THREE ESSAYS IN INTERNATIONAL FINANCE

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate School of

The Ohio State University

By

Rodolfo Martell, M.A.

The Ohio State University

2005

Dissertation Committee: Approved by

Professor René M. Stulz, Adviser

Professor G. Andrew Karolyi _________________

Adviser
Professor Bernadette A. Minton Graduate Program in Business Administration
 ABSTRACT

Recent research in international finance focuses on the extent to which markets

are integrated across countries, how shocks propagate from one country to another and

how firms in foreign countries react to country level shocks. This dissertation provides

empirical evidence on the degree of integration in international bond markets, on the

propagation of extreme shocks between cross-listed shares and domestic markets and on

the dispersion in capital market reactions across firms to sovereign rating changes.

In the first dissertation essay, I study the determinants of credit spread changes of

individual U.S. dollar denominated bonds – domestic and foreign sovereign – using

fundamentals specified by structural models. Credit spreads are important determinants

of the cost of debt for all issuers and are fully determined by credit risk in structural

models. I construct a new dataset of domestic corporate and sovereign U.S. dollar bonds,

which I use to find that changes in spreads not explained by fundamentals have two large

common components that are distinct for each type of debt I study. Using a vector

autoregressive (VAR) model, I find that domestic spreads are related to the lagged first

component of sovereign spreads. Consequently, even though there is no

contemporaneous common component in bond spreads, there seems to be a common

component when focusing on the dynamics of these spreads. Traditional macro liquidity

ii
 variables are related to the common components found in domestic and sovereign

spread changes. My findings suggest possible explanations for the common component

documented by previous research in domestic debt spreads. My research shows that, after

taking into account the dynamics of the common components in credit spreads across

debt types, the cost of debt for firms and countries depends to some extent on shocks that

affect all types of debt.

The second dissertation essay studies the extreme linkages between Latin

American equities and the US stock market using tools from Extreme Value Theory

(EVT). Bivariate extreme value measures are applied on six different country pairs

between the U.S. S&P500 Index and each of the following countries: Argentina, Brazil,

Chile, Colombia, Mexico and Venezuela. I find evidence of: a) asymmetric behavior in

the left and right tails of the joint marginal extreme distributions, and b) differences in

extreme correlations for different instruments (investing in ADRs vs. investing directly in

the local stock markets) when no difference was to be expected. There is also evidence of

a structural change in the correlations for the Mexican case before and after the 1995

Mexican crisis.

The third dissertation essay studies the effect of sovereign credit rating changes

issued by Standard and Poor’s and Moody’s on the cross section of domestically traded

stocks. I first establish, consistent with earlier literature that analyzed similar phenomena

in the U.S. (e.g. Holthausen and Leftwich, 1986; Goh and Ederington, 1993), that local

stock markets react only to news of sovereign credit rating downgrades. Cumulative

abnormal returns of stock indices also show that investors react only to rating

announcements made by Standard & Poor’s and not to those by Moody’s. I then study the

iii
cross sectional variation of the abnormal returns of individual firms associated with

sovereign credit rating changes. I find that larger firms experience larger stock price

drops after a sovereign credit downgrade. Also, firms located in more developed

emerging countries experience smaller stock price reductions following sovereign credit

downgrades. Finally, I document that firms that had access to international capital

markets experience larger abnormal returns than firms that do not have access to

international financial markets.

iv
Dedicated to my family

v
 ACKNOWLEDGMENTS

I wish to thank my adviser, René Stulz, for intellectual support, encouragement,

and enthusiasm which made this dissertation possible, and for his patience in correcting

both my stylistic and methodological errors.

I thank Andrew Karolyi for stimulating discussions, guidance, and

encouragement, not only with this dissertation but throughout my graduate studies.

I am grateful to Bernadette Minton for discussing with me various aspects of this

thesis, and for her insightful feedback.

I also wish to thank Mike Cooper, Craig Doidge, Jean Helwege, Francis

Longstaff, and seminar participants at Drexel University, Fordham University, Ohio State

University, Purdue University, Queen’s University, and University of Virginia for helpful

comments and suggestions.

vi
 VITA

July 9, 1972.................Born – Puebla, Puebla, Mexico

1996............................Bachelor of Arts in Economics, Udla-Puebla, Mexico

1996 – 1999 ...............Analyst,

Bancrecer
Petroleos Mexicanos

2000............................Master of Arts in Economics, Ohio State University

PUBLICATIONS

Research Publication

1. R. Martell and R. Stulz, “Equity Market Liberalizations as Country IPOs.”

American Economic Review, 93(2), 97, (2003)

FIELDS OF STUDY

Major Field: Business Administration

Concentration: Finance

vii
 TABLE OF CONTENTS

Abstract ............................................................................................................................... ii

Dedication ........................................................................................................................... v

Acknowledgments.............................................................................................................. vi

Vita.................................................................................................................................... vii

List of tables....................................................................................................................... xi

List of figures................................................................................................................... xiii

Chapter 1: Introduction ....................................................................................................... 1

Chapter 2: Understanding common factors in domestic and international bond spreads... 6

2.1. Introduction.............................................................................................................. 6
2.2. Debt spreads of sovereign bonds ........................................................................... 10
2.2.1. Sovereign debt literature. ................................................................................ 11
2.2.2. Implications of the literature and proxies used to test them. .......................... 14
2.2.2.1 Bond-specific variables............................................................................. 15
2.2.2.2. Country-specific variables ....................................................................... 15
2.2.2.3. U.S. interest rate term structure. .............................................................. 16
2.2.3. Data description .............................................................................................. 16
2.2.4. A model for sovereign spreads ....................................................................... 20
2.3. Debt spreads of domestic bonds ............................................................................ 22
2.3.1. Domestic debt literature.................................................................................. 23
2.3.2. Theoretical determinants of domestic debt spreads ........................................ 25
2.3.2.1. Bond specific variables ............................................................................ 25
2.3.2.2. Firm specific variables............................................................................. 25
viii
 2.3.2.3 U.S. interest rate term structure ............................................................ 26
2.3.3. Data description .............................................................................................. 26
2.3.4. A model for domestic debt spreads................................................................. 28
2.4. Analyzing the common factor................................................................................ 29
2.4.1. Establishing the existence of common factors................................................ 29
2.4.2. Explanatory power of the extracted components............................................ 33
2.5. Looking into the information content of the common factors ............................... 34
2.5.1. Lead-lag relations............................................................................................ 34
2.6. Conclusions and future work ................................................................................. 39

Chapter 3. Latin American and U.S. equities return linkages: An extreme value approach

........................................................................................................................................... 41

3.1 Introduction............................................................................................................. 41
3.2. Literature review.................................................................................................... 44
3.2.1. The univariate case ......................................................................................... 47
3.2.2. The bivariate case ........................................................................................... 50
3.3. Data ........................................................................................................................ 51
3.4 A small test for the Mexican pairs .......................................................................... 55
3.5. Concluding remarks ............................................................................................... 55

Chapter 4. The effect of sovereign credit rating changes on emerging stock markets ..... 58

4.1. Introduction............................................................................................................ 58
4.2. Literature review.................................................................................................... 65
4.3. The effect of sovereign rating changes on stock market indices ........................... 71
4.3.1. Data ................................................................................................................. 72
4.3.2. Methodology ................................................................................................... 74
4.3.3. Discussion of index level results..................................................................... 75
4.4. Impact of sovereign rating changes at the firm level............................................. 79
4.5. Conclusions............................................................................................................ 86

Chapter 5: Conclusions ..................................................................................................... 88

Bibliography ..................................................................................................................... 91

ix
Appendix A. A comparison of sovereign bond coverage on Datastream and the NAIC . 99

Appendix B. Tables ........................................................................................................ 103

Appendix C. Figures ....................................................................................................... 134

x
 LIST OF TABLES

Table 1. Expected signs on explanatory variables for sovereign sample ....................... 104

Table 2. Summary statistics for sovereign sample.......................................................... 105

Table 3. Sovereign spreads fixed effect regressions....................................................... 106

Table 4. Expected signs on explanatory variables for domestic sample......................... 107

Table 5. Summary statistics for domestic sample........................................................... 108

Table 6. Domestic spreads fixed effect regressions........................................................ 109

Table 7. Correlation structure of residuals...................................................................... 110

Table 8. Principal component analysis of residuals........................................................ 112

Table 9. Sovereign and domestic regressions including the common factors ................ 114

Table 10. Vector autoregression model with exogenous variables................................. 115

Table 11. Summary statistics .......................................................................................... 116

Table 12. Extreme correlations using different number of tail exceedances.................. 117

Table 13. Sovereign rating changes by Standard & Poor's............................................. 118

Table 14. Sovereign rating changes by Moody's ............................................................ 119

Table 16. Stock index results using Moody's ratings...................................................... 121

Table 17. Stock index results using initial ratings .......................................................... 122

Table 18. First ratings for Argentina............................................................................... 123

Table 19. Stock market reaction to the first rating by either agency .............................. 124

xi
Table 20. Cumulative Abnormal Returns (CAR) for stocks with international financing

......................................................................................................................................... 125

Table 21. Cumulative Abnormal Returns (CAR) for all stocks following a sovereign

rating downgrade ............................................................................................................ 126

Table 22. Cumulative Abnormal Returns (CAR) for all stocks following a sovereign

rating upgrade ................................................................................................................. 129

Table 23. Countries included in this comparison............................................................ 132

Table 24. Coverage for sovereign bonds on Datastream and Warga databases. ............ 133

xii
 LIST OF FIGURES

Figure 1. First common component ................................................................................ 135

Figure 2. Second common component............................................................................ 135

Figure 3. Q-Q Plots for the left tail and the right tail of the dollar return of the Mexican

equity index..................................................................................................................... 136

Figure 4. Q-Q Plots for the left tail and the right tail of the dollar return of the Mexican

ADR equally weighted portfolio..................................................................................... 137

Figure 5. Q-Q Plots for the left tail and the right tail of the dollar return of the S&P 500

equity index..................................................................................................................... 138

Figure 6. Excess mean graphs for the left tail and the right tail of the dollar return of the

Mexican equity index...................................................................................................... 139

Figure 7. Excess mean graphs for the left tail and the right tail of the dollar return of the

Mexican ADR equally weighted portfolio...................................................................... 140

Figure 8. Excess mean graphs for the left tail and the right tail of the dollar return of the

S&P 500 equity index ..................................................................................................... 141

Figure 9. Correlation between S&P and the Mexican stock market index and correlation

between S&P and Mexican ADRs.................................................................................. 142

Figure 10. Correlation between S&P and the Chilean stock market index and correlation

between S&P and Chilean ADRs ................................................................................... 142

xiii
Figure 11. Correlation between S&P and the Venezuelan stock market index and

correlation between S&P and Venezuelan ADRs........................................................... 143

Figure 12. Correlation between S&P and the Colombian stock market index and

correlation between S&P and Colombian ADRs............................................................ 143

Figure 13. Correlation between S&P and the Brazilian stock market index and correlation

between S&P and Brazilian ADRs ................................................................................. 144

Figure 14. Correlation between S&P and the Argentinean stock market index and

correlation between S&P and Argentinean ADRs.......................................................... 144

Figure 15. Correlation between S&P and the Mexican stock market index and correlation

between S&P and Mexican ADRs before the 1995 Mexican crisis ............................... 145

Figure 16. Correlation between S&P and the Mexican stock market index and correlation

between S&P and Mexican ADRs after the 1995 Mexican crisis .................................. 145

Figure 17. Sovereign Downgrades (S&P) ...................................................................... 146

Figure 18. Sovereign Upgrades (S&P) ........................................................................... 146

Figure 21. Sovereign Downgrades (Moody’s)................................................................ 147

Figure 20. Sovereign Upgrades (Moody’s) .................................................................... 147

xiv
 CHAPTER 1

INTRODUCTION

The last twenty years have witnessed large reductions in regulations and barriers

that prevented financial integration across countries. As markets slowly became more

integrated, brand new fields for financial research opened up. Not only could we study if

foreign markets behaved in a similar way to U.S. markets, but we also could study issues

surrounding the integration of U.S. and international financial markets. The first

dissertation essay investigates whether common factors that explain credit debt spread

changes for domestic and sovereign debt after taking into account fundamentals are

related, and then proceeds to analyze the determinants of these common factors. This

dissertation essay looks at two groups of assets that had previously been studied only

separately. It focuses on credit spread changes of U.S domestic bonds and sovereign

bonds, making it the first paper to bring together these two groups of individual bonds to

study their joint dynamics.

Previous research in spread changes of U.S. domestic bonds identified a common

component unrelated to credit risk in the time-series and cross-section of the unexplained

portion of the spreads (Collin-Dufresne, Goldstein and Martin, 2001; Huang and Huang,

2003). If the U.S. and overseas market for dollar-denominated credit-risky bonds is

1
integrated, the information present in the unexplained portion of U.S. dollar sovereign

debt spread changes should be related to unexplained portion of U.S, domestic bonds

spread changes. This especially should be the case if that common component can be

explained by liquidity shocks, since such shocks are pervasive across markets (Chen,

Lesmond, and Wei, 2002; Chordia, Sarkar, and Subrahmanyam, 2003; Kamara, 1994).

Existing research investigates separately the existence of common components in

changes in credit spreads for domestic credit-risky debt (Collin-Dufresne, Goldstein and

Martin, 2001) and dollar-denominated sovereign debt (Scherer and Avellaneda, 2000;

Westphalen, 2003). The contribution of this dissertation essay is to study the relation

between the common components identified in domestic debt and the common

components found in sovereign credit spreads.

To conduct this analysis, a new dataset comprised of all domestic industrial and

U.S. dollar-denominated sovereign debt is constructed. This dataset contains data for 233

non-callable, non-puttable bonds issued by 37 emerging countries and 3097 domestic

corporate bonds issued by 649 different companies that traded between January 1990 and

January 2003. Results obtained help to discriminate between competing explanations for

the common component previously documented for domestic debt, and also might

suggest new explanations.

I find strong evidence of the existence of two common factors unrelated to credit

risk in debt spread changes of U.S. denominated sovereign debt and in the debt spread

changes of domestic bonds. While principal component analysis shows no evidence of

contemporaneous correlation between the two domestic and the two sovereign factors, a

vector autoregressive (VAR) model shows that domestic spread changes are related to the

2
lagged sovereign spread’s first principal component. Finally, I find that all four common

factors are related to the flows of money going into equity and bond funds, and the

second common component of each group is related to the net borrowed reserves from

the Federal Reserve, a macroeconomic measure of liquidity.

The second dissertation essay analyses the extent of the financial and economic

integration between Latin American countries and the United States by focusing on the

behavior of linkages between financial assets using a statistical technique known as

Extreme Value Theory (EVT). EVT is the study of outliers or extremal events. Since

large movements in returns are usually characteristic of financial crisis and since these

large movements can be considered outliers, the use of EVT seems to be warranted. This

approach has several advantages. First among these are the well-known results on

asymptotic behavior of the distribution of very high quantiles. Second, no assumptions

are needed about the true underlying distribution that generated data in the first place.

Since financial contagion usually occurs during periods of very high distress, it seems to

be best analyzed using techniques that focus on the tails of a distribution function. (Bae,

Karolyi and Stulz, 2003)

The financial assets I analyze are American Depositary Receipts (ADRs) and their

domestic counterparts in Latin America. Latin American firms can cross-list their shares

in the U.S. via ADR programs, and at the end of 2001 there were 1,322 non-U.S. firms

with sponsored programs, including 623 trading on American stock exchanges with a

total trading volume of $752 billion.

This chapter documents evidence of the asymmetric transmission of shocks from

U.S. stock markets into domestic markets. It builds on the work of Longin (1996) and

3
Longin and Solnik (2001) applying EVT in finance by examining the linkage between

financial assets available to U.S. investors looking for international exposure before and

after main events such as the 1995 Mexican crisis. It also adds to the growing literature

on financial contagion by employing a statistical technique more “appropriate” than

current approaches based on elliptic distributions for the often temporary, but large,

movements in prices.

The third dissertation essay studies the effect of sovereign credit rating changes

on the cross-section of locally-traded firms. A sovereign credit rating reflects the rating

agency’s opinion on the ability and willingness of sovereign governments to service their

outstanding financial obligations and it reflects macroeconomic factors related to political

and financial stability. Sovereign credit ratings have large effects that spread to firms

located within their borders, and changes to these ratings constitute country-wide shocks

that can have sizable effects on the terms under which firms obtain financing and the

overall cost of capital.

This chapter contributes to the existing literature by extending our understanding

of how much information sovereign rating changes convey to individual stocks within

domestic markets. Specifically, I investigate if and why a country rating matters for firms

within a country. I show that sovereign rating changes affect the terms on which a

domestic firm can get credit, creating an exogenous change in the cost of capital. I divide

the results into two parts: I first present the effect of rating changes at the aggregate level

using national stock indices, and then proceed to study the effect of those sovereign

rating changes on the individual firms located within those countries. Index level results

are consistent with the extant literature on the effect of credit-rating changes on U.S.

4
firms. I do find evidence of a significant negative stock price reaction to sovereign rating

downgrades while I find no evidence of a stock price reaction to sovereign rating

upgrades. Further, I document that local stock markets react only to news of sovereign

rating downgrades issued by Standard & Poor’s.

To conduct this analysis I collected all sovereign rating changes issued by

Standard and Poor’s (S&P) and Moody’s on 29 emerging countries from 1986 until 2003.

I study the stock price reaction to 136 downgrades (81 from S&P and 55 from Moody’s)

and 100 upgrades (57 and 43 from S&P and Moody’s respectively). I also collect

information on 1281 individual firms located in 29 emerging countries. After computing

abnormal returns for each firm, cross-sectional regressions of those abnormal returns are

run on firm-specific characteristics and country-specific variables. I document how the

size and wealth of the country where a firm is domiciled are related to the extent to which

that a firm will be affected by a sovereign-rating change. More importantly, I find that

previous access to international capital markets is an important determinant of the extent

to which a firm is affected by a sovereign credit rating change.

5
 CHAPTER 2

UNDERSTANDING COMMON FACTORS IN DOMESTIC AND

INTERNATIONAL BOND SPREADS

2.1. Introduction.

In this chapter I analyze the determinants of credit spread changes of individual

U.S domestic and sovereign bonds. Previous research has focused on one type of bonds at

a time, making this paper the first one to bring together the credit spreads on these two

types of debt to study their joint dynamics. If the market for dollar-denominated credit-

risky bonds is integrated, we can expect credit and non-credit related shocks to affect all

bonds, i.e. the information present in the time series cross-section of the unexplained

portion of U.S. dollar sovereign debt spread changes should be related to the common

component unrelated to credit risk identified by previous research in spread changes of

U.S. domestic bonds (Collin-Dufresne, Goldstein and Martin, 2001; Huang and Huang,

2003). This should especially be the case if that common component can be explained by

liquidity shocks, since such shocks are pervasive across markets (Chen, Lesmond, and

Wei, 2002; Chordia, Sarkar, and Subrahmanyam, 2003; Kamara, 1994). In this chapter, I

investigate whether common factors that explain credit spread changes for domestic and

sovereign debt after taking into account fundamentals are related and analyze the

determinants of these common factors.

6
 Existing research investigates separately the existence of common components in

changes in credit spreads for domestic credit-risky debt and dollar-denominated

sovereign debt. Scherer and Avellaneda (2000) identify the existence of two common

factors for sovereign debt spread changes. Westphalen (2003) finds evidence of a

common factor for sovereign debt spread changes of bonds denominated in several

currencies after controlling for country risk proxies. Research on changes in domestic

bond credit spreads by Collin-Dufresne, Goldstein and Martin (2001) finds one common

component after controlling for fundamentals. The relation between these common

components has not been examined in the literature.

I extend the research on common components present in bond spreads by

examining whether the information in the dynamics of U.S. dollar denominated sovereign

debt spreads is associated with the common component found in U.S. corporate bond

spreads. Specifically, I estimate different models of spread changes for each type of

bonds – domestic and sovereign – because these two groups vary in their source of credit

risk. Using principal component analysis for each debt type, I extract common factors

from the unexplained portion of credit spread changes from these models. I investigate

whether the common factors in U.S. dollar denominated sovereign debt are related to the

common factors present in U.S. corporate debt spread changes using both regressions

explaining contemporaneous changes in spreads and a dynamic model of changes in

spreads. Finally, I attempt to provide an economic interpretation for the relations I

uncover.

To conduct this analysis, I construct a new dataset that is comprised of all

domestic industrial and U.S. dollar-denominated sovereign debt. This dataset contains

7
data for 233 non-callable, non-puttable bonds issued by 37 emerging countries and 3097

domestic corporate bonds issued by 649 different companies that traded between January

1990 and January 2003. This dataset is different from the ones used by earlier studies in

at least three ways. First, extant bond studies that use Datastream bond data do not

include ‘dead’ issues, i.e., bonds that have matured or were retired, while I include them

to avoid a survivorship bias. Second, the Fixed Income Database used in some other

studies has a limited coverage of high-yield issues since it mainly covers investment-

grade bonds (Huang and Kong, 2003). I do not have this problem because my dataset

contains data for the complete universe of bonds covered by Datastream.1 Finally, this

dataset covers a longer time period than any previous study.

My results help to discriminate between competing explanations for the common

component previously documented for domestic debt, and also suggest new explanations.

I find strong evidence of the existence of two common factors unrelated to credit risk in

debt spread changes of U.S. denominated sovereign debt and in the debt spread changes

of domestic bonds. While principal component analysis shows no evidence of

contemporaneous correlation between the two domestic and the two sovereign factors, a

vector autoregressive (VAR) model shows that domestic spread changes are related to the

lagged sovereign spread first common component. Finally, I find that all four common

factors are related to the flows of money going into equity and bond funds, as measured

by the Investment Company Institute (ICI), while only the second common component of

1
Informal conversations with Datastream’s customer service revealed that several large banks, including
Lehman Brothers, were among their providers for bond data. Since Lehman Brothers was the provider for
the FISD, we feel confident Datastream’s data includes what is covered in the FISD and has broader
coverage of high-yield bonds because of the additional data providers. A comparison between FISD and
Datastream sovereign bond data can be found in Annex A.
8
each group is related to the net borrowed reserves form the Federal Reserve, a

macroeconomic measure of liquidity.

