
-10,38
2,58
Omission
(1.36)
(1.39)
(2.32)
(2.29)
(2.39)
(3.06)
(2.52)
(1.23)
(1.04
(1.54)
(0.76)
(0.16)
1,95
-2,01
-4,73
-5,72
-6,90
-9,41
-10,86
-5,85
-11,48
-20,84
-12,95
0,72
Repetition
(1.37)
(1.41)
(2.35)
(2.32)
(2.42)
(2.95)
(2.41)
(0.92)
(1.47)
(1.89)
(0.96)
(0.05)
In the particular case of the stock market index from Argentina (MERVAL), two events were
analyzed: “Corralito” and “Devaluation”. The "Corralito" event displays a significant fall from day -
120, which could indicate that the market anticipated the event. But, two results appear for the
"Devaluation" event, depending on the correction method used. The method of "Omission"
("Repetition") presented a statistically significant CAR between days -30 and –4 (-30 and –10), which
indicates that during that period the market anticipated a negative event such a devaluation, but this
signal was diluted near the event day.
In the Brazilian case, clear evidence can be observed. The market began to anticipate the event
around 30 days before it happened, since the CARs become negative and statistically significant at that
moment, a situation that lasts until the day of the devaluation.
In the Mexican case, when the stock market index is used as a proxy, results show negative
CARs at day -10, but it is diluted in the days previous to the devaluation.
Figure 1: Local Stock Market Cumulative Abnormal Returns, CAR, 120 days before the announcement of
events. The event "Devaluation" for the market index “Merval” displays two different series
depending on the correction method used.

---

International Research Journal of Finance and Economics - Issue 15 (2008)
118
The Brady Bond CARs results from Argentina, Brazil and Mexico, for each event, appear in
Table 2 and Figure 2. The results show clear evidence of a strong fall in Brady Bond’s abnormal
returns for the windows, with exception of the "Devaluation" in Brazil, which displays positive
abnormal returns most of the time, but not for the previous days to the announcement. For Mexico, we
observe negative CARs for each of the windows analyzed (all of them are statistically significant),
which shows that the Mexican Brady Bonds anticipated a possible crisis. Finally, for the Argentinean
case, the results for the two events studied are similar to the Mexican ones, since both display negative
abnormal returns, suggesting that the bond market anticipated the crisis; whereas for the event
"Devaluation" in Argentina, the negative statistical significance of CARs Brady Bond appear between
weeks -17 and the -13.
Table 2:
Brady Bond Accumulated Abnormal Returns, CAR, in percentages. The CARs are calculated using
the Dimson (1979) Adjusted Market Model where the market portfolio was the average of the Brady
Bond return for the period –17 to –31 weeks. t-tests are shown in parenthesis.
Week
(0,0)
(-1,0)
(-2,0)
(-3,0)
(-4,0)
(-8,0)
(-13,0)
(-17,0)
Argentina
-5,31
-11,65
-13,80
-16,87
-16,80
-24,81
-33,15
(-23,86
Corralito
(-1.28)
(-2.81)
(-2.35)
(-2.35)
(-2.02)
(-2.11)
(-2.22)
(-1.39)
4,30
3,98
-1,78
-3,80
-8,94
-19,77
-31,79
-34,62
Devaluation
(1.06)
(0.98)
(-0.31)
(-0.54)
(-1.10)
(-1.73)
(-2.18)
(-2.07)
Brazil
-0,24
-1,41
-0,68
4,16
2,70
7,14
40,60
61,20
Devaluation
(-0.04)
(-0.21)
(-0.07)
(0.36)
(0.21)
(0.38)
(1.71)
(2.25)
México
-11,80
12,51
-13,75
-11,95
-12,90
-14,92
-17,38 (-
-21,91
Devaluation
(-5.67)
(-6.01)
(-4.67)
(-3.31)
(-3.10)
(-2.53)
2.32)
(-2.55)
Figure 2: Brady Bond Cumulative Abnormal Returns, CAR, during a period of approximately 120 days before
the event announcement.
We analyzed the EMBIG index for Argentina, Brazil and Mexico and their respective events, as
proxies of country risk. The EMBIG index was created by JP Morgan Chase Bank to measure the
default risk in emerging economies. The expected abnormal returns around the event for the EMBIG
index were expected to be positive, which would provide evidence of an increase in the country risk, or
default probability, anticipating a crisis. The results on the events studied of the EMBIG abnormal
daily returns for the three countries are shown in Table 3 and Figure 3. Unfortunately, we could not