This chapter is the first one to bring together these two types of credit-risky

dollar-denominated debt to study the joint dynamics of the common factors in their credit

spreads. The results I obtain improve our understanding of the determinants of the cost of

debt for foreign countries and for domestic firms. For example, my results suggest that

the cost of debt for foreign countries and domestic firms is not only a function of their

own creditworthiness but also depends on shocks that affect the price of all debt.

Further, these results help us understand better the extent to which the sovereign

and domestic corporate bond markets are integrated. In a fully integrated dollar debt

market, we would expect the relation between domestic corporate credit spreads and

sovereign credit spreads to be contemporaneous. Further research should investigate

whether the lack of a contemporaneous relation is due to differences in liquidity and

infrequent trading or if this reflects a market inefficiency.

Finally, the lack of a relation between the common components of domestic

corporate credit spread changes and sovereign credit spread changes suggests that the

cost of debt for emerging markets depends mostly on country and emerging-market

specific considerations. This is surprising in light of a considerable literature that

emphasizes the impact of developed country developments for capital flows into

emerging markets (Calvo, Leiderman, and Reinhart, 1993; Chuhan, Claessens, and

Mamingi, 1998). Further investigation of the robustness of my results might shed greater

insight into this issue.

9
 This chapter proceeds as follows. Section II describes the literature, sample,

variables and methodology used to model credit spread changes for sovereign bonds.

Section III does the same for credit spread changes for domestic corporate bonds. I

investigate, using a variety of techniques, the existence and nature of the factors affecting

debt spread changes in section IV. Section V analyzes the dynamics of the common

factors and investigates whether liquidity and/or demand related variables are related to

them. Section VI concludes.

2.2 Debt spreads of sovereign bonds

In order to examine whether a common factor is associated with the variation in

U.S. domestic corporate and U.S. dollar denominated sovereign spreads, the unexplained

variation in each spread (i.e. residuals) must be calculated. My choice of variables to

compute the credit risk portion of debt spread changes is based on the determinants of

bond spread changes specified by structural models. For sovereign bond spreads, I expect

bond-specific characteristics to be associated with bond spreads. Additionally, I expect

bond spreads to be related to macro or country-specific factors as well as systematic

factors. In this section, I review the relevant literature on U.S. dollar denominated

sovereign bond spreads (section 2.1), and then discuss the testable implications of the

extant literature and describe the proxies that are used to test the hypotheses derived from

it (section 2.2). I describe the sovereign bond sample next (section 2.3), present a model

to estimate debt spreads, discuss the results, and explain the computation of residuals

(section 2.4).

10
2.2.1 Sovereign debt literature.

The international debt market changed dramatically in the past 25 years. In the

1980s bank loans were the principal instrument of this market. By the end of that decade,

reckless lending and borrowing caused outstanding debt balances to skyrocket to

unsustainable levels. The crushing pressure of debt payments forced several emerging

market countries to the verge of default. To avoid the ripple effects of such a default on

the world’s financial system –which was still recovering from the 1987 stock market

crash-- the U.S. government helped put in place a plan that would allow these countries

to orderly restructure their debt schedule. The Brady plan, formulated in 1989 by then

Secretary of the Treasury Nicholas Brady in association with the World Bank and IMF,

called for the issuance of sovereign bonds to replace the loans of commercial banks.2

Brady bonds opened a vast and untapped market for emerging market countries hungry

for U.S. dollars to help finance their growth, commercial deficits or simply to cover

current expenses. Bank loans, while still an important component in sovereign debt

balances, gave way to sovereign bonds as the principal financing instrument for emerging

countries in the 1990s. Bonds were clearly preferred for several reasons, for instance the

dispersion of creditors and the existence of a market where these bonds could be actively

traded, which provided investors with a transparent benchmark measure of country risk.

2
These bonds were coupon bearing (fixed, floating or hybrid), long maturity (ten to thirty years) issued in
registered or bearer form, whose principal and part of the interest were guaranteed by collateral of U.S.
Treasury bonds and other high grade securities. Some of them included special recovery rights (warrants)
that could be detached and traded separately. This last characteristic made the computations of yields for
these bonds especially tricky.
11
The sovereign spread, or credit spread, computed now from bond yields, continued to be

such a benchmark measure of country risk.3

Starting in the 1980s, the determinants of sovereign debt spreads has been studied

by Eaton and Gersovitz (1981), where governments trade off the cost of paying debt

versus reputation costs or exclusion from capital markets, and Bulow and Rogoff (1989),

who provide rational explanations for international lending and model the costs of debt

repudiation as direct sanctions. Edwards (1984) analyzed the macroeconomic

determinants of the debt spread measured as the difference between the interest rate

charged to a particular country and LIBOR (London InterBank Offered Rate).

Hernández-Trillo (1995) uses a measure openness, unexpected shocks to GDP,

international reserves and the risk free rate to explain the probability of default in

sovereign loans. The international episodes of financial contagion experienced in the

second half of the 1990s attracted even more attention to this area, as researchers started

to devote more time to the study of periods of increased co-movements among

international financial markets. For instance, Cantor and Packer (1996) and Eichengreen

and Moody (1998) study the determinants of bond spreads at the issue level, finding that

agency ratings include most of the information existing in macroeconomic variables.

More recently, Scherer and Avellaneda (2000), Joutz and Maxwell (2002) and Cifarelli

and Paladino (2002) study selected series from several emerging markets using principal

component analysis and vector-autoregressions.

3
The credit spread is often referred to as yield spread, debt spread or simply spread. These terms are used
interchangeably in this paper.

12
 It was previously mentioned in this chapter I take a structural approach to the

modeling of debt spreads. It is important to mention, though, that sovereign debt is

different from corporate debt. One of the most important characteristics of any debt

contract is the guarantee provided by the legal framework to creditors that allows them,

in the case of default, to take possession of collateral and/or to liquidate the defaulting

debtor’s assets. There is no enforceable bankruptcy code for sovereign bonds, making it

effectively impossible for a creditor to successfully pursue a claim on a defaulting

country’s assets. Acknowledging the endogenous default decision that countries face in

this framework, Gibson and Sundaresan (1999) present a model in which creditors can

impose trade sanctions and capture some fraction of the defaulting country’s exports, and

Westphalen (2002) extends their model to include rescheduling in the form of a bond

exchange. Finally, Westphalen (2003) applies a methodology used to study corporate

credit spread changes (Collin-Dusfrene et.al. 2001) to a sample of sovereign bonds issued

in different foreign currencies.4

So far, research on sovereign debt spreads has focused more on how spreads are

determined at issue than on the study of the dynamics of the cross-section. There are two

reasons for this. First, thin trading in many of these bonds produces relatively fewer

sovereign bond transactions data. As a result, some data vendors resort to provide matrix

prices (e.g. Bloomberg), which are not useful for research purposes.5 Second, in the early

1990s, when the market for sovereign debt was in its infancy, countries started by issuing
4
Another approach to the study of sovereign spreads has been implemented through the use of models
based on an exogenously specified intensity process, known as reduced-form models. Merrick (2000)
studies the implied recovery rations in Argentinean and Russian bonds. Pagès (2001) fits the joint Libor
structure and discount Brady bond prices to a reduced-form model using a two factor affine-yield model.
Duffie, Pedersen and Singleton (2003) conduct an analysis of Russian debt.
5
Actual quotes and/or transaction prices are available from different providers in the Bloomberg terminal
through an additional subscription service.
13
few bonds. As their credibility improved, reinforced by the implementation of structural

reforms in their economies, and investors got acquainted with this new supply of bonds,

sovereign issuers increased the number and amount of debt offerings. Therefore, it took

some years for this market to be sufficiently diverse and liquid enough to allow the

construction of a data panel suitable for research purposes. Today, the sovereign debt

market is more developed –we have more bonds with longer time series each one- and

there are better and more alternatives to obtain bond data – several information services

provide access now to observed pricing data, although some remain very expensive.

2.2.2 Implications of the literature and proxies used to test them.

Structural models of sovereign debt have identified macroeconomic variables that

affect sovereign debt spreads.6 Based in part on previous literature, I put together three

groups of variables that should capture most of the debt spread variation. The first group

contains bond-specific variables, i.e., variables that vary within bond issues, e.g. years to

maturity. The second group contains variables that vary from country to country but are

the same for all bonds from a given country (country-specific variables). The third group

contains variables that are the same for all bonds in the sovereign sample, and try to

capture changes in the U.S. interest rate term structure.

6
One problem with most empirical work exploring the relation between macroeconomic variables and debt
spreads is that they conduct static analysis, i.e., only study the cross-section of spreads at one point in time,
usually at issuance. For instance, GDP growth has been theoretically and empirically shown to have
significant explanatory power over issue level spreads. This is not useful in this context since most of the
data used in this paper is released monthly, quarterly or even annually in some countries.
14
2.2.2.1 Bond-specific variables.

The bond-specific variable used is years to maturity. By definition, a bond’s life

to maturity duration measures how long an investor has to wait before getting their

money back. Sovereign bonds pay (relatively) large coupons and therefore a large

proportion of the cash flows are paid throughout the life of these bonds, thus we have to

consider the possibility that years to maturity could be an overstated proxy of a bond’s

average life.

2.2.2.2 Country-specific variables.

These variables are chosen to capture a measure of a country’s distance-to-

default, i.e., a country’s ability (and/or willingness, depending on the model of reference)

to keep servicing its debt. Following Eaton and Gersovitz (1981), Bulow and Rogoff

(1989), Krugman (1985, 1989), Gibson and Sundaresan (1999) and Westphalen (2002), I

collect data on exports and total debt outstanding to construct a debt-to-exports ratio.

Borrowing from Krugman and Rotemberg’s (1991) speculative currency attacks model, I

use international reserves to construct a debt-to-reserves ratio as an alternate proxy of

distance-default. Westphalen (2003) uses a political risk measure which I also use here. I

also collect the Standard and Poor’s (S&P) ratings history for each country. I use the

monthly volatility of local stock returns as a proxy for volatility in a country’s wealth or

value. This measure is also used because it is a good proxy for local risk. One direct

testable implication of this is that spreads should increase with volatility. Finally, the

local stock market return in U.S. dollars is included.

15
2.2.2.3 U.S. interest rate term structure.

Since all the bonds in my dataset are denominated in U.S. dollars, I care about

factors that affect the U.S. yield curve term structure. From Litterman and Scheinkman

(1991) we know that the U.S. yield curve level and slope are important explanatory

factors of the term structure. Further, in this framework, if a country’s wealth follows a

stochastic process analogue to a firm’s value process, the risk neutral drift will be

positively related to the risk-free rate. An increase (decrease) in the risk-free rate should

increase (decrease) the country’s wealth over time, making default less (more) likely to

happen. Since an upward-sloping yield curve slope is, according to the expectations

hypothesis theory of the term structure,7 predicting higher interest rates in the near future,

I expect this slope to have some effect on spreads today. Also, a positively sloped interest

rate term structure is perceived as signaling increased economic activity in the near

future.

Table 1 presents the predicted correlation signs between the variables previously

mentioned and debt spreads.

2.2.3 Data description.

I collect monthly data on all U.S. dollar denominated bonds with Datastream

coverage. Datastream’s yields are calculated using average market maker prices provided

by the International Securities Market Association (ISMA). I am able to identify 5270

‘live’ and 3451 ‘dead’ bonds8 issued by foreigners that traded between January 1990 and

7
Bodie, Kane and Marcus (1999), pp. 446.
8
One important feature of Datastream’s coverage of bonds is that only ‘live’ issues (i.e., issues that are
currently trading) appear in their bond lists. Therefore, to make sure I had all available data, I conducted a
16
January 2003. I eliminate from the dataset all bonds that were callable and/or puttable at

borrower’s option, all that had an early redemption feature and/or were extendible at the

bond holder’s option, and all that were not issued by a sovereign entity.9 This leaves my

dataset with 181 live and 52 dead bonds. Also, I eliminate all observations with less than

one year to maturity because, as these bonds approach their maturity date, they are less

traded, which in turn dries up their liquidity and distorts prices and yields.10 After all

these adjustments, I come up with a sample that contains 9,275 monthly observations

from 233 bonds issued by 37 different countries, which did trade between January 1990

and January 2003.

For each bond, I collect the monthly redemption yield (datatype 4 in Datastream).

I also collect the monthly U.S. Treasury yield curve. Then, I compute debt spreads as the

difference between the redemption yield of the sovereign bond and the value of a linear

interpolation of the U.S. Treasury yield curve to obtain the yield of a U.S. instrument

with identical maturity as the bond being analyzed.11 I collect years to maturity time

series for each bond. As proxies for the U.S. Treasury yield curve’s level and slope, I

collect monthly annualized yields for the on-the-run two and ten year Treasury notes.12

country by country search of U.S. dollar denominated bonds in the ‘Dead bonds’ (not trading anymore
because they were retired or they matured) section of the Datastream Extranet web site.
9
I am not using Brady bonds in my analysis because their characteristics are inherently different from
regular sovereign bonds. The existence of collateral as well as the existence of value recovery rights
attached to Brady bonds makes them a class on their own. Further, the tendency is for sovereigns to retire
par and discount Brady bonds, so that movements in Brady bond’s prices might be reflecting low volume
and thin trading problems and not changes associated with the underlying value of the issuer and the overall
liquidity of the market For instance, Mexico’s Ministry of Finance and Public Credit announced on April 7,
2003 that it was calling US$3,839 million of its dollar-denominated Series A and B Brady Par Bonds,
which were the last outstanding series of Mexican Brady Bonds denominated in dollars.
10
Sarig and Warga (1989). This effect is even more pervasive when considering that liquidity was not great
in the first place.
11
I also collected the monthly U.S. Treasury yield curve using CMT (constant maturity treasuries) to
calculate spreads and our results are insensitive to the choice of U.S. benchmark curve.
12
The use of CMT yields for those maturities did not affect our results at all.
17
Monthly exports data expressed in nominal U.S. dollars come from the IMF’s

International Financial Statistics. Debt outstanding and foreign reserves data are obtained

from the joint BIS-IMF-OECD-World Bank statistics on external debt. Quarterly data on

the total amount outstanding of bank loans, of debt securities issued abroad and of Brady

bonds is obtained from this source, as well as monthly data on the amount of international

reserve assets, excluding gold.13 One shortcoming of this database is that not all series are

available on a quarterly basis and there are some gaps in the data, especially in the earlier

1990s.

The Economist Intelligence Unit (EIR) started publishing in March 1997 a

measure of country risk for emerging markets. It measures political, economic policy,

economic structure, currency, sovereign debt and banking sector risks. This index can be

used as a guide for the general risk of a specific country. It provides help in assessing the

risk of investing in the financial markets of those economies as well as the risks involved

in direct investment. The values are derived from measuring the risk associated with four

aspects of the country –political risk, economic risk, economic structure risk and liquidity

risk.14

To get a measure of monthly local wealth volatility, I use an equity volatility

measure as proxy. Ideally, I wanted to use MSCI country indices, since they are

calculated for each country using the same methodology. However, MSCI country

indices were not available daily going back to the early 1990s for many of the countries

13
All figures are expressed in current U.S. dollars.
14
The overall risk rating is measured on a scale from 1 to 100 where 1 denotes the least risk and 100 the
most risk possible. For example, in December 2002, the value of the index was 78 for Argentina, 63 for
Brazil and 48 for Mexico.

18
included in this paper. Therefore, I used Datastream local equity indices. For more than

half of the countries in the sample (twenty one), I collect daily data for the local

Datastream equity index. For eight additional countries, I collect daily data from their

own local equity indices. For the remaining countries Datastream’s world total return

index was used. To correct for differences in the scales of the indices the coefficient of

variation (sample standard deviation over sample mean) was computed.

I also collect the available history of Standard & Poor’s (S&P) country ratings

from Bloomberg, and follow Eom, Helwege and Huang (2003) for translating S&P

ratings into numerical values, where a rating of AAA has a value of 1, AA+ a value of 2

and so on.

Table 2 has summary statistics for the sovereign sample. Observations are

grouped in five different categories according to their S&P rating. It is evident from panel

A that all groups display a high degree of non-normality. Also, as expected, spreads

increase as we move down in ratings. The mean debt spread in the overall sample is 483

basis points, the maximum spread is 3939 basis points and the minimum is 1.9 basis

points. Interestingly, the standard deviation also increases as the rating deteriorates. Over

the sample period, the standard deviation is on the order of 25.3 to 809 basis points.

There is evidence of extreme movements in each group as the 90% and 10% values are

away from the mean by several times the standard deviation.

Panel B has the mean values, by group and for the overall sample, of some

country specific variables. Debt-to-reserves, debt-to-exports and political risk all increase

in value as move down in rating to signal a worsening of a country’s situation. I expect

19
these variables to have on average higher values as we move form high to low ratings,

and that is precisely what I find.

2.2.4 A model for sovereign spreads.

I estimate the following equation for each bond observation in the sample:

∆Spreadi,t = Constant + β1*∆Debt to foreign reserves ratioi,t + β2*∆Country risk

measurei,t + β3*∆U.S. Treasury yield curve level,t + β4*∆U.S. Treasury yield curve (1)

slope,t + β5*∆Local volatilityi,t-1 + β6*Local returnt-1 + β7*∆Years to maturityi,t + εi,t

Following earlier research, I estimate regressions on debt spread changes.15 To

estimate this equation, I decided to use an OLS model with Newey-West adjusted

errors.16 A priori, I expect the coefficients to have the signs described in Table 1. Table 3

shows the results of estimating equation (1) in four different rating groups. These groups

are similar to those presented in Table 2, except that the first and second groups from that

table were grouped together in Table 3.

The model seems to have a good fit, as measured by R-squared measures, which

range from 19% to 30%. For brevity, I will discuss only the results for the overall sample.

The debt-to-reserves ratio and the political risk measure both have a positive coefficient

15
Some previous research has been conducted on spread levels, for instance, Houweling et. al. (2002).
Cantor and Packer (1996) and Eichengreen and Moody (1998) run regressions on the log of the yield
spread.
16
I experimented with several other methodologies. I estimated equation 1 using OLS fixed effects,
grouping our sample by bond, by country, and by region. I also estimated FGLS (Feasible Generalized
Least Squares), OLS with panel corrected standard errors and OLS with Huber/White standard error
correction. All methodologies produced quantitatively and qualitatively similar results; however results
were more consistent using OLS coefficients with Newey-West adjusted errors. Results obtained with other
methods are not reported in this paper and are available from the author.
20
(as expected) and are highly significant. Two lags of the political risk variable were

included to account for the possibility of autocorrelation in this variable. These variables

measure the ability to service debt and the overall political and economic environment of

the issuer. An increase in political risk would signal higher instability and/or the

possibility of expropriation and therefore should be associated with a higher spread. An

increase in the debt-to-reserves ratio could be caused by an increase in the nominal debt

amount or a decrease in international reserves, both of which should be associated with a

higher spread. I also find that the coefficient estimates when using debt-to-exports in

place of debt-to-reserves are not significant and have the wrong sign, so they are not

reported.

The coefficient associated to the U.S. Treasury yield curve level is negative and

highly significant. Previous work had obtained insignificant positive coefficients (Cline

and Barnes, 1997; Min, 1998; and Kamin and Von Kleist, 1999), and significant negative

coefficients (Eichengreen and Mody, 1998). One interpretation of these negative

coefficients is that, as interest rates go up, low rated countries find it less convenient to

issue debt. Also, most structural models predict a negative relation because higher

interest rates increase the drift of the process followed by the firm’s (in this case,

country’s) value.17 A higher firm (country) value should be associated with a smaller

spread and hence the negative sign.

The coefficient associated with the U.S. Treasury slope term is always positive

and significant, however, this is unexpected. Following the expectations theory of interest

rates, a positively sloped yield curve signals higher future rates, which should be

17
Longstaff and Schwartz (1995).
21
associated with smaller spreads. Two reasons for this effect were previously mentioned.

On one hand, we could expect the average quality of sovereign issuers to increase

because low rated countries decide not to issue debt and this increase in overall quality

puts downward pressure on spreads. On the other hand, higher rates will mechanically

increase the distance to default in most Merton-based structural models which would also

lead to a decrease in spreads.

Local volatility is positive and highly significant, as expected. The local stock

return has the expected (negative) sign and also is significant. The coefficient on changes

of years to maturity is negative and not significant. I interpret this coefficient as evidence

of the existence of a survivorship bias in which only relatively better countries make it to

issue longer term debt, as explained by Helwege and Turner (1999) for the domestic case.

It may be the case that investors think that in the case of a default, short term maturities

are more risky than long term maturities since countries will usually default first on

issues with closer maturities, making short term issues riskier. The lack of consistent

cross-default clauses in some countries allows them to default or re-schedule debt

payments selectively. Finally, for a country facing financial difficulties, a longer time

horizon will provide the necessary time and maneuvering room to enact reforms and

measures that will allow the country to return to fiscal stability, effectively making longer

term debt less risky.

2.3 Debt spreads of domestic bonds.

In this section, I review the relevant literature on U.S. dollar denominated

domestic bond spreads (section 3.1). Then I discuss the variables used in the computation

22
of domestic spreads (section 3.2), and I describe the characteristics of the domestic bond

sample (section 3.3). I then proceed to estimate domestic debt spreads, discuss the results

and compute residuals (section 3.4).

2.3.1 Domestic debt literature.

The first structural model of risky debt is by Merton (1974). In this paper, Merton

used an option pricing approach to include systematic and idiosyncratic risk in the

calculation of the value of a put option on the firm’s value.18 In Merton’s model a firm

defaults on its debt when its assets are not enough to cover its outstanding obligations.

Default occurs when the firm’s value crosses from above a given threshold. The initial

model allowed for default only at maturity and was extended by Black and Cox (1976) to

allow for earlier default. Another extension was introduced by Longstaff and Schwartz

(1995) by incorporating stochastic interest rates. Strategic default was introduced in

models by Anderson and Sundaresan (1996) and Mella-Barral and Perraudin (1997).

Modeling endogenous corporate default was introduced by Leland (1994) and Leland and

Toft (1996). As these models need a fair amount of abstraction to achieve tractability, it

is not surprising that they prove to be difficult to implement and then almost always with

disappointing results (see Eom, Helwege and Huang (2003) for a review of the problems

and limitations faced by structural models).

This lack of results motivated some researchers to try another approach, using

reduced-form models, or intensity-based models. These models ignore firm-specific

18
Specifically, Merton’s (1974) model states that a risky zero-coupon bond has the same payoff structure
as a risk-free bond plus being short a put option on the firm’s value with a strike price equal to the face
value of the debt.
23
fundamentals and do not explicitly model the processes followed by the firm’s leverage

and/or value. Reduced-form models assume an unpredictable default process governed by

an exogenous hazard rate. For instance, Duffie and Singleton (1997) use a generic point

process and Lando (1998) uses a Cox process. Through extensive calibration, reduced

form models generally produced better results at explaining and forecasting yield spreads

than structural models.

More recently, Elton, Gruber, Agrawal and Mann (2001) tried to explain

corporate spreads using explanatory factors that included the probability of default, the

loss given default, and the difference in tax regimes. Collin-Dufresne et. al. (2001) tried

to explain changes in the credit risk portion of corporate spreads using data on spot rates,

reference yield curve slope, firms leverage and volatility, estimates for jumps in the

firm’s value and a proxy for the general business climate. Both papers, the former being

more of a reduced-form approach and the latter using a variables specified by a structural

framework, find similar results in that their models left a large portion of the cross-

sectional time variation of spreads unexplained, and further, they find that a single

common unknown factor could explain up to 75% of the residual variation. Huang and

Huang (2003) calibrate several classes of structural models to be consistent with the

recent history of observed defaults. They find that different models could generate the

wide range of credit spreads observed in the recent past, and further they provide some

evidence about the predictive power of such models.

24
2.3.2 Theoretical determinants of domestic debt spreads.

Structural models of domestic debt have identified variables that affect debt

spreads. In a manner consistent with the previous section, I put together three lists of

variables that should capture most of the debt spread variation. As with the sovereign

case, the first list also contains bond-specific variables, i.e., variables that vary within

bond issues, e.g. years to maturity. The second list contains variables that vary from firm

to firm but are the same for all bonds issued by firm (firm-specific variables). The third

list contains variables that are the same for all bonds in the domestic sample, and try to

capture changes in the U.S. interest rate term structure as well as changes in the U.S.

economic climate.

2.3.2.1 Bond-specific variables.

The bond-specific variable is years to maturity. The same arguments from section

2.2.1 apply here.

2.3.2.2 Firm-specific variables.

I choose two firm-specific variables following the basic spirit of Merton’s model

as presented in Stulz (2003). The first variable, leverage, has been used in previous

research as a successful proxy of a firm’s financial health. The second variable is the

volatility of a firm’s equity. A priori I expect a negative relation between each of these

two variables and debt spreads, since an increase on any of them would make default

more likely.

25
2.3.2.3 U.S. interest rate term structure.

The domestic sample is denominated in U.S. dollars. Therefore, I care about

factors that affect the U.S. yield curve term structure. Similar to the sovereign case, I use

the U.S. yield curve level and slope as explanatory factors of the term structure

(Litterman and Scheinkman, 1991). The arguments used in section 2.2.3 also apply here.

Since I collect data only on bonds issued by U.S. industrial firms, I assume their exposure

to the economic cycle is better captured by the S&P 500 index and therefore I collect

monthly returns for this index. Table 4 presents the predicted relations between the

variables previously mentioned and debt spreads.

2.3.3. Data description.

The domestic sample contains all U.S. denominated bonds issued by industrial

domestic firms. Applying the same selection criteria as those for the sovereign sample, I

end up with 2,493 live and 604 dead bonds issued by 649 different firms during the

January 1990 – January 2003 period for a total of 71,831 usable observations. This

sample differs from previous studies in at least three aspects. First, it covers a larger time

period than previous research. Second, I collect data for the entire universe of bonds

issued in U.S. dollars by domestic industrial firms, not only for those traded by any

specific group of investors. Third, the Fixed Income Database used in earlier studies like

Collin-Dufresne et al (2001) mainly covers investment grade bonds, and so results

obtained for high-yield bonds using that database might not be representative (Huang and

Kong, 2003).

26
 Debt spreads are computed in the same way as for the sovereign sample. Years to

maturity data is collected for each bond. The measures of the U.S. Treasury yield curve’s

level and slope are the same as the ones used in the previous section. The proxy for the

U.S. economic climate is the S&P 500 total return index from Datastream.

To compute the leverage ratio I collect book value of debt from COMPUSTAT

(items 45 and 51) and the market value of equity from CRSP. Leverage ratios are then

computed, following earlier literature, as:

( Book Value of Debt )

( Book Value of Debt + Equity Market Value)

Table 5 shows descriptive statistics for the domestic sample. Although it would be

desirable to classify this sample also by rating, however, Datastream’s coverage of

ratings is sketchy at best. In light of that problem, I decided to classify the data according

to leverage. As can be seen from the leverage columns of panel A, credit spreads increase

as firms become more levered. Further, the standard deviation of credit spreads also

increases with leverage. Data on the ‘No leverage data’ column refers to firms that are

either private or are not covered by COMPUSTAT. The presence of heavy tails in each

category is evident from the dispersion observed in the max, min, 10% and 90% values,

where the 10% and 90% values are several standard deviations away from the mean.

27
2.3.4 A model for domestic debt spreads.

I estimated the following equation for domestic bonds:

∆Spreadi,t = Constant + β1*∆Leverage ratioi,t + β2*∆Stock return volatilityi,t +

β3*∆U.S. Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + (2)

β5*∆S&P index returnt-1 + β6*∆Years to maturityi,t + εi,t

The results from these regressions are reported on Table 6. I estimated equation

(2) for each leverage group and for the overall sample. Clearly, the model performs better

in the highly leveraged group, as evidenced by the higher R-squared value. This result is

consistent with previous studies which find that structural models perform better for

longer maturity, lower rated sub-groups. As with the sovereign case, I will only go over

the overall sample results. On the coefficient of changes in years to maturity, the negative

but insignificant coefficient is consistent with previous results of Helwege and Turner

(1999) and with the basic Merton (1974) model predictions, as described in Stulz (2003),

for conservative levels of debt. The coefficients of lagged leverage and stock return

volatility –contemporaneous and lagged- are positive and strongly significant. The sign of

the U.S. yield curve level is also as expected and similar to the results obtained by earlier

studies.

In contrast to the sovereign sample, nothing conclusive can be said about the sign

and significance of the coefficient estimated for changes of the U.S. Treasury slope is

significantly positive in every specification. Finally, the S&P index return is significant

and with a negative sign, just as predicted by the theory.

28
2.4. Analyzing the common factor.

As previously mentioned, the goal of this chapter is to investigate whether the

common factor identified in domestic credit spread changes also is present in sovereign

debt spread changes. In this section, I establish the existence of a common factor in both

the residuals from the regressions on sovereign and domestic debt spread changes.

2.4.1 Establishing the existence of common factors.

In order to investigate whether common factors are present in the unexplained

variation in spreads, I use principal components analysis. This is a statistical technique

for data reduction whose objective is to find unit-length linear combinations of the

original variables that capture the maximum variance. I apply principal component

analysis to the residuals obtained from the regressions discussed in previous sections to

verify whether the unexplained variation is truly noise or whether there is evidence of a

common factor driving this unexplained portion of the variance of credit spread

changes.19

The first problem faced when applying principal components analysis is how to

organize unbalanced panels in the most efficient form. Research in this area conducted by

Boivin and Ng (2003) shows that more data is not always better when conducting this

type of factor analysis. In fact, in their forecast exercise they show that factors extracted

from as few as 40 variables could be more informative than factors extracted from all 147

series in their setup. Basically, their result obtains because of large cross-correlation in

19
The serial correlation of the residuals from the regressions for sovereign yield changes is -0.1513 with p
value of 0.2290 and 0.0068 with p value of 0.9345 for domestic yield changes.
29
errors and of small variability of the common components. Sadly, until today there is no

guide as to what data should be included in a principal component analysis or what is the

optimal number of series to include in this exercise. Recent work from Scherer and

Avellaneda (2000) applies principal component analysis to eight variables only,

effectively using one or two bonds per country in their study. While this might be a

solution for the sovereign case where it is easier to identify benchmark bonds for each

issuer, this is not feasible in the domestic sample. This sample has bonds from 649

different firms, and in many cases there are tens of bonds outstanding from a given firm.

Also, applying principal component analysis to all the bonds in the domestic sample is

not a good idea, since that would most certainly only increase the amount of statistical

noise while adding very little or no new information at all.

I decide to follow the approach implemented by Collin-Dufresne et. al. (2001) and

create groups or ‘bins’ of data to efficiently summarize the information content of the

residuals. I divide each sample (sovereign and domestic) in three maturity categories and

three leverage (debt-to-reserves, in the sovereign case) categories, creating a total of nine

‘bins’ in each sample. Then, each observation is assigned to a bin. I estimate again

equations (1) (for the sovereign bins) and (2) (for the domestic bins), compute the

residuals, and calculate averages across residuals for each bin. Table 7 shows the

correlation structure for the average domestic residuals (panel A), the average sovereign

residuals (panel B) and for all averages –domestic and sovereign (panel C). The average

correlation for the sovereign sample is 0.75, and 0.87 for the domestic sample. To

investigate whether the relatively high correlations found in panel A and C are caused by

a common component, I proceed to conduct principal component analysis.

30
 Table 8 shows the results of applying this eigenvalue decomposition to the bins

constructed earlier. Panel A shows strong evidence of the existence of a common factor

in sovereign spreads. The first common factor explains 76.09% of the variation, as shown

by the proportion of the first eigenvalue. The second common component explains the

remaining variance that is orthogonal to the first common component. It is difficult to

interpret the second component because its eigenvalue is well below the value of the first

eigenvalue and is much closer to the third eigenvalue. However, if this is to be interpreted

as evidence of a second common factor, it would explain an additional 20.37%.

According to Scherer and Avellaneda (2000), a number between 65% and 80% for the

first common component would be considered as indicative of strong co-movement

characterized by a high correlation in the spread changes. My results are consistent with

their result – obtained with spreads computed from Brady issues for selected countries –

in which they found evidence of two common factors driving most of the variation in

spread changes.

Panel B of Table 8 also shows strong evidence of the existence of a common

factor to all domestic spreads. The first common component explains 86.23% of the

variance. There is weak evidence on the existence of a second component which explains

an additional 8.53% of the variance. The existence of a first common factor that explains

such a large portion of the variance is consistent with previous research, e.g., Collin-

Dufresne et. al. (2001). The existence of a second common factor has not been

documented for domestic debt before, but this could be due to the fact that I am using a

larger dataset and that I am looking at a longer time period that earlier studies.

31
 Finally, panel C has the results of looking at the common components of both

groups of bonds, sovereign and domestic. Interestingly, I find no evidence suggestive of

the existence of a common factor to both groups of bonds. The first common factor

explains 42.06% of the residual variance of spread changes, while the second factor

explains an additional 33.12%. As mentioned before, Scherer and Avellaneda (2000)

consider a value of 65% for the first common component as the lower boundary for a

weak coupling, or correlation, between spread changes. The result I obtain is puzzling

because if the market for dollar-denominated credit-risky bonds is integrated, and if the

common components I find can be explained by liquidity shocks, then such shocks

should be pervasive across markets (Chen, Lesmond, and Wei, 2002; Chordia, Sarkar,

and Subrahmanyam, 2003; Kamara, 1994). According to panel C, this is not what is

happening.

In order to shed more light on the issue of whether the common factor identified

in both samples is indeed the same in both groups, I extract the first and second common

components of each sample to compare them. These common factors are plotted in figure

1. The interpretation of the units in the y-axis is as follows. To compute the principal

components, I analyzed the correlation matrix. This is equivalent to all the variables to

having mean 0 and standard deviation 1. Thus, the common factors are expressed in

terms of these standardized variables. Loosely speaking, the units on the y-axis in both

figures can be interpreted as percentage points.

The pattern seems to suggest a lead-lag relation between the first factor from the

domestic sample and the first factor extracted from the sovereign sample. Figure 2 shows

the second common component extracted from both samples. The figure seems to suggest

32
a weak contemporaneous relation. These issues will be investigated further in the next

sections.

2.4.2 Explanatory power of the extracted components.

In this section, I examine whether these common factors have explanatory power

over the cross-section of debt spreads for the other type of debt. I estimate again

equations (1) and (2) including in each equation the two common components extracted

from the other group, i.e., I include the factors extracted from the sovereign sample into

the domestic sample regression and vice versa. Results are shown on Table 9.

First, I will talk about the sovereign regression when the domestic common

components are included. Looking at rating categories, the explanatory power of the

equation for the lower rated group (B- to C) increases, as measured by the increases of R-

squared statistic from 22% to 31%. This sub-sample is the one with the least number of

observations. There is, however, no gain in explanatory power in the overall sample.

Further, in every case only the contemporaneous value of the first component from the

domestic sample is significant. In the case of the second component extracted from the

domestic sample, only the lagged value of the second factor is significant for all sub-

groups.

The domestic sample has strikingly different results. In this case, I included in the

domestic spread changes equation both common components extracted from the

sovereign sample. The explanatory power of the overall equation is increased almost by

30%, from an R-square value of 0.09 to 0.12. The only significant common component

coefficient is the contemporaneous effect of the first common component from the

33
sovereign sample. The explanatory power in most domestic sub-samples increases by a

similar percentage as the R-squared value of the overall sample. Overall, I interpret these

results as evidence of the existence of a relation between the first common component

extracted from the sovereign spread changes and domestic debt spread changes. The

dynamics of the relation between the common components extracted from each type of

debt is investigated in the next section.

2.5. Looking into the information content of the common factors.

The principal component analysis conducted neither provides information on the

dynamics of the factors identified nor provides an economic interpretation of them. In

this section I investigate the contemporaneous and inter-temporal relation between factors

and also investigate whether these factors might be capturing liquidity and/or

supply/demand shocks.

2.5.1 Lead-lag relations.

The picture shown in figure 1 suggests the possibility of an intertemporal relation

between the first factor extracted from the sovereign sample and the first factor extracted

from the domestic sample. Using a vector-autoregression approach, I investigate the

possibility of one of these markets acting as an early signal for potential problems that

can affect the bond market in general. Previous work like Joutz and Maxwell (2002) and

Cifarelli and Paladino (2002) have applied VAR procedures in a credit spread framework

to study the relation between credit spreads from different countries. More recently,

Longstaff, Mithal and Neis (2003) applied a VAR framework to study the relation

34
between bond and credit derivatives markets. To explore the lead-lag relation between

the sovereign and the domestic factors, the following simple vector-autoregression

specification is used:

k k
FacSovt = a1 + ∑ β1 j FacSovt − j + ∑ γ 1 j FacDomt − j + δ 1 X t +ε 1
j =1 j =1

k k
FacDomt = a2 + ∑ β 2 j FacSovt − j + ∑ γ 2 j FacDomt − j + δ 2 X t +ε 2
j =1 j =1

Table 10 shows the results for the simple case when k is equal to two. Both the

Akaike Information and Schwartz criteria suggest that a VAR system of two lags is

warranted by the data. I first run the VAR model without exogenous variables to have an

initial idea of the lead-lag structure. For brevity, I only report the R-squared value for

each equation and also because the basic lead-lag relation is unchanged when the

exogenous variables are included. I then run the VAR model with exogenous variables.

These exogenous variables are chosen to capture liquidity and supply/demand effects.

Most previous studies dealing with credit spreads specifically abstain from liquidity

effects because of the lack of consensus on how to measure and model liquidity premium

affecting spreads (Chen, Lesmond and Wei, 2003). Longstaff, Mithal and Neis (2003)

study the consistency of the price of credit risk between the bond and derivatives

markets. They find that the implied cost of credit is higher in the bond market than in the

credit derivative market, and advance a possible explanation for this based on the

existence of a liquidity component in debt spreads. Their measure for this liquidity

35
premium is the difference between the price of credit risk in the bond and credit

derivative markets.20

Since all the bonds in the sample are denominated in U.S. dollars and they all

trade in U.S. financial markets, I am interested in variables that measure the overall

liquidity in these markets. I use two general measures of liquidity. The first proxy is the

difference in yield between the on-the-run21 thirty year U.S. Treasury bond and the most

recent off-the-run bond is computed. Off-the-run bonds are bonds that whilst not being

the most recently issued in a certain maturity range, are very similar to the on-the-run

issue in all respects. Therefore, any differences in prices –and therefore in yields- is

usually considered to be due to liquidity. As liquidity dries up, this difference is expected

to decrease.

The second proxy for general liquidity in the market is the net borrowed reserves

from the Federal Reserve,22 which is considered a measure of the monetary stance. A

loose monetary policy usually implies an increase in liquidity via the decrease of credit

constraints. Harvey and Huang (2002) showed that the Federal Reserve, through its

ability of changing the money supply, impacts the trading of bonds and currencies.

Following Chordia, Sarkar and Subrahmanyan (2003) I define net borrowed reserves as

total borrowing minus extended credit minus excess reserves, divided by total reserves.

Since borrowed reserves represent the amount that banks are short to satisfy the Fed’s

requirements, a lower value of this measure indicates looser monetary conditions.

20
Collin-Dufresne et. al. (2001) point that Chakravarty and Sarkar (1999), Hotchkiss and Ronen (1999) and
Schultz (1999) found evidence of the existence of relatively high transaction costs and low volume in bond
markets, and therefore, Collin-Dufresne et. al. (2001) interpret these results as evidence of a liquidity
premium.
21
An on-the-run bond is the most recently issued (and typically the most liquid) government bond with a
given maturity.
22
See Chordia, Sarkar and Subrahmanyan (2003).
36
 To capture possible supply/demand shocks I collect data from the Investment

Company Institute (ICI) on the monthly flows into mutual funds. ICI’s statistics are

collected from approximately 8,300 mutual funds, and are divided in flows into equity

funds and bond funds. These measures could potentially capture changes in investor’s

attitudes towards risk or any other supply/demand shocks unrelated to overall market

liquidity.23

Table 10 reports the results of the VAR model with the exogenous variables. It

seems that the sovereign factors have explanatory power over the domestic factors but

not the other way around. I will discuss each one of the four equations in the VAR model,

starting with the first common domestic factor. The second lag of the first sovereign

factor is significant in the regression for the first domestic factor. The flows to stocks and

flows to funds variables are negative and significant. This equation is the one with the

smallest gain in R-squared when including the exogenous variables.

The first sovereign factor seems to be slightly autoregressive from the barely

significant coefficient for its own first lag. Coefficients for the domestic factor lags are

not significant, while all the coefficients for the exogenous variables are highly

significant. It seems as if this factor is capturing both liquidity and demand shocks as

implied by the coefficients associated with the exogenous variables. This equation also

has the highest increase in explanatory power, since the R-squared increased from 0.05 to

0.37 when the exogenous variables are added to the specification.

23
Of course, we have to consider the possibility of endogeneity in our variables, since for instance, a
change in the Fed’s stance could make certain markets more attractive and influence the flows into those
markets. No effort is made at this time to address this concern.
37
 The second domestic factor equation shows significant coefficients for the first

lag of both sovereign factors as well as for its own second lag. The flow variables come

out both significant, as well the net reserves coefficient.

The second sovereign factor equation has the highest R-squared value, at 0.67.

Second lag coefficients for both sovereign factors are significant, as are both lags of the

second domestic factor. The flow variables and the net reserves measure are also highly

significant.

Overall, the exogenous variables of the VAR do capture a significant portion of

the time variation of the factors extracted from the sovereign and the domestic sample.

There also is evidence in support of the aforementioned result that sovereign common

components are related to domestic spread changes but not the other way around, because

the lags of the sovereign common components have significant coefficients in the

equations for the domestic common components but the domestic common components

do not appear to have explanatory power in the equations for sovereign common factors.

Also, there is evidence of an inter-temporal relation going from the first sovereign

common component to the first domestic common component. This is expected from

figure 1. I find that all four common factors are related to the flows of money going into

equity and bond funds, as measured by the Investment Company Institute (ICI), while

only the second common component of each group is related to a macroeconomic

measure of liquidity, namely the net borrowed reserves form the Federal Reserve. These

results are consistent with previous literature that has concluded that the unexplained

variation in credit spreads could be caused by liquidity and supply/demand shocks.

38
 The pattern displayed by the first common factors in figure 1 raises the concern

that my VAR results could be driven by the large spike observed around October 1998.

As a robustness check, I re-estimate the system excluding the observations for

September, October and November 1998.24 As expected, the significant loading of the

second lag of the first sovereign common component onto the first domestic common

component disappears. Interestingly, however, the coefficient associated with the second

lag of the second sovereign factor now becomes significant, whereas it was not

significant before. The explanatory power of the exogenous variables and the overall R-

squared values of the system remain unchanged. Overall, the evidence suggest that the

asymmetric relation observed between the first common factor from the domestic sample

and the common factors from the sovereign sample remains even after excluding the

large shock of October 1998 from the sample.

2.6. Conclusions and future work

The availability in recent years of a panel of observations on sovereign bond

yields provides with a unique instrument whose dynamics can shed some light in the

study of the determinants of debt for countries and firms. In this chapter I identified the

existence of a two strong common components, unrelated to credit risk and distinct for

each type of debt, in credit spreads of sovereign and domestic bonds. Using a vector

autoregressive (VAR) model, I find that domestic spreads are related to the lagged first

common component of sovereign spreads. While there is no contemporaneous common

component in bond spreads, there seems to be a common component when focusing on

24
Results are not reported but are available upon request.
39
the dynamics of these spreads. Traditional macro liquidity variables are related to the

common components found in domestic and sovereign spread changes.