---

119
International Research Journal of Finance and Economics - Issue 15 (2008)
validate our expectations, since Argentinean and Mexican EMBIG indexes (for the majority of the
windows) showed negative abnormal returns, meaning the default probability diminished before the
crisis. Therefore, the anticipation of the crisis in Argentina and Mexico, using the EMBIG index as
country risk proxies did not exist but it did indeed become apparent for the Brazilian crisis.
Figure 3: The EMBIG Cumulative Abnormal Returns, CAR, during a period of 120 days before the event. For
Argentina’s case the adjusted EMBIG index is also shown.
EMBIG
2
1.5
Corralito
1
Deval.Arg
0.5
Corralito aj.
0
Deval.Arg.aj.
-0.5
Deval.Brasil
Deval.México
-1
-1.5
Day

---

---

121
International Research Journal of Finance and Economics - Issue 15 (2008)
Table 4 shows the results when applying the methodology to each one of the events in the
study, using the ADR portfolio from Argentina and Brazil1. In the case of Brazil, the only statistically
significant window was the day of the devaluation (t=0), but the non-parametric test is not statistically
significant. For the Argentinean ADRs, we did not find meaningful results, which confirm the ADR
market did not anticipate the crisis. When we studied the Argentinean devaluation, we did not obtain a
significant level with the nonparametric test, so we cannot accept abnormal returns in this critical
period.
Table 4: The ADR portfolio Cumulative Abnormal Returns, CARs, in parenthesis appears its corresponding
t-test. The CARs are calculated using the Dimson (1979) Adjusted Market Model where the market
portfolio was the S&P500 return.
Window

(0,0)
(-5,0)
(-10,0)
(-20,-0)
(-30,0)
(-60,0)
(-80,0)
Brazil
13,63%
-2,55%
1,80%
-3,14%
-1,37%
3,04%
-0,50%
15/01/1999
(3,93)
(-0,23)
(0,16)
(-0,906)
(-0,39)
(0,88)
(-0,15)
Argentina
1,17%
-1,04%
-0,01%
-2,45%
1,31%
1,65%
0,30%
03/12/2001
(0,56)
(-0,16)
(-0,01)
(-1,180)
(0,63)
(0,79)
(0,15)
Argentina
-2,24%
-5,09%
7,14%
-3,25%
0,78%
3,36%
-0,84%
07/01/2001
(-1,13)
(-0,81)
(1,14)
(-1,635)
(0,39)
(1,69)
(-0,42)
IV. Conclusions
The crises in Argentina, Brazil and Mexico have been widely studied, indicating that their causes were,
to a great extent, unanticipated. We analyzed whether the stocks indices, country risk proxies, and
ADRs anticipated the critical financial events lived by those economies during the 90´s. Our results are
mixed; as not one of these three variables was single-handedly able to predict the crises in all three
economies.
The anticipation ability of the country risk proxies is dissimilar for the EMBIGs and the Brady
Bonds. These results allow us to indicate that the anticipation of a crisis by means of country risk
analysis is, at best, difficult. This information though, continues to be a well relied upon and used as a
relevant indicator for practitioners; in spite of its low ability to anticipate the traumatic event lived in
Argentina, Brazil and Mexico during the 90´s. In and of itself, we suggest that this points to the need
for additional research in this area, so as to reconcile its wide use and the lack of predictive power of
these variables as seen in this study.
In the case of the stock market index as a tool to anticipate a crisis, the results were as expected
for Brazil and Mexico, but not in the case of Argentina whose CARs was positive, but not statistically
significant. Once again, we suggest that it might be useful to further pursue this question with
additional investigations, trying to reconcile these results, and/or explore different circumstances or
intervening variables that might bring about different findings.
One of the most unexpected results was the evidence from the ADR market; where the
conclusion is that there were no negative CARs. This is quite surprising, because these securities are
traded in a highly sophisticated and well-informed market, with enormous financial resources to
supervise the investments. This means that the devaluation announcements where not anticipated by
the ADR market, neither in the case of Argentina nor Brazil. Again, in line with the opinions above,
the results would suggest the merit of further analyzing how is it that such sophisticated instruments
fail to capture crisis information, or maybe, if they are indeed completely inappropriate for such task;
or beter yet: how, why, and where is that the case.
One might therefore conclude, given our findings and the extant body of research, that the key
to anticipate or better understand the economic crises addressed, would require a combination of
macroeconomic and international economy factors, together with the financial information contained
1
The ADRS of México were not included because of a problem of sample size.