I will conduct further research to shed light on why the relation between domestic

corporate credit spreads and sovereign credit spreads is not contemporaneous, as

expected in a fully integrated market. This could be due to differences in liquidity and

infrequent trading or maybe it reflects a market inefficiency. My results also are

surprising since they suggest that the cost of debt for emerging markets depends mostly

on country and emerging-market specific considerations, and this is surprising when

considering results obtained by previous literature emphasizing the impact of developed

country developments for capital flows into emerging markets.

In this chapter, I contributed to the literature by showing that current structural

models of debt spreads can be improved if these findings are incorporated in them. To the

extent that investors depend on these models to hedge the credit risk of their bond

positions, they can benefit from a better understanding of the determinants of credit

spreads changes. My research also shows that, after taking into account the dynamics of

the common components in credit spreads across debt types, the cost of debt for firms

and countries depends to some extent on shocks that affect all types of debt.

40
 CHAPTER 3

LATIN AMERICAN AND U.S. EQUITIES RETURN LINKAGES: AN EXTREME

VALUE APPROACH

3.1. Introduction

The last fifteen years of the past century witnessed an important increase in

financial and economic integration between Latin American countries and developed

countries. Due to geographical and trade considerations, most of the ongoing integration

took place between Latin America and the United States. This is not to say that

integration with other developed countries, for instance countries members of the

European Union, is not significant, but rather that integration with the U.S. happens to be

more important both in the public perception and in relative terms.

One of the many benefits of this increased integration is the newly available

supply of financial assets exhibiting low correlation with the U.S. financial markets.

Mean-variance optimizing investors looking to diversify their portfolios have poured vast

amounts of resources into Latin American stock markets as they were being liberalized.

Among the new instruments available to U.S. investors pursuing the benefits of

international diversification were closed-end funds, open-end funds, American

Depositary Receipts (ADRs), and foreign indices mimicking funds.

41
 However, there was a dark side to financial integration and openness. Financial

crisis, in the form of negative financial markets shocks, also moved easily across borders.

In addition, the propagation of financial crisis was facilitated by the newly increased

correlations. This phenomenon, usually referred to as financial contagion, has attracted

much attention in recent years from researchers. Recent experiences with the Mexican

devaluation of 1994 – “the tequila effect”-, the 1997 East Asian crisis, the 1998 Russian

meltdown -“the vodka effect”- and the subsequent Brazilian devaluation –the “samba

effect”-, which sent most financial markets around the world tumbling, stressed the

importance for both practitioners and academics of studying and identifying the forces

behind these phenomena.

This chapter addresses the behavior of linkages between financial assets using a

statistical technique known as Extreme Value Theory (EVT). EVT is the study of outliers

or extremal events. Since large movements in returns are usually characteristic of

financial crisis, and these large movements can be considered as outliers, the use of EVT

seems to be warranted. This approach has several advantages. First among these are the

well known results on asymptotic behavior of the distribution of very high quantiles.

Second, no assumptions are needed about the true underlying distribution that originated

data in the first place. Since financial contagion usually comes into scene during periods

of very high distress, it is a natural area on which to use a technique that focuses on the

tails of a distribution function.

My results show that for the six emerging countries analyzed, only in the case of

Mexico the correlations in the extreme are higher between the locally traded stocks and

the S&P than for the corresponding ADRs and the S&P. This result suggest that the

42
contagion mechanism for Mexico is different than for the rest of the countries analyzed

here, in that the shocks appear to be propagated directly into the Mexican stock exchange

rather than through New York via the ADR issues trading there. Another finding

This chapter contributes in the literature in a number of ways. First, it builds on

the work of Longin (1996) and Longin and Solnik (2001) applying EVT in finance by

examining the linkage between financial assets available to U.S. investors looking for

international exposure before and after main events such as the 1995 Mexican crisis.

Second, it adds to the growing literature on financial contagion by employing a more

“appropriate” statistical technique for the often temporary but large movements in prices.

Third, related to the first contribution, this chapter also contributes to the empirical asset

pricing literature because of its focus on a relatively new statistical application to the

transmission of information and prices. This is useful as sectors of the U.S. markets

become more volatile. Fourth, this chapter contributes to the risk management literature.

The basic analysis one EVT can easily applied to the other assets, such as energy-based

and commodities derivatives, which also exhibit temporary but sharp price movements.

This is important as the government contains to deregulate selective energy markets.

Finally, it contributes to the research on international asset pricing and allocation. The

analysis on U.S. and Latin American financial assets also is useful for professional

international portfolio managers.

The chapter is organized as follows: the next section discusses some basic results

from Extreme Value Theory and how they are applied in a financial framework. Section

III presents an empirical analysis of the data. In this section, bivariate extreme value

measures are applied on six different country pairs between the U.S. S&P500 Index and

43
each of the following countries: Argentina, Brazil, Chile, Colombia, Mexico and

Venezuela. Section IV presents an extension of the analysis, searching for structural

changes in linkages following the Mexican crisis. Section V concludes and outlines some

ideas for future research.

3.2. Literature Review

In recent years, extreme value theory techniques have been applied successfully to

three related fields in finance: value at risk (VaR), the financial contagion literature and

the literature that studies the return-volume relation. Brooks, et al (2005) test several

EVT models by fitting them to different series of futures contracts in order to determine

which model is more efficient for value at risk purposes. In an univariate analysis, they

find that a model that treats the tail differently while at the same time incorporating

information from the rest of the distribution yields superior results to a generic GARCH

(1,1) model. Gencay and Selcuk (2004) apply univariate EVT to investigate performance

of VaR models in emerging markets, documenting the thresholds and probabilities of

stock crashes. Poon and Lin (2000) use univariate EVT to show that an internationally

diversified portfolio will dominate a US portfolio. The reach that conclusion looking at

probability of loss for each market using a tail index to study the distribution of returns.

In a related exercise, Susmel (2001) applies the safety-first principle developed by

Roy (1952) to compute levels of (negative) returns that would wipe out an investor’s

wealth. He applies univariate EVT analysis to study the tail distribution of several Latin

American markets. He finds that investors can benefit from inclusion of these markets in

their portfolios –regardless the fatter tails exhibited by Latin American markets. His

44
results suggest an optimal 15% allocation to these markets. Hartmann, Straetmans, and de

Vries (2001) study stock-bond contagion within and across five developed countries

applying an innovative a non-parametric measure of dependence in the extremes of the

return distributions. They find that simultaneous crashes in stock markets are twice as

likely as simultaneous crashes in bond markets. Also, stock-bond and bond-stock

contagion is equally as likely. Finally, cross-border linkages are very similar to national

linkages.

Qi (2001) focus is to answer whether unexpected volume provides information

and therefore moves prices and to do this, he studies the return-volume relation in the

tails using bivariate EVT. In almost all cases the results point to a positive correlation

between absolute return and trading volume. This relation is not symmetric, i.e., it is

stronger in the right tail than in the left tail. Reich and Wegmann (2002) analyze a sample

of Swiss stocks in order to study the relation between market value and the relative bid-

ask spread. Using bivariate EVT and using Streatmans’ non-parametric approach to

extreme linkages they document that extreme movements in both variables are not

independent. Marsch and Wagner (2004) use bivariate EVT to reject the null of

independence in the extremes of returns and volume in five of seven developed countries

in their sample. Further, they find that relation to be symmetrical only in the US, i.e., the

same for negative and positive returns.

Jondeau and Rockinger (2001) study stock returns in 20 countries using univariate

EVT. Surprisingly, they cannot reject that the left and right tail indices are the same, and

they cannot reject that the tail index is the same for all countries. They do find disparity

with respect to where the extremes are located (i.e., a 10% price drop might not be an

45
extreme observation in every country). Poon, Rockinger and Tawn (2002) use non-

parametric measures to identify and quantify tail dependence among international stock

markets. Using data from 1968 to 2000 for US, UK, Germany, France and Japan they

find left-tail dependence to be stronger that right tail dependence. They also document

that these stock index returns do not exhibit asymptotic dependence. Schich (2002) looks

at European stock indices (Germany, UK, France, Netherlands and Italy) from 1973 to

2001. Using Starica (2000) measure of dependence in a bivariate EVT framework he

finds that measures of dependence have grown over time and are higher for negative than

for positive returns. He also finds that the bivariate correlation with Germany is similar

for all the countries in the sample. Chan-Lau, Mathieson and Yao (1998) study how

contagion differs across countries. They conclude that contagion patterns differ

significantly within and across regions, that Latin America shows the largest increases in

contagion and that contagion is higher for negative returns than for positive returns.

Several studies have focused on the study of firms within one country. Tolikas

and Brown (2003) apply univariate EVT on individual Greek stocks and find that the tails

have become less fat tailed over time. Brännäs, Shahiduzzaman and Simonsen (2002)

also use univariate EVT to determine that the tails of Swedish stocks are better described

using a Fréchet distribution. Gencay and Selcuk (2001) study overnight borrowing rates

in an interbank money market in Turkey and describe the characteristics of the tail of

their distribution.

Straetmans, Verschoor, Wolff (2003) look at the U.S. stock markets before and

after 9/11. They find no evidence for a structural change in downside risk measured

before and after 9/11. They do find, though, that the probability of joint co-exceedances

46
over thresholds between different sectors and the market portfolio has risen when taking

9/11 as the sample midpoint.

3.2.1. The Univariate Case

Extreme value theory (EVT) has been extensively studied in mathematics and

statistics. One of its earliest applications, and in fact one of the branches of science that

motivated much of the work in EVT, is hydrology. Classical applications are the

calculations for the optimal height of dykes in the Netherlands as well as the calculations

of level of tides in the U.K. Engineers also are fond of EVT when it comes to calculate

how resistant some structures are to wind and sea forces. In finance, EVT is a useful

supplementary risk measure because it provides more appropriate distributions to fit

extreme events and it has found a niche in value at risk (VaR) and insurance applications.

Basically, EVT involves modeling the tail of a marginal distribution as well the

dependence structure of extreme observations. This modeling of the asymptotic

distribution can be done without making assumptions about the form of the original

underlying distribution that produced the data in the first place. This is very desirable

since the original distribution is generally unknown. Readers interested in a more

technical discussion are encouraged to read the seminal work by Gumbel (1961).

Since this chapter makes no attempt at contribution to the existing EVT theory,

we will only go over the main results of the theory, highlighting the results that are

important for our analysis. We will describe the main results of univariate EVT. Then

building on those results, we will briefly explain the main results extending to bivariate

EVT.

47
 For any time series, we can define extremes in two ways. The first approach is to

divide the series in blocks of equal size and take the maxima (the analysis for the minima

can be done by simply using the negative of the series) of each block. For instance, if we

have daily returns for a period of many years, we can divide the data in annual semi

annual blocks and then proceed to take the maximum value from each block. The other

approach to define extremes is to define a high threshold and consider all observations

that exceed the threshold. The first significant result is, for the first case (taking

maxima), that the asymptotic distribution of the maxima is shown -under certain

conditions- to converge to a Gumbel, Fréchet or Weibull distribution. A standard form of

these distributions is known as a Generalized Extreme Value distribution (GEV). The

second significant result, for the threshold approach where one is interested in modeling

the behavior of the exceedances, is that the limiting distribution is a Generalized Pareto

Distribution (GPD).

We will now talk about the first approach. Let {X1,...,Xn} be a sequence of

identically independent distributed random variables. The asymptotic distribution of a

series of maxima (minima) has been shown (Fisher and Tippet, 1928; Gnedenko, 1943)

to converge to a Generalized Extreme Value (GEV) distribution of the form:

⎧ exp( − [1 + ξ ( x − µ ) / σ ] −1/ξ ) ifξ ≠ 0

H (ξ ,µ ,σ ) = ⎨ − ( x − µ )/σ
⎩ exp( − e ) ifξ = 0

The parameters µ and σ represent, as usual, a scalar and a tendency. The

parameter ξ is called the tail index and it is positively related to the thickness of the tail,

i.e., the larger the tail index, the thicker the tail. Depending on the value of the tail index

we have three possible scenarios: ξ >0 which corresponds to a Fréchet distribution, ξ =0

48
which represents a Gumbel distribution and ξ <0 which represents a Weibull distribution.

Empirically, it seems to be the case that the distribution is the best fit for fat tailed

financial series.

The second approach, known in the literature as the POT (peaks over thresholds)

approach, involves estimating the conditional distribution once a high threshold is

reached.

Let Fu be the distribution of exceedances of X over a threshold u, such that

F u ( x ) = P( X − u ≤ x X > u )

The determination of the threshold u is not trivial and will be briefly addressed

further ahead. For now, we will only mention that it is subject to minimizing the mean

squared error in the parameter estimates. If we set a high threshold, you get a less reliable

estimate due to the lack of data. On the other side, if we set a threshold to low, you get

more precise, albeit biased, estimates since we included some points that did not belong

in the tail. Once the threshold has been determined, we can write

F u ( x ) ≈ G ξ , β ( u ) ( x ), u → ∞

where

− 1/ ξ
⎧ ⎛ ξx⎞
⎪1 − ⎜ 1 + ⎟ ifξ ≠ 0
Gξ ,β ( u ) ( x ) = ⎨ ⎝ β ⎠
⎪ 1 − e − x/β ifξ = 0
⎩

This result shows how the conditional distribution Fu is approximated by a

Generalized Pareto Distribution (GPD). The case where ξ = 0 gives us the exponential

distribution. Any value of ξ > -0.5 is considered a fat tailed distribution.

49
 Application of EVT requires some previous analysis to make sure the data

exhibits fat tails, justifying the use of EVT. In this early stage analysis we usually

perform a visual inspection of the data with the help of Q-Q plots and mean excess

function graphs. A mean excess function graph describes the expected overshoot of a

threshold given that an exceedance occurs. Fat tailed distributions show mean excess

functions tending towards infinity for high thresholds u (linear shape with positive slope).

As mentioned before, the choice of a threshold is subject to the usual trade off

between variance and bias. One solution would be to use a mean excess graph and choose

a threshold that yields a reasonable linear shape. Another –and most common solution– is

to calculate the value of the Hill estimator. The Hill estimator is a maximum likelihood

estimator for a GPD and it is a very popular estimator of ξ. Because the Hill estimator is a

maximum likelihood estimator, we know that the parameter values are chosen to

maximize the joint probability density of the observations.

3.2.2. The Bivariate Case

The first problem faced in extending EVT to a bivariate case is that such

extension is not straightforward. The reason is that in higher order Euclidean space there

is no standard notion of order and thus no standard notion of extremes (Embrechts, et. al.,

1999). The current multivariate EVT allows only for low dimensional problems. Truly

multivariate EVT is not under the current theory reach. Longin (1999) established that the

joint distribution of extreme marginal distributions is not necessarily the distribution of

the extremes for an aggregate position. Still, the bivariate estimation is done in two parts:

univariate estimation of tail indices first, followed by the estimation of a “uniformised”

50
dependency function. Even though no natural parametric family exists for the dependence

function, Longin and Solnik (2001) proposed the use of a logistic dependence function to

fit the joint tails in a bivariate EVT framework. The logistic dependence function was

first proposed by Gumbel and therefore is also referred to in the literature as a Gumbel

dependence function. This is its functional form:

Dl ( y1 , y2 ) = ( y1 −1/α + y 2 −1/α ) α

The main reason for using this dependence function is that the dependence

coefficient α is related to the correlation coefficient of extremes, ρ, via the following

functional form:

ρ = 1− α 2

Summarizing the bivariate case, we will first get univariate estimates for each

distribution, then fit the dependence function. As we use different thresholds to calculate

the tail indices, we will also get different values of ρ. This property is crucial in that it

will allow us to calculate in a straightforward manner how the correlation changes as we

move further into the tails.

3.3. Data

We collected daily data from Datastream for the period between 12/29/1989 and

12/29/2000 --2872 days-- for six Latin American countries and the U.S. The countries are

Argentina, Brazil, Chile, Colombia, Mexico and Venezuela. For each Latin American

country, we collected the return index of the local stock market and for every live ADR

issue trading in the U.S. We also collected the return index of the S&P 500. For every

asset, log returns in dollars were obtained. Also, an equally weighted basket of ADRs for
51
each country was calculated. ADRs were chosen because they represent an alternative to

investors who want to get foreign stock market exposure. Table 11 shows the summary

statistics of the variables.

The next step was to conduct an exploratory data analysis or visual inspection of

each series, two for each country plus the S&P index. Through the use of Q-Q plots and

mean excess graphs we verified that the series exhibit fat tails and we got initial ranges

for the tail indices.

The Q-Q plot (graph of quantiles) helps to assess the goodness of fit of data to a

parametric model. The more linear the Q-Q plot, the better the goodness of fit. It is also

helpful because on it outliers can easily be detected. The following is the mathematical

characterization of a Q-Q plot:

⎧ −1 ⎛ n − k + 1 ⎞ ⎫
⎨ X k ,n , F ⎜ ⎟ , k = 1,..., n⎬
⎩ ⎝ n ⎠ ⎭

Figures 3 to 5 show the Q-Q plots for the left and right tail of the Mexican equity

dollar return, the Mexican ADR equally weighted portfolio and the S&P500 index. In all

of them the presence of fat tails is obvious based on visual inspection.

The next step of the exploratory analysis was to graph the mean excess function

e(u). e(u) is the mean excess over the threshold u. As discussed above, it describes the

expected overshoot of a threshold given that an exceedance occurs. Fat tailed

distributions show e(u) tending towards infinity for high threshold u (linear shape with

positive slope).

52
 This is exactly the pattern that is observed on figures 6 to 8 for the same variables

as in the Q-Q plots. Mathematically, e(u) is:

e( u ) = E ( x − u X > u )

and the plot is given by the points in:

{X k ,n }
, en ( X k ,n ), k = 1,..., n

The complete set of Q-Q plots and mean excess graphs for all variables is

available from the author.

As outlined in the previous section, univariate analysis is a pre-requisite of

bivariate analysis. An interesting product of the univariate analysis is the possibility of

calculating probabilities for out of sample events. In plain terms, we can calculate for

instance, given a 20 year time series of daily returns, the maximum loss that is to be

exceeded every x years. McNeil (1997) has a very good example on this.

However, this chapter is interested in asset linkages, and therefore we will not talk

about the conditional univariate probabilities of exceeding a certain threshold. Our final

goal is to be able to make inferences on the pattern displayed by the correlation in the

extremes, using a framework à la Longin and Solnik (2001). Our exercise here is to

calculate pairwise correlations between the S&P and the local domestic stock indices and

compare these correlations with the correlations obtained from another pair wise

calculation using the S&P and a basket of ADRs. Since in theory an investor can get the

same international exposure from buying shares in the local stock markets than from

buying ADRs in the U.S., we would not expect to find any differences in the behavior of

correlations as we move further into the tails of the marginal distributions. Curiously, this

53
seems not to be the case for some countries considered like the most representative of

economic and financial liberalization, namely Mexico, Chile and Brazil. Figures 2 to 8

show the main product of this chapter. Each graph shows how extreme correlations

(measured in the vertical axis) change value as I increase the threshold used to obtain the

sample (measured in the horizontal axis). Points to the right of the vertical axis show us

the behavior of the positive -right- tail, whereas points to the left of the vertical axis show

the correlations in the negative -left- tail.

As just mentioned, Mexico, Chile and Brazil seems to show asymmetric

correlations which we think are worth further research. Colombia seems promising but

the lack of data makes the inference process dubious. Finally, it is also worth

investigating what is it (or was?) about the Argentinean market that make it behave in

such a “well behaved” form during the period in question.

In the ideal case of bivariate normal behavior, we would expect to see a well

behaved bell shape. However, some of the countries offer some evidence of increased

correlations in the negative tails, consistent with all the previous work on financial

contagion. Whether this increased negative extreme correlations are caused by liquidity

problems, heteroskedastic time series or some other explanation is beyond the scope of

this work. For now, we will just show that correlations display an asymmetric behavior

and that S&P - ADR correlations do not always behave in the same way as S&P - local

index correlations. We consider this result significant and we hope that further research

will allow us to draw conclusions about whether the transmission mechanism of financial

crisis goes through the U.S. stock markets -via ADRs- or moves directly to the local

stock markets.

54
3.4. A Small Test for the Mexican Pairs.

Since the Mexican pairs displayed differences both between left and right tail

correlations and between S&P - ADRs and S&P - local stock index, we decided to use

this country to test whether the extreme correlations changed after the 1995 Mexican

crisis. Figures 9 and 10 show the experiment. We divided the sample in two, the first

sample going from 12/29/1989 to 12/31/1994 and the second sample running from

1/3/1995 to 12/29/2000. Absent some formal test, we limit the analysis to a visual

inspection of the behavior of the extreme correlations. At first sight, it seems that the

asymmetric pattern exhibited by the whole sample is driven primarily by the post 1995

period. Let us be very clear about the fragility of any assumption stated in these sections.

As mentioned before, extreme value parameters depend heavily on the size of the sample,

namely the extreme observations. Dividing the original sample in two probably did hurt

the accuracy of the estimates. More work on this area is in progress and it will be

included in future versions. However, the visual pattern is suggestive of evidence of

changes in the value of extreme correlations.

3.5. Concluding remarks

This work is still in its very early stages. However, some promising results are

already showing. Applying bivariate extreme value techniques to pairs formed by Latin

American local stock indices vs. the S&P and Latin American ADRs vs. the S&P we

found visual evidence of non-bivariate normal behavior. More formal Wald test are in the

agenda for future research. Also, we neglected in this chapter a third channel through

55
which a U.S. investor can gain foreign exposure: country-funds. We intend to include

pairs formed by country-funds vs. the S&P and compare the behavior of their extreme

correlations with the ones already obtained and check whether they can offer more clues

about the transmission mechanism of financial shocks.

Finally, the last point in the research agenda involves a departure from the

Gumbel logistic dependence function. Hartmann, Straetmanns and de Vries (2001) show

that other dependence functions do a much better job at fitting the joint marginal extreme

distributions. Since the only foundation for the use of a Gumbel dependence function was

its ease of interpretation, it is certainly worth the effort to check whether other

dependence functions do a better job fitting this sample of Latin American countries. The

rationale for studying other dependency functions is as follows. Blyth (1996) and Shaw

(1997) have made the point that linear correlation cannot capture the non linear

dependence relationships that exist between many real world risk factors. They point that

the use of correlation as we commonly do is ok when risks have a multivariate normal

distribution, jointly. This is, when the underlying process is governed by elliptical

distributions, i.e., distributions whose density is constant on ellipsoids and whose contour

lines of the density surfaces are ellipses in a two dimension representation.

However, they point that the use of correlation is not ok when we are outside the

elliptical world because marginal distributions and correlation do not determine the joint

distribution. Other important facts in this non-elliptical case are that perfectly dependent

variables do not necessarily have a correlation of 1 and vice versa, and zero correlation is

not equal to independence. Finally, it is important to recall that correlation is only defined

when the variances of risks are finite.

56
 The aforementioned facts provide with the following game plan for future

research. We can, for instance, use rank correlation, i.e., the linear correlation of

probability transformed variables. It still is very limited, since the structure of

dependence is not fully described and it cannot be easily manipulated.

The use of copulas comes as a natural alternative. Copulas are a way of trying to

extract the dependence structure from the joint distribution. For instance, a joint

distribution can be written as F(x1,..., xn) = C(F1(x1),...Fn(xn)), where

Cβ (u, v ) = exp[ − {( − log u) 1/ β + ( − log v ) 1/ β } β ]

and C is a copula function can be thought of as a multivariate distribution function with

standard uniform marginal distributions. C is the dependence structure of F.

This approach has the advantage of not summarizing dependence in a single

number (which can be a dangerous simplification) and rather utilizes a model of the

dependence structure to provide with more info. More work in this area is needed and

projected as future extensions to the present work.

57
 CHAPTER 4

THE EFFECT OF SOVEREIGN CREDIT RATING CHANGES ON EMERGING

STOCK MARKETS

4.1. Introduction

In this chapter, we study the effect of sovereign credit rating changes on the cross

section of locally traded firms. Standard and Poor’s defines a credit rating as “a current

opinion of the creditworthiness of an obligor with respect to a specific financial

obligation, a specific class of financial obligations, or a specific financial program.”25

A sovereign credit rating, then, reflects the rating agency’s opinion on the ability

and willingness of sovereign governments to service their outstanding financial

obligations in full and on time – it is basically an estimate of probability of default and/or

likelihood of repayment. These sovereign ratings reflect factors such as a country's

economic status, transparency in the capital markets, levels of public and private

investment flows, foreign direct investment, foreign currency reserves, and the ability of

a country's economy to remain stable despite political change.

An important difference between sovereign credit ratings and corporate credit

ratings is that sovereign credit ratings have large effects and implications that spread to

25
http://www.standardandpoors.com/
58
other entities besides the one being rated. Cantor and Packer (1996) wrote: “Sovereign

ratings are important not only because some of the largest issuers in the international

financial markets are countries, but because these assessments affect the ratings assigned

to borrowers of the same nationality. For example, agencies seldom, if ever, assign a

rating to a (...) private company that is higher than that of the issuer's home country.”

Investors with interests in foreign firms pay attention not only the foreign firm’s credit

rating –when it is available- but also to the credit rating of the country where the firm is

domiciled.

This chapter contributes to the existing literature by extending our understanding

of how much information do sovereign rating changes convey to domestic stock markets.

Specifically, we investigate if and why a country rating matters for firms within that

country. We show that sovereign rating changes affect the terms on which a domestic

firm can get credit, creating an exogenous change in the cost of capital. We divide our

results in two: we first present the effect of rating changes at the aggregate level using

national stock indices, and then proceed to study the effect of those sovereign rating

changes on the individual firms located within those countries. Our index level results are

consistent with the extant literature on the effect of credit rating changes on U.S. firms.

We do find evidence of a significant negative stock price reaction to sovereign rating

downgrades while we find no evidence of a stock price reaction to sovereign rating

upgrades. Further, we document that local stock markets react only to news of sovereign

rating downgrades issued by Standard & Poor’s.

When we look at the effect of sovereign rating changes on the cross section of

individual firms, our results suggest that sovereign credit rating changes affect larger

59
firms more. We also find that firms in poorer emerging countries experience larger drops

in the price of their shares. The fact that ratings affect asset prices is surprising based on

Wakeman (1992). In that paper, he explained that the real function of bond rating

agencies is to reduce costs of financing at issuance. According to him, a bond rating does

not determine but rather mirrors the market’s assessment of a bond’s risk, and thus

ratings should not affect but rather reflect the market’s estimation of a bond’s value.

Therefore, there should be no impact of rating changes on stock prices.

Empirical work in this area (e. g., Holthausen and Leftwich, 1986, Cornell et al

1989) shows, however, that credit ratings convey valuable information to the bond and

stock markets. The idea is that firms disclose more information to rating agencies, which

in turn incorporate this information in their ratings in such way that the privileged

information is conveyed but not disclosed. In an international setting, one could argue

that rating agencies, in conducting their research to rate a country, develop information

that would also affect the future prospects of domestic firms. Cornell et al (1989) proved

that U.S. firms with more intangibles -which are more difficult to value- are affected

more by the information conveyed by rating changes. More recently, Jorion et.al (2004)

tested the effects of the introduction of Reg FD (Regulation Fair Disclosure) on the

informational advantage of rating agencies. Using standard methodology, they show that

both credit rating downgrades and upgrades have larger effects on stock prices after the

introduction of Reg FD.

There are at least three reasons why stocks in foreign markets react to credit

changes of their sovereign government. Although sovereign credit ratings are not country

ratings, it is more often than not the case that the rating assigned to a non-sovereign

60
foreign entity is the same or lower than the credit rating assigned to the sovereign where

it is domiciled. This is the so-called ‘sovereign ceiling’. The concept is based on the

assumption that a sovereign default will force domestic issuers also to default because

most circumstances leading to national debt crises –e.g. balance of payment crises and

terms of trade shocks- directly affect the debt servicing capacity of private borrowers. For

instance, private firms might be forced to default in the payment of their international

obligations if their government imposes exchange controls that prevent access to foreign

currency. All these additional risks exogenous to the firm are the basis for the existence

of a cap on the ratings assigned to foreign firms, namely the sovereign ceiling.26

The sovereign ceiling is relevant for firms when it is binding, because in that case

firms are facing costs of capital artificially higher than those they should face, i.e., a

binding sovereign ceiling means the firm is paying a higher yield on its debt than it

otherwise should. When a foreign firm has stronger credit characteristics than the

sovereign where it is located and when the risk of the imposition of debt-service-limiting

foreign exchange controls is less than the risk of the sovereign defaulting, it could be

possible to observe foreign firm credit ratings higher than those assigned to the sovereign

– this is known among practitioners as breaking the sovereign ceiling.27

26
A common misconception is that the sovereign rating and the sovereign ceiling are synonyms. There are
examples of countries – for instance, some east European countries- where monetary ties make it harder for
the country to impose defaulting conditions on its firms. In those cases, it is not uncommon to observe a
sovereign rating lower than the sovereign ceiling. In many countries, however, the sovereign rating remains
the best proxy to the sovereign ceiling (Durbin and Ng, 2004). The idea is based on the notion that because
firms operate under the economic framework promoted by the government, an economic downturn would
reflect itself in the domestic firm’s ability to repay their financial obligations abroad. For instance, in many
emerging countries, economic crisis were almost always associated with currency devaluations, making
foreign obligations more difficult to meet. The IMF (1991) defined this risk as ‘transfer risk’, since a
government can transfer its problems –through greater taxes, imposing currency controls or assets
expropriation- to an otherwise completely healthy firm.
27
The sovereign ceiling has not always been binding. The most notable exception is the Argentinean case,
where Standard & Poor’s allowed 14 firms in April 1997 to have higher debt ratings than that of the
61
 The second reason why local stock markets might react to sovereign credit rating

changes has to do with the difficulties in collecting reliable information for most foreign

firms. Many foreign countries –emerging markets mostly- have legislations that are not

as conducive to the free flow of information from firms to its investors –actual and

potential- as the U.S. legal framework. Faced with such difficulties to obtain detailed

firm-level data, investors tend to rely on sovereign ratings as convenient and intuitive

aids in valuing projects in emerging countries where firm-level information is scarce,

effectively ‘painting all issuers with the same brush’. This effect seems to be even

stronger for below investment-grade than for investment-grade issuers (Cantor and

Packer, 1996).

Finally, a third reason is that sovereign credit rating changes communicate

information about the country and firms depend in many ways on the country where they

are located. Large domestic firms in these countries can typically access capital more

cheaply abroad than at home. However, a sovereign credit downgrade reduces this

advantage. In that case, the firms that can borrow abroad would be less likely to do so,

which would lead them to borrow more from at home and crowd out smaller firms.

Alternatively, large domestic firms could find it impossible to raise suitable amounts at

home because domestic financial markets are not deep enough. In this case, if foreign

borrowing becomes more expensive (i.e., higher costs of capital), firms may end up

borrowing less, or not borrowing at all and not undertaking investments.

Republic of Argentina. At that time, Moody’s severely criticized the move, calling it irresponsible, and
refused to follow suit. Interestingly, when Argentina defaulted in 2001, all these Argentinean firms
defaulted too.
62
 To conduct this analysis we collected all sovereign rating changes issued by

Standard and Poor’s (S&P) and Moody’s on 29 emerging countries from 1986 until 2003.

We study the stock price reaction to 136 downgrades (81 from S&P and 55 from

Moody’s) and 100 upgrades (57 and 43 from S&P and Moody’s respectively). We also

analyze the impact of the first rating change to a given event, e.g., the first rating change

from either agency to the Mexican crisis of 1994 or to the Russian crisis of 1998.

This chapter is the first to study the effect of sovereign credit changes on the stock

prices of domestic firms. To do this, we collect information on 1281 individual firms

located in 29 emerging countries. After computing abnormal returns for each firm, we

run cross-sectional regressions of those abnormal returns on firm-specific characteristics

and country-specific variables. We document how the size and wealth of the country

where a firm is domiciled are related to the extent to which that firm will be affected by a

sovereign rating change. More importantly, we find that previous access to international

capital markets is an important determinant of the extent to which a firm is affected by a

sovereign credit rating change.

Our results are important for several reasons. Recent renowned bankruptcies –

Enron, World Com, United Airlines- have raised legitimate questions about the value we

should place on credit ratings.28 Further, recently there has been a concern that rating

agencies are late in issuing sovereign rating changes. Some authors claim that by lagging,

rating agencies contributed to boom-bust cycles. If that is the case, we should not capture

much of an effect in our empirical exercise. Yet, we find evidence of an effect that could

have been –arguably- anticipated. We interpret our results as evidence of the

28
A good reference to illustrate the failure of credit ratings in capturing deteriorating conditions of the
firms mentioned in the text is: http://www.moodyskmv.com/research/UAL.html
63
informational content of sovereign rating changes, even when they are not fully

unanticipated events. Also, regulators -foreign and domestic- that conduct assessments of

risk can benefit from an improved understanding of sovereign ratings changes and their

effects on many of the firms they monitor.

Our results also contribute to the research pioneered by Morck, Yeung and Yu

(2000). In their work, they show that markets do not do a good job of differentiating

among firms within emerging markets. Specifically, they find that “stock prices in

economies with high per capita gross domestic product (GDP) move in a relatively

unsynchronized manner. In contrast, stock prices in low per capita GDP economies tend

to move up or down together.” In this chapter, we ask whether the price reaction of stocks

to sovereign rating changes in 29 countries is related to firm characteristics and/or

country characteristics. We do find that larger firms experience worse stock price

reactions following sovereign downgrades. Firms that have accessed international capital

markets also experience more negative reactions to downgrades of their sovereign

government. Finally, we also find that firms located in richer countries (GDP per capita)

and in countries with more developed financial markets (stock market capitalization to

GDP) experience smaller price reductions following sovereign downgrades.

This chapter will proceed as follows. In section II we will review the existing

literature. Section III will show that there is an impact on the stock market at the index

level. We proceed to analyze the impact of sovereign rating changes on individual firms

in section IV. Section V will conclude.

64
4.2. Literature review

Earlier literature in the 1970s and all through the 1990s focused on the study of

the impact of rating changes of individual U.S. firms on their bond and stock prices.

Griffin and Sanvicente (1982), using monthly data, were the first ones to find evidence of

significant negative stock price reactions to rating downgrades while finding no evidence

of significant reactions to rating upgrades. Holthausen and Leftwich (1986) is the first

paper that used daily data to study the stock price reaction to credit rating changes. The

main contribution of their work was to establish that rating downgrades by Moody’s and

Standard & Poor’s provide information to the markets and impose costs to the firm by

reducing the stock’s price. They did not find a significant stock price reaction for rating

upgrades. Subsequently, many papers confirmed the findings by Holthausen and

Leftwich (1986) under many different specifications and conditions (Wansley and

Clauretie (1985), Cornell, Landsman and Shapiro (1989), Hand, Holthausen and Leftwich

(1992), Goh and Ederington (1993), Goh and Ederington (1999), Dichev and Piotroski

(2001)).

Cornell, Landsman and Shapiro (1989) test whether a stock’s price response to

rating changes is related to the nature of the firm’s assets, in particular whether the stock

price response is related to the firm’s net intangible assets. They propose two hypotheses

for the existence of such a relation, both of them based on the assumption that intangible

assets are more difficult to value than tangible assets. Rating agencies, in conducting their

business, develop expertise in the valuation of intangibles which in turn gives them an

edge when it comes to firm and cash flow valuation. Their first hypothesis (investor

hypothesis) assumes that a rating change should be more informative for firms with

65
relatively large proportions of intangible assets. A second hypothesis is the non-investor

stakeholder hypothesis. Here, stakeholders in the firm –clients, suppliers, and employees-

have implicit claims that are revalued at the arrival of news on credit rating changes. The

revaluation of these implicit claims has immediate effects on a firm’s cash flows.

Although rating changes are usually anticipated, they are not fully discounted by the

market because there is uncertainty about the exact timing of the announcement. Both

hypotheses imply that the impact of new information about rating changes on a firm’s

stock price is likely to depend on the firm’s intangible assets. They do find supporting

evidence for their hypotheses when analyzing a sample credit rating changes of U.S.

firms.

Goh and Ederington (1993) looked further into the reaction to rating downgrades

by analyzing if a bond rating downgrade conveys good or bad news for stockholders.

They divided the sample of rating downgrades in those that are mechanically triggered by

increases in leverage and those that are caused by weaker prospects for the firm,

expecting to find negative stock price reactions stemming from the latter reason and

positive or no stock reactions to downgrades prompted by the former reason. Consistent

with their predictions, they find that rating deteriorations due to worse future prospects

cause a negative stock price reaction, whereas the stock price did not react on average to

credit rating changes due to changes in leverage. Hull, Pedrescu and White (2004)

analyze the relation between credit default swap spreads and credit ratings. They find that

only reviews for downgrades convey information to default swaps, confirming that

downgrades convey information to investors and upgrades do not.

66
 Although most studies consider a rating downgrade to be bad news, not all

authors agree that downgrades are bad news for shareholders. An argument initially put

forth by Holthausen and Leftwich (1986) and Zaima and McCarthy (1988) explains why

we might observe positive stock price reactions to rating downgrades. The argument is

based on a Merton-model for firm value point of view. In this case, equity holders hold

an option on the value of the firm with an exercise price equal to the par value of the

firm’s debt, and therefore an increase in the variance of the firm’s cash flows would

trigger a downgrade and redistribute wealth from bondholders to stockholders.

Goh and Ederington (1999) analyze the cross sectional variation in the stock

market reaction to bond rating changes, showing the stock markets react more negatively

to bond rating downgrades within and into junk status. They also find that downgrades

tend to follow periods of negative returns, which is interpreted as evidence of partial

predictability in ratings.

To investigate whether capital markets consider that credit ratings convey any

information beyond what is publicly available on a firm’s prospects and fundamentals,

Kliger and Sarig (2000) conduct a unique experiment. On April 26, 1982, Moody’s

modified its rating classification to include finer ratings. This unannounced change was

simultaneously implemented on all bonds followed by Moody’s, thus providing a unique

opportunity to study the informational content of credit ratings. They analyzed bond,

stock and option prices observed before and after the classification change and concluded

that rating information is valuable to capital markets.

Dichev and Piotroski (2001) examine long-run stock returns using all Moody’s

bond rating changes between 1970 and 1997. They look at both cumulative abnormal

67
returns and buy-and-hold returns, and after controlling for size and book-to-market, find

no evidence of abnormal returns following upgrades, whereas they confirm substantial

negative abnormal returns following downgrades. Further, they document poorer returns

for downgrades of small and low-credit-quality firms. They also find evidence that

suggests a role of downgrades as predictors of future deteriorations in earnings.

More recently, Odders-White and Ready (2003) take a different approach to

determine if credit ratings contain information beyond the one that is publicly available.

Using panel regressions including popular measures of adverse selection, they show that

firms with more adverse selection problems have lower ratings. Further, they find a

significant negative relation between the components of the adverse selection measures

related to private information and debt ratings. This relation is interpreted as evidence

that ratings contain information beyond that available in other published financial data

and that is not captured by other variables. Finally, they also find that rating agencies

often fail to react to changes in uncertainty immediately, thus causing some predictability

in rating changes. The fact that rating changes are predictable is a source of potential

concern to market regulators.

On the sovereign side, Cantor and Packer (1996) looked at the determinants of

sovereign credit ratings using ratings issued by Moody's and S&P. After conducting a

variety of cross-sectional analyses, they concluded that ratings subsume all relevant

macroeconomic information. They also found that the announcement effect on bond

spreads is much stronger for below investment-grade (junk) than investment-grade

issuers.

68
 Elayan, Hsu and Thomas (1999) conducted a comparison of the informational

content of credit rating announcements in New Zealand and in the U.S. They document a

reaction to market credit rating announcements in small markets that is different from the

reaction in large markets; in particular they find evidence of New Zealand firms’ stock

prices reacting significantly to both good and bad news conveyed by rating upgrades and

downgrades. Kaminsky and Schmukler (2002), look into the effects of rating and

outlook changes on bond and stock returns. Their methodology is different from ours.

Further, they focus on the macroeconomic consequences of rating changes whereas we

are interested in the individual firm effects. Using a panel specification they find

evidence of rating upgrades taking place after market rallies and rating downgrades

occurring after periods of poor performance, concluding that rating agencies play a de-

stabilizing role in emerging markets.

Patro and Wald (2005) study the returns experienced by local firms following

equity liberalizations. They find that small, high book-to-market (BM) and low beta firms

experience higher returns following liberalization events, showing that stock prices of

individual firms react in different ways to country-wide events. Miller and

Puthenpurackal (2005) show that firms accessing international debt markets via global

bond offerings do experience reductions in their overall cost of capital. We conjecture in

our chapter that previous access to international capital markets could be related to the

magnitude of the stock price impact a firm experiences after a sovereign credit change.

Some papers asked whether capital markets differentiate between rating agencies.

Beaver, Shakespeare, and Soliman (2004) study the differences in ratings provided by

certified agencies like Moody’s and credible, non-certified agencies. They find that

69
ratings by certified agencies are more conservative and less timely than those issued by

non-certified rating agencies. Mollemans (2004) studies the effect of rating

announcements in Japan and concludes that Japanese stock returns are responsive to S&P

rating changes but not to Moody’s rating changes.

Rating agencies have been criticized for the role they played in currency and/or

debt crises in the 1990s. Critics claim that by lagging in their rating changes, agencies

contributed to boom-bust cycles, inducing excessive flows of funds into and out of

emerging countries. Sy (2003) asks whether rating agencies anticipate currency and/or

debt crises for the sovereigns they cover. He concludes that ratings do not predict

currency crises and react ex-post to them, consistent with rating agencies’ assertion that

they measure probability of debt default, not probability of currency crises. He also finds

that lagged ratings and rating changes are useful in anticipating sovereign distress. Amato

and Furfine (2003) study if rating agencies behave in a counter-cyclical way. They find

that ratings issued by S&P do not have excessive sensitivity to the business cycle – which

is consistent with the normative view that ratings should have a long term perspective.

The ‘through-the-cycle’ methodology S&P uses is better suited for long-term

perspectives, so that a rating agency only changes a rating when it feels confident the

changes in a company’s risk profile are likely to be permanent. Brooks, Faff, Hillier and

Hillier (2004) study the aggregate stock market impact of sovereign rating changes from

several rating agencies. They conclude that national stock markets react to news of

sovereign downgrades but not to sovereign upgrades.

Governments not only seek credit ratings to facilitate their access to international

capital markets, but also because these assessments affect the ratings of other borrowers
70
of the same nationality. As mentioned in the introduction, one way through which

sovereign ratings affect domestic firms is the sovereign ceiling. Durbin and Ng (2004)

study the instances in which the sovereign ceiling rule is broken. They match bonds from

companies based in emerging markets and match them with a corresponding sovereign

bond to measure the difference between corporate and sovereign spreads. In 20 cases (out

of a reduced sample of 28 bonds) they identify several instances in which the sovereign

ceiling rule is broken. A casual review of the characteristics of these firms suggests that

the ability to generate revenue abroad, the existence of ties with the government or an

affiliation with a foreign firm are common characteristics of these firms. Cruces (2001)

shows that sovereign credit ratings are key in determining the cost and availability of

international financing for an economy. He reasons that because there is limited ability to

enforce debt contracts subject to the regulatory authority of a foreign government, and

because governments are sovereign in their territories and have few assets beyond their

borders that can be seized by foreign court order, measures of host government sovereign

risk contain critical information for investors contemplating international investment.

4.3. The effect of sovereign rating changes on stock market indices

Following earlier literature (Kamisnky and Schmukler, 2002; Brooks, Faff, Hillier

and Hillier, 2004), we will analyze separately the impact of S&P and Moody’s rating

changes on stock indices. We will start with this section with data and methodology

descriptions.