---

International Research Journal of Finance and Economics - Issue 15 (2008)
122
by the local stock market, the ADRs market and the country risk proxies. Yet, the mixed results of this
investigation, also suggests that there is indeed a considerable number of issues that warrant
researching the roles of stock market indices, country risk measures and ADRs in predicting crises.
Among others, additional combinations of these indicators, alternative models, the role – if any – of
intervening variables -- and/or extensions of these financial measures and their the fundamental
predictive abilities appear to be warranted.
V. Ackonowledgement
The authors wish to thank Natalia Burgos, David Diaz, Marcelo Gutierrez and Paul Naranjo for their
assistance. We also thank the participants at the Balas conference in Madrid, Spain for their comments
and suggestions.
References
[1]
ADES, A. and M. BUSCAGLIA (1999), “Argentina: In Transition”, Goldman Sachs Global
Economies Paper, Nº 22
[2]
AGÉNOR, R., M. MILLER, D. VINES and A. WEBER (1999), “The Asian Financial Crises:
Causes, Contagion and Consequences”, Cambridge University Press, Cambridge.
[3]
BAMBACI, J., SARONT T. and TOMMASI, M. (2002), “The Political Economy of Economic
Reforms in Argentina”, Journal of Policy Reform, vol. 5 Nº 2, PP. 75-88.
[4]
BARCINSKI, A. and M. GARCÍA (1998), “Capital Flows to Brazil in the Nineties:
Macroeconomic Aspects and the Effectiveness of Capital Controls”, Department of Economics,
PUC-RIO (1998).
[5]
BUSCAGLIA, M.A. (2003), “The Political-Economy of Argentina’s Debacle”, William
Davidson Institute Working Paper, Publication Nº 594, July.
[6]
BUSTELO, P. (1998), “The East Asian Financial Crises: An Analytical Survey”, ICEI Working
Papers, Nº 10, October, Complutense University of Madrid, Spain.
[7]
BUSTELO, P., C. GARCÍA and I. OLIVIÉ (1999), “Global and Domestic Factors of Financial
Crises in Emerging Economies: Lessons from the East Asian Episodes (1997-1999)”, ICEI
Working Papers, Nº 16, November, Complutense University of Madrid, Spain.
[8]
BUSTELO, P. (2002), “Los Orígenes de la Crisis Financiera en Argentina: Una Comparación
con las Crisis Asiáticas”, Boletín Económico de ICE, Nº 2715, 7-13, January, pp. 9-14.
[9]
BUSTELO, P. (2004), “Capital Flows ad Financial Crises: a Comparative Analysis of East Asia
(1997-98) and Argentina (2001-02)”, Working Paper, Nº 2004-017, October, Faculty of
Economics, Complutense Univrsity of Madrid, Spain.
[10]
CALVO, G. A. (1996), “Varieties of Capital-Market Crises”, University of Maryland, May,
Typescript.
[11]
CARNEIRO, D. y M. GARCÍA (1994), “Flujos de Capital y Control Monetario Bajo
Sustitución Doméstica de Dinero: La reciente Experiencia Brasileña”, in R. Steiner, ed.,
Afluencia de Capitales y Estabilización en América Latina, Fedesarrollo, Bogotá, Colombia.
[12]
CHANG, R. and A. VELASCO (1998), “Financial Crises in Emerging Markets: A Canonical
Model”, Federal Reserve Bank of Atlanta Working Papers, Nº 98-10, July.
[13]
CHANG, R. and A. VELASCO (2001), “The 1997-98 Liquidity Crisis: Asia versus Latin
America”, Working Paper Nº 120, November, Chile Central Bank.
[14]
CHOWDURY, A. R. (1999), “The Asian Currency Crises: Origins, Lessons and Future
Outlook”, Research for Action, Nº 47, UNU-WIDER, Helsinki.
[15]
CORBETT, J. and D. VINES (1999), “The East Asian Financial Crises: Lessons from
Vulnerability, Crises and Collapses”, APSEM Working Paper, Nº 99-3, Australian National
University, Canberra.

---