71
4.3.1 Data

We start by collecting all information available in Bloomberg on sovereign rating

changes. There are three certified rating agencies that monitor emerging countries and

issue credit ratings for a number of them. These agencies are Fitch, Moody’s and

Standard and Poor’s (S&P). In this chapter we use ratings issued by Moody’s and S&P

because they cover more countries for a longer period of time. For all emerging countries

covered by each rating agency, we collected debt rating history for long-term, foreign

currency denominated obligations.29

To rate issuers, Standard and Poor’s uses 24 different ratings, ranging from D

(lowest) to AAA (highest), with a “+” or “-“ to denote issuers above or below the mean in

each category. Moody’s has 27 ratings, from a lowest rating of C to a highest rating of

Aaa (Moody’s adds the number 1, 2, or 3 after the rating to signal an issuer above the

mean, on the mean or below the mean in each category, respectively). Each rating agency

has a slightly different threshold for so-called ‘investment grade’ issuers. Standard and

Poor’s considers everything above a rating of BBB- (inclusive) as investment grade,

whereas Moody’s considers Baa3 (inclusive) and above as an investment grade rating.

Tables 13 and 14 summarize the information collected on sovereign rating

changes. Table 13 shows the 32 emerging countries covered by Standard and Poor’s.

Korea and Thailand were among the first countries rated (1988 and 1989, respectively),

while Venezuela was the first Latin American country rated (in 1990). Bulgaria, Jamaica

and Ukraine are the most recently added countries to their watch list, in 1998, 1999 and

2001, respectively. As expected, countries that experienced episodes of heightened

29
We use Standard and Poor’s Emerging Market Database as our criteria to decide which countries are
emerging markets. If a country is included in their list, it is included in ours.
72
financial turmoil in the 1990s are among the countries that experienced most credit rating

changes, so it is no surprise to see Indonesia (16), Korea (11), Russia (11), and Argentina

(9) with the largest number of rating changes.

It is interesting to observe the evolution of ratings in some countries. Portugal, for

instance, had a slow but steady increase in credit worthiness. Other countries, like Korea

and Venezuela, had credit ratings that were all over the place, experiencing credit

upheavals that made them get very high as well as very low credit ratings. Only 8

countries (Chile, China, Czech Republic, Greece, Malaysia, Portugal, Qatar and

Thailand) have always been rated as ‘investment grade’, whereas 10 additional countries

have enjoyed that status at some time or another.

Table 14 shows information from rating changes announced by Moody’s.

Argentina (1986), Brazil (1986), and Malaysia (1986) are among the first countries

covered. The most recent additions to their list of covered countries are Ukraine (1998),

Chile (1999) and Egypt (2001). Although the same countries are covered, there are

important differences as to how long they have been covered and also –we can only

speculate on this- about the level of attention given to them by the rating agencies. For

instance, Indonesia (with 16 rating changes from Standard and Poor’s) only has 4 rating

changes from Moody’s. Argentina, on the other side (with 9 changes from Standard and

Poor’s) has 19 rating changes from Moody’s. Argentina (19), with Malaysia (14) and

Russia (10), are the countries with the largest number of rating changes from Moody’s.

Daily stock index data was collected from Datastream for 29 countries.

Specifically, we collected the return index (datatype RI) for Datastream’s local stock

indices in U.S. dollars. Indices not available in U.S. dollars were converted from their

73
local currency into U.S. dollars using the prevailing exchange rates available also on

Datastream. For 21 countries we use Datastream’s local stock indices, whereas for 8

more countries we use their local stock exchange index. We could not find information

for the remaining 3 countries.

4.3.2 Methodology

In conducting our study, we run into a typical problem when interpreting

announcement effects: in efficient markets, the announcement effect of the event we want

to study will measure the difference between the post-announcement effect and what was

expected beforehand. If investors had a high likelihood beforehand of an announcement

occurring, the updating element is small and the announcement effect underestimates the

impact of the event. In the case of debt ratings, rating agencies are famous for being

secretive about their decision process. Therefore, although rating changes are sometimes

anticipated, there is uncertainty about when, if at all, a rating change will occur (Cornell,

Landsman and Shapiro, 1989).

We follow the standard event study methodology by Brown and Warner (1985).

The methodology developed in that paper has been successfully applied to a wide variety

of events, for instance mergers and acquisitions and the cross listings of shares in new

markets. A common denominator of those applications is the fact that the event being the

focus of the study is rarely a ‘sudden’ occurrence. Usually, news about a merger or a

cross listing of shares is leaked or is even publicly announced prior to them taking place.

In this chapter, we are interested in the stock price reaction when news of a sovereign

rating change is made public.

74
 Event studies are by definition joint tests of hypotheses. To be able to measure

abnormal returns, one has to define what a normal return is, i.e. make an assumption on

the return-generating process. More often than not, the market model is used to compute

normal returns.30 The null of event studies is that there should be no significant abnormal

average returns if the event is uncorrelated with the stock return. We will compute

abnormal returns according to the following specification:

Ai ,t = Ri ,t − αˆ i − βˆi Rm,t

where α̂ i and βˆi are the OLS parameters obtained from the (t-244, t-6) estimation

window, Ai ,t is the abnormal return for firm i, Ri ,t is the return for firm i, and Rm ,t is the

world market return from Datastream. We then compute the significance of the average

abnormal return for each date in the estimation window using a test statistic. The statistic

is computed as the ratio of the mean abnormal return to the estimated standard deviation

from the time series of mean abnormal returns.

4.3.3 Discussion of index level results

Tables 15 and 16 show the results from event studies using rating changes from

Standard and Poor’s and Moody’s, respectively. Using S&P data, we have a total of 160

rating changes, out of which 81 are rating downgrades, 57 are upgrades and 22 are initial

30
Other measures of normal returns used in the literature are mean adjusted returns (i.e. using the simple
average of a security’s daily return in some pre-defined estimation window) and market adjusted returns
(where the return of the market is subtracted from each individual firm return). Brown and Warner (1985)
showed that the OLS market model is well-specified in most cases and that it outperforms the other two
return-generating processes when some assumptions (normality, autocorrelation, etc.) are relaxed. Using
250 simulations of 50 randomly chosen stocks with daily data Brown and Warner (1985) showed that
methodologies based on the OLS-market model are well-specified.
75
rating announcements.31 Table 15, panel A has the results of the event study conducted

using S&P downgrades. There seems to be evidence of news leaking to the market a few

days before the actual announcement, which is consistent with anecdotical evidence as

well as with information collected through informal conversations with practitioners.

Rating changes usually don’t strike like bolts of lightning, but rather many times they are

‘telegraphed’ to the market days in advance. This could explain the fact that abnormal

returns for t-3, t-2, t-1 and 0 are statistically significant. Also, cumulative abnormal

returns (CAR) for the (-5, +5) and (-1,+1) are negative and statistically significant at the

1% level. The 11-day CAR for rating downgrades is -2.8% (t-stat of -2.37), while the 3-

day CAR is -1.8% (t-stat of -2.95). Figure 17 illustrates a negative trend of both the daily

abnormal return (bars) and the cumulative abnormal return (line) starting at t-5.

Panel B of Table 15 has the results from the analysis of rating upgrades issued by

S&P. Consistent with earlier domestic literature we find no evidence of a significant

stock price reaction to rating upgrades. No single abnormal return in the (-5, +5) period is

significant, and the visual inspection of figure 18 –which displays abnormal returns and

cumulative abnormal returns- provides no further clues.

Table 16 presents the results of the analysis using Moody’s rating changes.

Moody’s has 53 downgrades, 45 upgrades and 19 initial rating announcements for the

countries in our sample. Panel A has the results for Moody’s rating downgrades.

Although the abnormal returns at t-4 and t-1 are negative and significant at the 1% and

5% level, the 11-day and 3-day cumulative abnormal returns are statistically

insignificant. These results seem to suggest that, for sovereigns, Moody’s ratings are
31
The reason we have only 22 initial rating announcements is that not all countries had index data
available at the time their initial sovereign credit rating was announced.
76
considered as less informative by the local stock markets.32 Panel B of Table 16 shows

the results of rating upgrades. Just as it was the case with Standard and Poor’s upgrades,

we find no evidence of significant abnormal returns.

Ours is not the first paper to find that outside the U.S. Standard and Poor’s ratings

are considered more informative than ratings issued by Moody’s. Mollemans (2004) finds

a similar result looking at Japanese firms, and Beaver, Shakespeare, and Soliman (2004)

document the different stock price reactions to ratings issued from certified and non-

certified rating agencies. Hu, Kiesel and Perraudin (2001) opted for S&P ratings over

Moody’s to estimate transition matrices for sovereign ratings because they were more

informative, although they noted that of a sample of 49 sovereigns rated by both agencies

at the time they wrote their paper, 28 had the same rating, 14 were apart by only one

notch and 7 were apart by two notches.

So far, we have looked separately at rating changes by agency, in a manner

consistent with the extant literature. Now we try two different approaches. One is to use

the initial rating announcement by either agency of a credit rating for a country, i.e., the

first time a country is rated. By definition, this is a good news event. Non-rated countries

–or investment bankers acting on their behalf- initiate contacts with rating agencies when

they want to access international capital markets for the first time, when they want to

improve the terms under which they access international capital markets or to attract

foreign investors into its local debt or equity markets. Countries would hire a rating
32
One difference that becomes evident from looking at the breakdown of rating change announcements is
that Moody’s makes more ‘outlook’ change announcements than Standard and Poor’s. An outlook change
is a warning issued by the rating agency, usually stating that a country’s rating is under review or under
stress. Although not all warnings materialize in actual rating changes (only about 60% of them do), and not
all changes are preceded by warnings, one possibility we had to consider is that stock prices react to
outlook announcements and not to actual rating changes. We re-run our analysis using outlook
announcements dates and still found no significant abnormal returns for Moody’s data.
77
agency when their prospects are good, and therefore when they have reasonable high

expectations of being rated favorably. 33 Viewed in this light, initial ratings are usually

good news, and our expectation is to find favorable stock price reactions to the

announcement of an initial rating, or at least stock price reactions similar to those

observed for rating upgrades. Table 17 has the results for this experiment. We have 17

countries with data going back far enough to cover the date of the initial rating. Of those

17 countries, 10 had initial ratings by S&P and 10 by Moody’s. We do not find any

evidence of a significant reaction across countries to these events, just as it was the case

with sovereign rating upgrades.

The second approach is to look at the impact of the first rating change issued by

either agency to a given event. This is not straightforward, as Moody’s and S&P rating

changes do not move in tandem. For instance, in the 1990-2003 period Moody’s had 19

rating changes for Argentina while S&P had only 9. We proceeded by defining pairs of

events of any two rating changes by both agencies that took place within a six month

period. The earliest announcement within each pair is what we consider as first rating.

Table 18 has the complete list of paired rating changes for Argentina.34 It shows which

agency is identified with providing –chronologically- the first rating if both agencies

changed a country’s rating within a six month period. We did this pairing of ratings for

each one of the 29 countries in the sample, and ended up with 40 downgrades (21 were

from S&P and 19 from Moody’s) and with 38 upgrades (21 from Moody’s and 38 from

S&P).

33
See Martell and Stulz (2003).
34
Tables for all the countries included in this paper are omitted for brevity and available upon request.
78
 Table 19 has the results of this analysis of first ratings. As with the complete

sample of rating changes, we only find significant results for rating downgrades. Panel A

has the results from analyzing sovereign downgrades by S&P (21) and by Moody’s (19).

The difference in the stock reaction to ratings issued by each agency is evident.

Downgrades issued by S&P are associated with a negative 11-day abnormal cumulative

return of 6%, significant at the 1% level. The 3-day CAR is also negative and significant.

No CAR computed from announcements by Moody’s is significant, and what’s more, the

11-day CAR is positive (although not significant). Overall, Table 19 provides evidence

that financial markets react only to sovereign credit downgrades issued by Standard and

Poor’s.

Summarizing our index level results, we find evidence of emerging stock indices

reacting to news of sovereign rating downgrades, i.e., we find significant 3-day and 11-

day negative cumulative abnormal returns following a sovereign rating downgrade. We

find no evidence of a significant stock market reaction to the news of sovereign upgrades.

We also find evidence of local stock markets reacting more to sovereign rating changes

issued by S&P than to those issued by Moody’s.

4.4. Impact of sovereign rating changes at the firm level

In the previous section we looked at the effect of sovereign rating changes on

domestic stock indices. This analysis, however, leaves several important questions

unanswered. For instance, does a change in a sovereign rating affect the terms on which a

domestic firm can get credit? If so, does it affect all firms in the same way? This begets

the question of why does the country rating matter for domestic firms. If a sovereign

79
rating change causes an exogenous change in the cost of capital, it is important to

understand whether that exogenous change is the same for all firms and whether firm

characteristics are related to the magnitude of the stock price impact of that exogenous

shock.

One possibility is that all firms are affected in the same way following a sovereign

credit change. For instance, a sovereign credit rating downgrade leads to an increase in

sovereign yield that can signal that the country as a whole is a riskier place, e.g. it is a

place with higher probability of expropriation, and/or can also signal an increased level of

market segmentation, all of which would mean that the expected return on all local stocks

should increase. This would be consistent with the work by Morck, Yeung and Yu

(2000). They showed that stock prices moved together more in emerging countries than

in richer countries and conclude that political events in low-income countries can create

market-wide stock price swings, mainly because poorer economies offer less

diversification opportunities to their firms. A sovereign downgrade is by definition a

worsening of rating that increases political risk, and can be seen domestically as a

political event with market-wide stock price reactions.

The first step is to investigate if prior access to international capital markets does

affect firms’ stock reactions to sovereign rating changes. We split our sample according

to whether a firm has an ADR trading abroad, an Eurobond issue, or both. Table 20

presents abnormal returns from sovereign downgrades (panel A) and upgrades (panel B).

As expected, all the action takes place in panel A. Oddly, CARs from firms with and

without ADR issues are nearly identical. CARs for firms with international debt,

however, are much more negative than those of firms with no international debt. This

80
suggests that firms that already have an international debt issue will probably need to tap

international markets again, either to refinance or to get new resources. However, a

sovereign downgrade will induce an exogenous worsening of the terms under which these

firms got their international credit. Therefore, these firms are more sensitive to sovereign

downgrades.

To investigate further which firm characteristics are associated with the observed

behavior of stock prices after sovereign credit changes, we will compute individual

abnormal returns and then estimate cross-sectional regressions of those returns on a group

of variables that capture firm and country characteristics. We will compute abnormal

returns for each firm included in Datastream’s stock indices using a simple CAPM

framework (i.e., market model). Our firm accounting data comes from Worldscope. For

each firm, we collected intangible assets (item 02649), total assets (item 02999), total

debt (item 03255), total revenues (item 07240), and cash plus short term investments

(item 02001). We also collected other items like foreign income as a percentage of total

income (item 08741) and total interest expense (item 01075), but decided not to use them

as their reduced availability impacted severely on our sample size. Following Morck et al

(2000) we collected GDP per capita and stock market capitalization of public firms as a

percentage of GDP from the World Bank’s Economic Indicators database. All variables

were converted to U.S. dollars using prevailing exchange rates collected from

Datastream. Cumulative abnormal returns (CARs) were computed in the same way as in

the previous section, using market model adjusted returns. We report z-statistics instead

of t-statistics to account for the varying number of firms in our computations.

81
 We also collected data on firm’s access to international equity and debt markets,

gathering information from Citibank and The Bank of New York on ADRs (all levels)

that traded in the U.S. from 1990 to 2003. Out of 385 firms with ADRs in the countries

included in this chapter, 156 are included in our sample. We also got data from SDC and

Datastream on all live and dead non-sovereign bonds from countries in our sample and

manually matched them. We ended up with 228 firms that had at least one bond trading

in international markets at the time of the sovereign rating change. As expected because

regulatory requirements are less stringent for debt issuance than for equity raising, we

have more firms accessing international debt markets than international equity markets.

Our reasoning for collecting information on access to international capital markets

goes as follows. Healthy firms might choose to go abroad to get financing because it

might be cheaper to do so or because their financial needs are not fulfilled in domestic

financial markets. When the government of the country where a firm is domiciled is

downgraded, these firms suddenly find –everything else equal- that the terms on which

they access international markets deteriorate. This will in turn raise the cost of capital for

new projects, making some of them unprofitable and therefore reducing future cash

flows, all of which translates into lower stock prices. Since these firms came to rely on

international financing on a regular basis, we hypothesize that firms that have accessed

international capital markets in the past –either equity markets via an ADR program or

debt via an international bond offering- would experience larger negative stock price

reactions following a sovereign downgrade and larger positive stock price reactions

following a sovereign upgrade.

82
 Table 21 shows results obtained from regressing individual firm cumulative

abnormal returns following sovereign downgrades on several firm and country variables.

These abnormal returns were computed in the following way:

Ai ,t = Ri ,t − αˆ i − βˆi Rm,t

where α̂ i and βˆi are the OLS parameters obtained from the estimation window,

Ai ,t is the abnormal return for firm i, Ri ,t is the return for firm i, and Rm ,t is the world

market return from Datastream, all of them denominated in dollars. We include country

dummies to control for country fixed effects, although for brevity we do not report the
35
results of each intercept. We are using the world market return to make these returns

comparable to those presented in the index-level discussion of the previous section.

Panels A and B show 3 and 11-day dollar CARs, respectively. Availability of firm

accounting data on Worldscope is of some concern, since our sample got reduced from

2,523 observations to 1,487 observations in the omnibus regressions. R-squared values

are much better than the ones obtained by previous research, although most of the

explanatory power comes from the fixed country effects specification. In panel A, the

minimum R-square value for 3-day abnormal returns is 24.7% and the max R-square

value is 28.21%, whereas in Panel B (11-day CARs) the min R-square is 15.29% and the

max is 21.72%. Both panels have a qualitatively similar story, although since results are

stronger in panel A, we will focus our discussion of results on that panel.

We started by estimating regressions on country dummies (this is equivalent of

running a fixed effects regression with country level effects) and a constant. The second

35
Detailed regression results are available upon request.
83
regression specification on panel A includes dummies to control for the existence of

ADRs and/or international bonds, and we find that only the latter has a negative

significant coefficient. This result, however, reverses strongly in all other regression

specifications when we include size proxies, log of GDP per capita and the country’s

stock market capitalization to GDP. Not only has the ADR dummy a significant and

negative coefficient, but its interaction with the log of total assets also has a significant

coefficient (positive). In all cases, the log of GDP per capita and the market capitalization

to GDP are significant at the 1% level and positive. Two proxies for size, the log of total

assets and revenues to total assets have negative coefficients, significant at least at the

10% level. Interestingly, the measure of intangible assets as proportion of total assets

never has a significant coefficient, although the negative sign is consistent with Cornell et

al (1989).

Even though we conducted regressions controlling for fixed country effects, the

log of GDP per capita and the proportion of stock market capitalization to GDP always

have positive and significant coefficients. These results suggest that firms located in

richer countries (as measured by GDP per capita) experience smaller stock price drops

following a downgrade of their government’s debt. The richer a country is, the more

diversified we expect its economy to be and therefore there will be a richer variety of

economic activities in which domestic firms can engage, making firms less sensitive to

systematic shocks (Morck et al, 2000). Also, firms located in countries with more

developed financial markets (measured by stock market capitalization to GDP)

experience smaller price reductions although the coefficients associated with this variable

84
are at least two orders of magnitude smaller than the coefficients associated with the log

of GDP.

The two variables we use to proxy for firm size, log of total assets and revenues

over total assets, are consistent in the story they tell. Both have negative and significant

coefficients in every specification in panel A. It is sensible to think that only larger firms

will outgrow local financial markets to the point that they need to go abroad to get the

needed financing, making them more sensitive to exogenous shocks to the cost of

international financing.

Table 22 present the results of similar regressions as those shown in Table 21

using CARs following sovereign credit upgrades. Although the index-level results did not

produce significant abnormal returns, this does not mean that there is no information in

the cross section. As with Table 21, panel A of Table 22 has the results from regressions

that use 3-day CARs as the dependent variable, while panel B uses 11-day CARs.

Although both panels have the same qualitative picture, panel B has a sharper picture and

our description shall focus on it. Interestingly, the coefficients for log of GDP per capita

and stock market capitalization to GDP have the opposite sign as before (now it is

negative). This result is significant at the 1% level for both variables. Both size proxies

have insignificant coefficients in the omnibus regressions. The proportion of inventories

to assets has a significant positive coefficient, as well as the proportion of total debt to

total assets. Most importantly, having had any type of access to international capital

markets makes no difference in the abnormal return experienced by these firms.

Overall, the picture that emerges from Tables 21 and 22 is one consistent with

sovereign credit changes having an asymmetric impact on the stock prices of domestic

85
firms. Larger firms experience larger reductions in their stock price following sovereign

downgrades yet they do not experience larger increases when a sovereign upgrade is

announced. We think this reflects the exogenous cost imposed on the firm by a sovereign

downgrade, especially in the case where the sovereign ceiling is binding. If a firm’s

rating is constrained by the sovereign rating, a drop in sovereign rating will force an

increase in the cost of international debt for that firm. An increase in sovereign rating, on

the other hand, does not reduce the cost of international debt. Larger firms are more

probable to experience this behaviour.

Firms with more debt as a proportion of total assets benefit more from a sovereign

upgrade yet their stock prices are not punished as much when their sovereign government

is downgraded. Firms located in richer countries and in countries with more developed

financial markets experience smaller losses in their stock prices following a downgrade

but also experience smaller gains following upgrades of their sovereign government.

Finally, having accessed international equity and/or debt markets makes a firm more

vulnerable to sovereign downgrades but does not grant any additional benefits when the

host government is upgraded. This effect is more important for larger firms, as evidenced

by the significant coefficient for the interaction of ADRs and size.

4.5. Conclusions

In this chapter we analyzed the impact of sovereign credit rating changes on local

stock markets. In the first part of the chapter, we established that local stock indices from

29 emerging countries react only to news of sovereign rating downgrades, in a manner

consistent with the results of existing domestic literature. We also established that

86
international markets consider rating changes from S&P more informative than rating

changes issued by Moody’s.

We then looked at the effect of sovereign rating changes on individual firm stock

prices. Analyzing the cross section of abnormal returns of 1281 firms, we found that

firms located in richer countries and in countries with more developed financial markets

experience smaller stock price reductions following a downgrade of their host

government. Our evidence also suggests that larger firms are more sensitive to sovereign

downgrades, as well as firms that have accessed international equity or debt markets.

Our results are relevant for domestic and foreign investors, firms located in

emerging countries and regulators. Local regulators can better assess the risks faced by

the firms they are supposed to regulate. Further, a better understanding of how these

country-wide shocks affect local firms can help them engage in more efficient ways to

lower their cost capital – both at home and abroad.

87
 CHAPTER 5

CONCLUSIONS

This dissertation contributes to three literatures in finance. The first essay sheds

light on the determinants of debt yield spreads for countries and firms. This essay

identified the existence of a two strong common components, unrelated to credit risk and

distinct for each type of debt, in credit spreads of non-U.S. sovereign bonds and domestic

U.S. corporate bonds. Using a vector autoregressive (VAR) model, I find that domestic

spreads are related to the lagged first common component of sovereign spreads. While

there is no contemporaneous common component in bond spreads, there seems to be a

common component when focusing on the dynamics of these spreads. Traditional macro-

liquidity variables are related to the common components found in domestic and

sovereign spread changes, raising the explanatory power of the VAR equations from a

minimum R-square value of 5% to a maximum value of 67%. Flows in and out equity

and bond funds explain more of the variation in the common components than net

borrowed reserves from the Federal Reserve.

The first essay contributes to the literature on the dynamics of credit spreads by

showing that current structural models of debt spreads can be improved if these findings

are incorporated in them. To the extent that investors depend on these models to hedge

88
the credit risk of their bond positions, they can benefit from a better understanding of the

determinants of credit spreads changes. This essay also shows that, after taking into

account the dynamics of the common components in credit spreads across different debt

instruments, the cost of debt for firms and countries depends to some extent on shocks

that affect all types of debt.

The second essay applied multivariate, extreme-value techniques to returns on

Latin American local stock indices, the S&P 500 and Latin American ADRs. I find

evidence of asymmetric behavior in the left and right tails of the joint marginal extreme

distributions for six Latin American countries. I also identified differences in extreme

correlations for different instruments (investing in ADRs vs. investing directly in the

local stock markets) where no difference was to be expected. Finally, this essay

documents evidence of a structural change in the correlations for the Mexican case before

and after the 1995 Mexican crisis.

The third and final dissertation essay analyzed the impact of sovereign credit

rating changes on local stock markets. I first established that local stock indices from 29

emerging countries react only to news of sovereign rating downgrades, in a manner

consistent with the results of the existing literature on U.S. ratings changes. I also

established that international markets consider rating changes from S&P more

informative than rating changes issued by Moody’s.

When I look at the effect of sovereign rating changes on individual firm stock

prices, interesting patterns arise. I document an average price drop of 8% for the firms in

my sample following a sovereign downgrade. This drop is more pronounced for firms

that have issued debt abroad (negative 12%). When I analyze the cross-section of

89
abnormal returns to sovereign upgrades and downgrades, I find that firms located in

richer countries and in countries with more developed financial markets experience

smaller stock price reductions following a downgrade of their host government. Also,

larger firms, as well as firms that have accessed international equity or debt markets, are

more sensitive to sovereign downgrades,. This finding suggests that larger firms – which

are more likely to find themselves financially constrained in their domestic financial

markets - rely more on external sources of financing. The cost of accessing these external

sources of financing is exogenously increased for these firms when news of a sovereign

downgrade reaches the markets. This effect is even more pronounced for firms that have

previously enjoyed access to international capital markets, as they tend to rely more in

this type of financing.

90
 BIBLIOGRAPHY

Amato, Jeffery D. and Craig Furfine, 2003, Are credit ratings procyclical?, BIS working
papers, no. 129, Bank for International Settlements Monetary and Economic Dept.
Anderson, Ron and Suresh Sundaresan, 1996, Design and Valuation of Debt Contracts,
Review of Financial Studies, Vol. 9, No. 1, pp. 37-68.
Bae, K., G.A. Karolyi and R.M Stulz, 2003, A new approach to measuring financial
contagion, Review of Financial Studies, Vol. 16(3), Fall 2003, pp. 717-764.
Beaver, William H., Shakespeare, Catherine and Mark T. Soliman, 2004, Differential
Properties in the Ratings of Certified vs. Non-Certified Bond Rating Agencies,
unpublished working paper, Stanford University.
Black, Fisher and John C. Cox, 1976, Valuing Corporate Securities: Some Effects of
Bond Indenture Provisions, Journal of Finance, Vol. 31, No. 2, pp. 351-367.
Bodie, Zvi, Alex Kane and Alan J. Marcus, 1999, Investments, Irwin-McGraw-Hill,
Fourth Edition.
Boivin Jean and Serena Ng, 2003, Are More Data Always Better for Factor Analysis?,
Working paper, Columbia University.
Brännäs, Kurt, Shahiduzzaman, Quoreshi and Ola Simonsen, 2002, Extreme-Value
Characteristics in Daily Time Series of Swedish Stock Returns, unpublished
working paper, Umea University.
Brooks Robert, Faff, Robert, Hillier, David and Joseph Hillier, 2004, The National
Market Impact of Sovereign Rating Changes, Journal of Banking and Finance,
Vol. 28, pp. 233 – 250.
Brooks, C.,Clare, A.D., Dalle Molle, J.W. and G. Persand, 2005, A comparison of
extreme value theory approaches for determining value at risk, Journal of
Empirical Finance, forthcoming.
Brown, Stephen J. and Jerold B. Warner, 1985, Using daily stock returns: The case of
event studies, Journal of Financial Economics, Vol. 14, pp. 3-31
Bulow, Jeremy and Ken Rogoff, 1989, A Constant Recontracting Model of Sovereign
Debt, Journal of Political Economy, Vol. 97, No. 1, pp. 155-178.
Calvo, Guillermo, Leiderman, Leonardo, and Carmen Reinhart, 1993, Capital inflows
and the real exchange rate appreciation in Latin America: The role of external
factors, IMF Staff Papers.

91
Cantor, Richard and Frank Packer, 1996, Determinants and Impact of Sovereign Credit
Ratings, Federal Reserve Bank of New York Review, October.
Chakravarty, Sugato and Asani Sarkar, 1999, Liquidity in U.S. fixed income markets: A
comparison of the bid-ask spread in corporate, government and municipal bond
markets, Working paper, Federal Reserve Bank of New York.
Chan-Lau, Jorge, Mathieson, Donald and James Yao, 1998, Extreme Contagion in Equity
Markets, working paper 02/98, International Monetary Fund.
Chen, Long, Lesmond, David and JasonWei, 2003, Bond Liquidity Estimation and the
Liquidity Effect in Yield Spreads, Working Paper.
Chordia, Sarkar and Subrahmanyan, 2003, An Empirical Analysis of Stock and Bond
Market Liquidity, Working paper, Federal Reserve Bank of New York.
Chuhan, Punam, Claessens, Stijn and Nlandu Mamingi, 1998, Equity and bond flows to
Latin America and Asia: the role of global and country factors, Journal of
Development Economics, 55, 439-463.
Cifarelli, Giulio and Giovanna Paladino, 2002, An Empirical Analysis of the Co-
Movement among Spreads on Emerging-Market Debt, Working Paper.
Claessens, Stijn, and George Pennachi, 1996, Estimating the Likelihood of Mexican
Default from the Mexican Prices of Brady Bonds, Journal of Financial and
Quantitative Analysis, Vol. 31, No. 3, pp. 109-126.
Collin-Dufresne, Pierre, Goldstein, Robert and Spencer Martin, 2001, The Determinants
of Credit Spread Changes, The Journal of Finance, Vol. 56, No. 6, pp. 2177 –
2205.
Cornell, Bradford, Landsman, Wayne and Alan Shapiro,1989, Cross sectional regularities
in the response of stock prices to bond rating changes, Journal of Accounting,
Auditing and Finance, Vol. 4, pp. 460 – 479.
Cruces, Juan José, 2001, Statistical Properties of Sovereign Credit Ratings, unpublished
working paper, University of San Andres.
Dichev, Ilia and Joseph Piotroski, 2001,The long run stock returns following bond rating
changes, Journal of Finance, Vol. 56, pp. 173 – 203.
Dornbusch, R., Y. Park, and S. Claessens, Contagion: How it spreads and how it can be
stopped?, unpublished paper, MIT, Cambridge, Ma.
Duffie, Darell, Pedersen, Lasse and Ken Singleton, 2003, Modeling Sovereign Yield
Spreads: A Case Study of Russian Debt, Journal of Finance, Vol. 58, No. 1, pp
119 – 159.
Duffie, Darrell and Ken Singleton, 1997, An econometric model of the term structure of
interest rate swap yields, Journal of Finance, Vol. 52, pp. 1287 – 1381.
Duffie, Darrell and Ken Singleton, 1999, Modeling term structures of Defaultable Bonds,
Review of Financial Studies, Vol. 12, pp. 687-720.

92
Durbin, Erik and David Ng, 2004, The sovereign ceiling and emerging market corporate
bond spreads, Journal of International Money and Finance, forthcoming.
Eaton, Jonathan and Mark Gersovitz, 1981, Debt with Potential Repudiation: Theoretical
and Empirical Analysis , Review of Economic Studies, Vol. 48, No. 2, pp. 289-
309.
Edwards, Sebastian, 1984, LDC Foreign Borrowing and Default Risk: An Empirical
Investigation, 1976-80, American Economic Review, Vol. 74, No. 4, pp. 726-734.
Eichengreen, Barry and Ashoka Moody, 1998, What Explains Changing Spreads on
Emerging-Market Debt: Fundamentals or Market Sentiment?, NBER Working
Paper No. w6408.
Elayan, Fayez, Maris, Brian and Phillip Young, 1996, The effect of commercial paper
rating changes and credit-watch placement on common stock prices, Financial
Review, Vol. 31, pp. 149-167.
Elton, Edwin, Gruber, Martin, Agrawal, Deepak and Christopher Mann, 2001, Explaining
the Rate Spread on Corporate Bonds, Journal of Finance, Vol. 56, No. 1, pp. 247
– 277.
Embrechts P, McNeil AJ and Straumann D, 2002. “Correlation and dependence in risk
management: properties and pitfalls”, in Risk management: value at risk and
beyond, edited by Dempster M, published by Cambridge University Press,
Cambridge
Embrechts, P., C. Klueppelberg and T. Mikosch, 1997. Modelling Extremal Events,
Springer, Berlin.
Embrechts, P., L. de Haan and X. Huang, 1999. Modelling multivariate extremes,
working paper.
Eom, Young Ho, Helwege, Jean and Jay Huang, 2004, Structural Models of Corporate
Bond Pricing: An Empirical Analysis", Review of Financial Studies, Vol. 17(2),
pp. 499-544
Fisher, R.A. and Tippet L.H.C., 1928. Limiting forms of the frequency distribution of the
largest or smallest number of a sample, Proc. Cambridge Philos. Soc., 24, pp.
180-190.
Forbes K. and R. Rigobon, 1998, No contagion, only interdependence: measuring stock
market co-movements, Sloan School of Management working paper.
Frankel, J. and S. Schmukler, Crises, Contagion, and Country Funds: Effects on East
Asia and Latin America, Center for Pacific Basin Monetary and Economic
Studies WP No. PB96-04, Federal Reserve Bank of San Francisco, September
1996.
Gencay, Ramazan and Faruk Selcuk, 2001, Overnight borrowing, interest rates and
extreme value theory, unpublished working paper, University of Windsor,
Canada.

93
Gencay, Ramazan and Faruk Selcuk, 2004, Extreme value theory and Value-at-Risk:
Relative performance in emerging markets, International Journal of Forecasting,
vol. 20, pp. 287 – 303.
Gibson, Rajna and Suresh M. Sundaresan, 2001, A Model of Sovereign Borrowing and
Sovereign Yield Spreads, Research Paper EFA 0044, PaineWebber Working
Paper Series at Columbia University.
Gnedenko, B.V., 1943. Sur la distribucion limite du terme d’une serie aleatoire, Ann.
Math. , 44, pp 423-453.
Goh Jeremy and Louis Ederington, 1993, Is a bond rating downgrade good news, bad
news or no news for stockholders? Journal of Finance, Vol. 48, pp. 2001 – 2008.
Goh Jeremy and Louis Ederington, 1999, Cross-sectional variation in the stock market
reaction to bond rating changes, Quarterly Review of Economics and Finance,
Vol. 26, pp. 101-112.
Griffin, Paul and Antonio Sanvicente, 1982, Common stock returns and rating changes:
A methodological comparison, Journal of Finance, Vol. 37, pp. 103-119.
Gumbel, E.J., 1961. Multivariate extremal distributions, Bulletin de l’Institut
International de Statistiques, Session 33, Book 2, Paris.
Hand, John, Holthausen, Robert and Richard Leftwich, 1992, The Effect of Bond Rating
Agency Announcements on bond and stock prices, Journal of Finance, Vol. 47,
pp 733-752.
Hartmann, Philipp, Stefan Straetmans and Casper G de Vries, 2001. “Asset Market
Linkages in Crisis Periods”, Centre for Economic Policy Research, working paper
No. 2916.
Hartmann, Philipp, Straetmans, Stefan and Casper de Vries, 2001, Asset Market Linkages
in Crisis Periods, working paper No. 71, European Central Bank.
Harvey, Campbell and Roger Huang, 2002, The Impact of Federal Reserve Bank's Open
Market Operations, Journal of Financial Markets, Vol. 5, No. 2, pp. 223-257.
Helwege, Jean and Christopher M. Turner, 1999, The Slope of the Credit Yield Curve for
Speculative-Grade Issuers, Journal of Finance, Vol. 54, No. 5, pp. 1869 – 1884.
Hernández-Trillo, Fausto, 1995, A Model-Based Estimation of the Probability of Default
in Sovereign Loan Markets, Journal of Development Economics, Vol. 46, no. 1,
pp 163-179 .
Holthausen, Robert and Richard Leftwich, 1986, The effect of bond rating changes on
common stock prices, Journal of Financial Economics, Vol. 17, pp. 57-90.
Hotchkiss, Edith and Tavy Ronen, 1999, The informational efficiency of the corporate
bond market: An intraday analysis, Working paper, Boston College.
Houweling, Patrick, Mentink, Albert and Ton Vorst, 2002, Is Liquidity Reflected I Bond
Yields? Evidence from the Euro Corporate Bond Market, Working paper,
Erasmus University.
94
Hu, Yen-Ting, Kiesel, Rudiger and William Perraudin, 2001,The Estimation of
Transition Matrices for Sovereign Credit Ratings, working paper, Bank of
England and CEPR.
Huang Jing-Zhi and Ming Huang, 2003, How much of the Corporate-Treasury Yield
Spread is Due to Credit Risk?, Working Paper.
Huang, Jing-Zhi and Weipeng Kong, 2003, Explaining Credit Spread Changes: Some
New Evidence from Option-Adjusted Spreads of Bond Indices, Working Paper,
Penn State University.
Hull, John, Pedrescu, Mirela and Alan White, 2004, The Relationship Between Credit
Default Swap Spreads, Bond Yields, and Credit Rating Announcements, Journal
of Banking and Finance, Vol. 28, pp. 2789-2811.
Jondeau, Eric and Michael Rockinger, 2001, Testing for differences in the tails of stock
market returns, unpublished working paper, HEC, Paris, France.
Jorion, Philippe, Shi, Charles, and Zhu Liu, 2004, Informational Effects of Regulation
FD: Evidence from Ratings Agencies, Journal of Financial Economics,
forthcoming.
Joutz, Frederik and William Maxwell, 2002, Modeling the yields on noninvestment grade
bond indexes, credit risk and macroeconomic factors, International Review of
Financial Analysis, Vol. 11, pp. 345 – 374.
Kamara, Avraham, 1994, Liquidity, Taxes and Short-Term Treasury Yields, Journal of
Financial and Quantitative Analysis, Vol. 29, No. 3, pp. 403-417.
Kamin, Steven and Karsten von Kleist, 1999, The Evolution and Determinants of
Emerging Market Credit Spreads in the 1990s, Board of Governors of the Federal
Reserve System, International Finance Discussion Papers, No. 653.
Kaminsky, G.L. and S.L. Schmukler, 1999, What triggers market jitters? A Chronicle of
the Asian crisis, Journal of International Money and Finance18, 537-560.
Kaminsky, Graciela and Sergio Schmukler, 2002, Emerging Market Instability: Do
Sovereign Ratings Affect Country Risk and Stock Returns?, World Bank
Economic Review, Vol. 16, pp. 171 – 195.
Kliger, Doron and Oded Sarig, 2000, The information value of bond ratings, Journal of
Finance, Vol. 55, pp. 2879 – 2902.
Krugman, Paul and Julio Rotemberg, 1991, Speculative Attacks on Target Zones, in P.
Krugman and M. Miller, eds., Exchange Rate Target and Currency Bands, Oxford
University Press.
Krugman, Paul, 1985, International Debt Strategies in an Uncertain World, in J.
Cuddington and R. Smith, eds., The World Debt Problem, World Bank.
Krugman, Paul, 1989, Private Capital Flows to Problem Debtors, in J. Sachs, ed., The
International Debt Problem, University of Chicago Press.

95
Lando, David, 1998, On Cox Processes and Credit Risky Securities, Review of
Derivatives Research, Vol. 2, pp. 99-120.
Ledford, Anthony W. and Jonathan A. Tawn, 1996. “Statistics for near independence in
multivariate extreme values”, Biometrika, 83, 1, pp. 169-187.
Leland, Hayne E. and K. B. Toft, 1996, Optimal capital structure, endogenous
bankruptcy, and the term structure of credit spreads, Journal of Finance, Vol. 51,
No. 3, pp. 987-1019.
Leland, Hayne E., 1994, Risky Debt, Bond Covenants and Optimal Capital Structure,
Journal of Finance, Vol. 49, No. 4, pp. 1213-1252.
Litterman, Robert and Jose Scheinkman, 1991, Common factors affecting bond returns,
Journal of Fixed Income, Vol. 1, pp. 54-61.
Longin, F. and B. Solnik, 1995, Is the correlation in international equity returns constant:
1960-1990?, Journal of International Money and Finance 14, 3-26.
Longin, F. and B. Solnik, 2001. “Extreme Correlation of International Equity Markets”,
Journal of Finance, 56, No. 2, pp 649-676.
Longin, F., 1996. The asymptotic distribution of extreme stock market returns, , 63, pp
383-408.
Longstaff, Francis and Eduardo Schwartz, 1995, A simple approach to valuing risky
fixed and floating rate debt, Vol. 50, No. 3, pp. 789-819.
Longstaff, Francis, Mithal, Sanjay and Eric Neis, 2003, The credit default swap market:
Is credit protection priced correctly?, Working paper, UCLA.
Marsch, Terry and Niklas Wagner, 2004, Return-Volume Dependence and Extremes in
International Markets, unpublished working paper, Haas School of Business, U.C.
Berkeley.
McNeil AJ and Frey R, 2000. “Estimation of tail-related risk measures for
heteroscedastic financial time series: an extreme value approach”, Journal of
Empirical Finance, 7: 271-300.
McNeil, A., 1997. On Extremes and Crashes, working paper.
Merrick, John J. Jr., 2000, Crisis Dynamics of Implied Default Recovery Rations:
Evidence from Russia and Argentina, Journal of Banking and Finance, Vol. 25,
No. 10, pp. 1921-1939.
Merton, Robert, 1974, On the pricing of corporate debt: The risk structure of interest
rates, Journal of Finance, Vol. 29, No. 2, pp. 449-469.
Miller, Darius and John Puthenpurackal, 2005, Security Fungibility and the Cost of
Capital: Evidence from Global Bonds, Journal of Financial and Quantitative
Analysis, forthcoming.
Ming, Hong G., 1998, The determinants of Emerging Market bond spread: Do economic
fundamentals matter?, Policy Research Working Paper 1899, The World Bank.

96
Mollemans, Michael, 2004, The Credit Rating Announcement Effect in Japan,
unpublished working paper, Macquarie University.
Morck, Randall, Yeung, Bernard and Wayne Yu, 2000, The Information Content of
Stock Markets: Why Do Emerging Markets Have Synchronous Stock Price
Movements?, Journal of Financial Economics, Vol. 58(1), pp 215-260.
Odders-White, Elizabeth and Mark Ready, 2003, Credit ratings and stock liquidity,
University of Wisconsin working paper.
Pagès, Henri, 2001, Can liquidity risk be subsumed in credit risk? A case study from
Brady bond prices, Bank for International Settlements, BIS working papers, No.
101.
Patro, Dilip and John Wald, 2005, Firm Characteristics and the Impact of Emerging
Market Liberalizations, Journal of Banking and Finance, forthcoming.
Poon, Ser_Huang, Rockinger, Michael and Jonathan Tawn, 2002, Modelling Extreme-
Value Dependence in International Stock Markets, unpublished working paper,
Lancaster University.
Poon, Ser-Huang and Han Lin, 2000, Taking Stock at the New Millenium, unpublished
working paper, Lancaster University.
Qi, Rong, 2001, Return-Volume Relation in the Tail: Evidence from Six Emerging
Markets, unpublished working paper, Columbia University.
Reich, Christian and Patrick Wegmann, 2002, Extremal Dependence between Market and
Liquidity Risk: An Analysis for the Swiss Market, unpublished working paper,
Wirtschaftswissenschaftliches Zentrum, Universität Basel.
Rigobon, R., 2000, Identification through heteroskedasticity: measuring “contagion”
between Argentinean and Mexican sovereign bonds, NBER working paper 7493.
Sarig, Oded and Arthur Warga, 1989, Bond Price Data and Bond Market Liquidity,
Journal of Financial and Quantitative Analysis, Vol. 24 (3), pp. 367 – 378.
Scherer, Paul K. and Marco Avellaneda, 2000, All for One ... One for All?: A Principal
Component Analysis of Latin American Brady Bond Debt from 1994 to 2000,
Working paper, New York University.
Schich, Sebastian, 2002, Dependencies between European stock markets when price
changes are unusually large, Discussion paper 12/02, Economic Research Centre
of the Deutsche Bundesbank.
Schultz, Paul, 2001, Corporate Bond Trading: A Peek Behind the Curtain, Journal of
Finance, Vol. 56, No. 2, pp. 677-698.
Straetmans, Stefan, Verschoor, Werner and Christian Wolff, 2003, Extreme U.S. stock
market fluctuations in the wake of 9/11, unpublished working paper, Maastricht
University.
Stulz, René M., 2003, Risk Management and Derivatives, Thomson-Southwestern, First
Edition.
97
Susmel, Raul, 2001, Extreme Observations and Diversification in Emerging Markets,
Journal of International Money and Finance, Vol. 20, pp. 971-986.
Sy, Amadou, 2003, Rating the Rating Agencies: Anticipating Currency Crises or Debt
Crises, IMF working paper No. 03/122.
Tawn, Jonathan A., 1988. “Bivariate Extreme Value Theory: Models and estimation”,
Biometrika, 75, 3, pp. 397-415.
Tolikas, Konstantinos and Richard Brown, 2003, The Distribution of Extreme Daily
Share Returns in the Athens Stock Exchange, unpublished working paper,
University of Dundee.
Wakeman, L. Macdonald, 1992, The Real Function of Bond Rating Agencies, in The
Revolution in Corporate Finance, ed. Joel M. Stern and Donald H. Chew, Jr.
Cambridge, MA: Blackwell Publishers, 1992
Wansley, James W. and Terrance M. Clauretie, 1985, The impact of Creditwatch
placements on equity re-turns and bond prices, Journal of Financial Research,
Vol. 8(1), pp 31-42.
Westphalen, Michael, 2002, Valuation of Sovereign Debt with Strategic Defaulting and
Rescheduling, Research Paper 43, FAME Research Paper Series.
Westphalen, Michael, 2003, The Determinants of Sovereign Bond Credit Spread
Changes, Working Paper, FAME.
Zaima, Janis and Joseph McCarthy, 1988, The impact of bond rating changes on common
stocks and bonds: tests of the wealth redistribution hypothesis, Financial Review,
Vol. 23, pp. 483 – 498.

98
 APPENDIX A

A COMPARISON OF SOVEREIGN BOND COVERAGE ON DATASTREAM

AND THE NAIC DATABASE.

99
 Researchers have access today to two main sources for international bond data

time series. One is Datastream International and the other one is the Fixed Income

Securities Database provided by the University of Houston (the Warga database). Both

databases differ in format, coverage, and data providers. While Datastream does not

make public the identity of their data provider due to legal technicalities, I found through

informal conversations with their support staff that Lehman Brothers and ISMA

(International Securities Market Association) are among their main data providers. The

inclusion of Lehman Brothers as a data provider for Datastream is re-assuring since prior

to 1998 Lehman Brothers used to be the main source of fixed income instruments data

(they discontinued support of that database in 1998).

To analyze Datastream and Warga databases’ coverage on sovereign bonds I

compare the set of bonds available on both. Ideally, I would have wanted to compute

correlations of monthly returns for the bonds on both databases, however, that is not

possible because the Warga database does not provide with time series data for the bonds

it covers. Before conducting the comparison I will make a brief description of the Warga

data. This $12,000 database comes in two CDs. The first one contains Microsoft Access

files that include lots of tables with exhaustive information (static) data for 157,488

different bonds issued by 10,057 different issuers. When I filter out sovereign issues

using the included ‘industry_code’ variable, I come up with 1,579 bond issues from 146

sovereign issuers. The files in the first CD, however, do not contain any price information

other than prices at launch. The second CD includes information from all insurance

companies’ buys and sells of instruments in the first CD for 1995-1999, provided by

NAIC –National Association of Insurance Commissioners. Although the web site for this
100
database (http://www.bauer.uh.edu/awarga/comp.html) says that price information is

available up until 2001, the last update purchased a few months ago by the finance

department at Ohio State includes only data up to the end of 1999. Since the price

information is only for the purchase and sell of securities, and because many of these

securities do not trade frequently, it is not possible to construct a regular time series –say

monthly- with prices for these issues. Further, the price data comes in four files: two files

for purchases and two files for sales. The first pair of files covers the 1995-1998 period

and the second set covers the 1998-1999 period. Datastream has continuous coverage

from the early 90s until today.

The only way to automatically link information from Datastream and the Warga

data is through the ISIN (International Security Identification Number) code, which is

available on both of them. For 35 emerging market countries included in my first

dissertation essay, Datastream has codes (a code is Datastream’s own id number) for 291

different bonds. Out of those 291 different sovereign bonds, there is ISIN information for

276 of them. Table 1 has the list of such countries.

Table 2 has information of the availability of data from those 35 emerging

markets on the Warga price files. There are total 105 different bonds across all four

pricing files. Table 2 shows that there 73, 58, 88 and 89 unique bonds in each file.

However, not all those bonds are present in Datastream. Only 40, 35, 55 and 56 bonds

from each Warga file respectively are in both databases. As mentioned before, there are

not nearly enough observations per bond to construct a time series. The average number

of observations per bond is 10, the minimum is 1 (and also the mode) and the maximum

is 30, although in those cases there are many observations clustered around certain
101
periods of time, and therefore having 30 observations does not mean we have 30 monthly

observations. So, we end up with 552, 221, 844 and 745 –respectively- usable data points

where we have clean price information for a specific bond on an the same date. Simple

clean price correlations range between 92% and 97%, whereas holding period return

correlations between Datastream and Warga data range from 73% to 89%.

These numbers make us feel more comfortable with the overall quality of

Datastream data, at least for sovereign data. In my job market paper I use yields from

Datastream, and being that yields are only complex transformations of clean prices, the

relevant correlations are the one between clean prices. Further, it seems clear that

Datastream has a better coverage of sovereign issues (276 vs. 105 bonds in the 35

emerging market countries analyzed here), plus a deeper one (this updated version of the

Warga database only has data from 1994 until the end of 1999). Hopefully, this quick

analysis should ease concerns about the integrity and quality of Datastream bond price

data.

102
APPENDIX B

TABLES

103
 Variables Expected sign Rationale

Life to maturity Uncertain Stulz (2003) explains how the relation between time to
maturity and credit spreads depends on the relative size of
debt and firm value. Helwege and Turner (1999) establish
that due to survivorship bias only relatively better rated
countries issue longer-term debt. Also, investors might
perceive shorter-term sovereign bonds as having higher
probability of default and therefore higher expected losses.
This belief is reinforced by the fact that some countries'
debt do not incorporate cross-default clauses, making easier
for countries facing financial distress default first on issues
with closer maturities. Longer-term bonds are, in this
setting, perceived as safer since countries could have time
to implement reforms that bring them out of financial
distress.
Debt to foreign Positive A higher ratio of any of these two measures implies a
reserves ratio smaller distance-to-default. So, larger values of them
and should be associated with higher spreads.
Debt to exports ratio

Country risk measure Positive This measure has a higher value for countries that are
perceived to have higher political risks, for instance, higher
expropriation risk. The larger the value of this variable, the
higher the debt spread.
Local stock market Positive This variable is an imperfect proxy of a country’s wealth
volatility volatility. Still, we expect a positive relation since more
volatility makes default more likely.
U.S. Treasury yield Negative Assuming that the country’s wealth follows a risk-neutral
curve level drift, higher rates should be associated with higher drifts
which in turn should reduce debt spreads. Also, Stulz
(2003) shows how debt value decreases with maturity. This
reduced the probability of default, and ergo, spreads.
U.S. Treasury yield Negative We assume a positive slope to signal higher future interest
curve slope rates. The previous arguments then suggest a negative
relation between spreads and the interest rate slope.
World return Negative A world index return is included as a proxy of the world
economic climate or business cycle. On average, we would
expect smaller spreads when the world as a whole is doing
well.
Table 1. Expected signs on explanatory variables for sovereign sample.

104
Panel A: Descriptive Statistics for Spreads
Credit Spread (%) All sample AAA to A+ A to BBB- BB+ to B B- to C No rating

Mean 4.834 0.488 2.098 4.971 12.365 4.478

Std. Dev. 4.555 0.253 1.369 3.261 8.090 3.860
Skewness 3.195 2.263 2.170 2.451 1.599 3.503
Kurtosis 18.001 15.840 12.672 14.145 5.063 18.620
Max 39.390 2.204 15.250 33.181 39.390 37.455
90% 39.173 1.332 10.142 30.967 39.173 23.681
10% 0.040 0.076 0.068 0.094 1.408 1.034
Min 0.019 0.038 0.019 0.046 1.241 1.019
No. of observations 9275 165 2213 5450 829 618

Panel B: Means for Selected Country Variables

Mean All sample AAA to A+ A to BBB- BB+ to B B- to C No rating

Debt-to-reserves 3.919 0.959 1.742 4.733 4.746 1.997

Debt-to-exports 40.603 3.349 12.330 52.654 39.359 8.207
Political risk 51.260 21.330 40.480 53.546 66.028 62.255
U.S. Treasury yield level (%) 5.285
U.S. Treasury yield slope 1.326

Panel C: Other data

Number of bonds 233

Number of countries 37
The sample includes all non-callable, non-puttable sovereign bonds in U.S. dollars that traded between January 1990 and January 2003.
All data is from Datastream. The spread over U.S. Treasuries is computed as the difference between the redemption yield of the
sovereign bond and the value of a linear interpolation of the U.S. Treasury yield curve to obtain the yield of a U.S. instrument with
identical maturity. Debt/Reserves is computed using all outstanding foreign debt (bank loans, Brady bonds and Eurobonds) divided by
the total number of international reserves in current U.S. dollars. Political risk is the value of The Economist Intelligence Unit's country
index. Local stock market volatility is the standard volatility computed each month from daily stock market returns in U.S. dollars. The
U.S. Treasury yield level is the yield of the 10 year U.S. Treasury note. The U.S. Treasury slope is computed as the difference between
the yield of the 10 year and the 2 year U.S. Treasury notes. The world stock return is the log return of Datastream's world total return
index.
Table 2. Summary statistics for sovereign sample.

105
∆Spread over U.S. Treasury AAA to BBB- BB+ to B B- to C Overall sample

Constant 0.452 -0.06 -2.769 -0.071

(1.74)* (0.49) (1.55) (0.65)
∆Years to maturity 5.85 -0.619 -34.553 -0.833
(1.87)* (0.42) (1.61) (0.64)
∆Debt to foreign reserves 0.048 -0.009 0.755 0.04
(0.85) (0.62) (7.79)*** (2.78)***
∆Political risk 0.006 0.052 0.133 0.075
(0.62) (4.46)*** (3.96)*** (7.86)***
∆Political risk lagged 0.03 0.002 0.008 0.016
(3.06)*** (0.15) (0.00) (1.64)
∆Political risk 2nd lag -0.013 0.002 0.041 0.018
(1.33) (0.14) (1.20) (1.85)*
∆Local volatility lagged 0.003 0.088 0.067 0.075
(0.05) (11.12)*** (3.00)*** (11.69)***
∆U.S. Treasury level -0.807 -1.088 -0.882 -0.978
(16.24)*** (12.85)*** (2.60)*** (14.32)***
∆U.S. Treasury slope 0.135 0.664 1.073 0.51
(1.75)* (5.17)*** (2.15)** (4.86)***
Local return lagged -0.79 -5.613 -4.663 -4.664
(5.22)*** (25.88)*** (7.03)*** (26.74)***

Observations 1630 3870 690 6316

R-squared 0.19 0.30 0.22 0.22

This table shows estimates from an OLS regression model with Newey-West adjusted errors. We estimated the following equation to
each bond observation: ∆Spreadi,t = Constant + β1*∆Debt to foreign reserves ratioi,t + β2*∆Country risk measurei,t + β3*∆U.S.
Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*∆Local volatilityi,t-1 + β6*Local returnt-1 + β7*∆Years to
maturityi,t + εi,t. We estimated eight different specifications of this basic equation, substituting duration for years to maturity and debt to
exports for debt to reserves. Years to maturity is the remaining life of a bond expressed in years, duration is a Macaulay’s duration
expressed in years, debt to reserves is the ratio of total debt outstanding (bank loans, Brady and Eurobond issues) denominated in U.S.
dollars divided by the total amount of international reserves also denominated in U.S. dollars. Debt to exports is the ratio of total debt
outstanding (bank loans, Brady and Eurobond issues) denominated in U.S. dollars divided by the nominal monthly value of exports in
U.S. dollars. Political risk is the value of The Economist Intelligence Unit's country index. Local stock market volatility is the standard
volatility computed each month from daily stock market returns in U.S. dollars. The U.S. Treasury yield level is the yield of the 10 year
U.S. Treasury note. The U.S. Treasury slope is computed as the difference between the yield of the 10 year and the 2 year U.S. Treasury
notes. The world stock return is the log return of Datastream's world total return index. Absolute value of t statistics are in parentheses;
*, **, *** denote significance at the 10%; 5%; and 1% level respectively.

Table 3. Sovereign spreads fixed effect regressions.

106
 Variables Expected sign Rationale

Life to maturity Uncertain Stulz (2003) explains how the relation between time to
maturity and credit spreads depends on the relative size of
debt and firm value. Helwege and Turner (1999) establish
that due to survivorship bias only relatively better rated
countries issue longer-term debt. Also, investors might
perceive shorter-term sovereign bonds as having higher
probability of default and therefore higher expected losses.
This belief is reinforced by the fact that some countries'
debt do not incorporate cross-default clauses, making easier
for countries facing financial distress default first on issues
with closer maturities. Longer-term bonds are, in this
setting, perceived as safer since countries could have time
to implement reforms that bring them out of financial
distress.
Leverage Positive A higher leverage ratio increases the probability of a firm
and facing financial distress. This should increase spreads.
Equity return Also, from a contingent claims approach, equity return
volatility volatility can proxy for firm's value volatility. A higher
volatility increases the chance of the firm's value process to
cross the threshold at which a firm defaults on its debt.
U.S. Treasury yield Negative Assuming that the firm's value follows a risk-neutral drift,
curve level higher rates should be associated with higher drifts which
in turn should reduce debt spreads. Also, Stulz (2003)
shows how debt value decreases with maturity. This
reduced the probability of default, and ergo, spreads.
U.S. Treasury yield Negative We assume a positive slope to signal higher future interest
curve slope rates. The previous arguments then suggest a negative
relation between spreads and the interest rate slope.
S&P 500 return Negative As the economic environment improves, measured by the
S&P return, we expect firms to do better and therefore to
reduce the probability of defaulting on their debt.

Table 4. Expected signs on explanatory variables for domestic sample.

107
Panel A: Descriptive Statistics for Debt Spreads

No
Overall Leverage Class leverage
Credit Spread (%) sample Low Medium High data

Mean 2.393 1.359 1.717 3.263 2.633

Std. Dev. 2.779 1.187 1.319 3.566 3.017
Skewness 4.543 5.621 5.210 3.947 4.007
Kurtosis 31.661 77.578 70.700 22.628 24.890
Max 29.989 26.846 29.636 29.847 29.989
90% 29.849 22.659 26.812 29.812 29.849
10% 0.0012 0.0031 0.0125 0.0279 0.0021
Min 0.0007 0.0012 0.0007 0.0062 0.0010
No. of observations 71831 11061 11247 11332 38191

Panel B: Means for Selected Variables

Mean

Leverage 0.343
Std. Dev. 0.023
U.S. Treasury yield level (%) 5.285
U.S. Treasury yield slope 1.326

Panel C: Other data

Number of bonds 2,930

Number of countries 649

The sample includes all non-callable, non-puttable domestic bonds issued by industrial firms in U.S. dollars that traded between
January 1990 and January 2003. All data is from Datastream. The spread over U.S. Treasuries is computed as the difference
between the redemption yield of the sovereign bond and the value of a linear interpolation of the U.S. Treasury yield curve to
obtain the yield of a U.S. instrument with identical maturity. Leverage is computed as the ratio of book value of debt divided by
the sum of book value of debt and market value of equity. Stock market volatility is computed monthly from daily stock market
log returns. The U.S. Treasury yield level is the yield of the 10 year U.S. Treasury note. The U.S. Treasury slope is computed as
the difference between the yield of the 10 year and the 2 year U.S. Treasury notes.

Table 5. Summary statistics for domestic sample.

108
 Leverage Class Overall
∆Spread over U.S. Treasury Low Medium High sample

Constant -0.046 -0.031 0.075 -0.019

(2.18)** (1.36) (0.87) (0.85)
∆Years to maturity -0.591 -0.141 0.834 -0.249
(2.35)** (0.52) (0.81) (0.93)
∆Leverage -0.447 -0.005 -0.027 0.018
(1.96)** (0.03) (0.09) (0.12)
∆leverage lagged 3.127 0.941 6.747 4.445
(13.80)*** (5.69)*** (22.32)*** (30.77)***
∆Stock return volatility 1.073 -0.164 3.133 2.149
(2.43)** (0.31) (4.18)*** (5.84)***
∆Stock return volatility lagged 4.141 3.323 11.866 8.582
(9.59)*** (6.28)*** (15.69)*** (23.42)***
∆U.S. Treasury level -0.218 -0.338 -0.421 -0.323
(14.53)*** (22.27)*** (11.46)*** (23.28)***
∆U.S. Treasury slope -0.002 0.023 0.116 0.031
(0.08) (0.86) (1.83)* -1.28
S&P return lagged -0.002 0.008 -0.007 -0.003
(1.99)** (0.09) (3.32)*** (4.05)***

Observations 9080 9679 8709 27468

R-squared 0.07 0.07 0.13 0.09

This table shows estimates from an OLS regression model with Newey-West adjusted errors. We estimated the
following equation to each bond observation: ∆Spreadi,t = Constant + β1*∆Leverage ratioi,t + β2*∆Stock return
volatilityi,t + β3*∆U.S. Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*S&P index returni,t-1 +
β6*∆Years to maturityi,t + εi,t. We estimated six different specifications of this basic equation, substituting duration and
modified duration for years to maturity. Years to maturity is the remaining life of a bond expressed in years, duration
is a Macaulay’s duration expressed in years, modified duration is estimated as duration divided by (1+ redemption
yield).Stock return volatility is the standard volatility of each firm's stock return computed each month from daily log
returns in U.S. dollars. The U.S. Treasury yield level is the yield of the 10 year U.S. Treasury note. The U.S. Treasury
slope is computed as the difference between the yield of the 10 year and the 2 year U.S. Treasury notes. The S&P
index return is the log return of Datastream's S&P 500 total return index. Absolute value of t statistics are in
parentheses; *, **, *** denote significance at the 10%; 5%; and 1% level respectively.

Table 6. Domestic spreads fixed effect regressions.

109
 Panel A: Sovereign bins
s11 s12 s13 s21 s22 s23 s31 s32 s33
s11 1.000
s12 0.826 1.000
s13 -0.007 0.511 1.000
s21 0.997 0.817 -0.004 1.000
s22 0.828 0.953 0.459 0.818 1.000
s23 0.367 0.781 0.895 0.369 0.773 1.000
s31 0.983 0.821 0.011 0.989 0.822 0.370 1.000
s32 0.891 0.896 0.259 0.886 0.966 0.624 0.897 1.000
s33 0.820 0.932 0.394 0.801 0.950 0.708 0.806 0.946 1.000

Panel B: Domestic bins

d11 d12 d13 d21 d22 d23 d31 d32 d33
110

d11 1.000
d12 0.978 1.000
d13 0.616 0.622 1.000
d21 0.967 0.968 0.744 1.000
d22 0.921 0.955 0.707 0.928 1.000
d23 0.685 0.673 0.856 0.747 0.747 1.000
d31 0.930 0.971 0.679 0.954 0.943 0.633 1.000
d32 0.859 0.920 0.712 0.893 0.973 0.659 0.960 1.000
d33 0.865 0.874 0.876 0.909 0.915 0.797 0.909 0.922 1.000

continued...
 Table 7 continued
Panel C: Sovereign and Domestic bins
s11 s12 s13 s21 s22 s23 s31 s32 s33 d11 d12 d13 d21 d22 d23 d31 d32 d33
s11 1.000
s12 0.826 1.000
s13 -0.007 0.511 1.000
s21 0.997 0.817 -0.004 1.000
s22 0.828 0.953 0.459 0.818 1.000
s23 0.367 0.781 0.895 0.369 0.773 1.000
s31 0.983 0.821 0.011 0.989 0.822 0.370 1.000
s32 0.891 0.896 0.259 0.886 0.966 0.624 0.897 1.000
s33 0.820 0.932 0.394 0.801 0.950 0.708 0.806 0.946 1.000
d11 0.227 0.119 -0.168 0.221 0.068 -0.087 0.205 0.101 0.100 1.000
d12 0.248 0.127 -0.173 0.245 0.117 -0.079 0.242 0.157 0.114 0.978 1.000
d13 -0.211 0.063 0.464 -0.182 0.092 0.388 -0.159 -0.046 -0.094 0.616 0.622 1.000
d21 0.064 0.073 0.020 0.065 0.082 0.054 0.065 0.070 0.057 0.967 0.968 0.744 1.000
d22 0.194 0.154 -0.004 0.212 0.150 0.074 0.210 0.142 0.036 0.921 0.955 0.707 0.928 1.000
111

d23 -0.156 0.193 0.572 -0.141 0.100 0.440 -0.124 -0.079 -0.005 0.685 0.673 0.856 0.747 0.747 1.000
d31 0.159 0.062 -0.143 0.160 0.129 -0.059 0.181 0.173 0.074 0.930 0.971 0.679 0.954 0.943 0.633 1.000
d32 0.161 0.086 -0.058 0.181 0.160 0.032 0.196 0.177 0.005 0.859 0.920 0.712 0.893 0.973 0.659 0.960 1.000
d33 -0.003 0.075 0.173 0.020 0.116 0.172 0.040 0.057 -0.063 0.865 0.874 0.876 0.909 0.915 0.797 0.909 0.922 1.000
This table presents the correlation structure of the residual bins. Each sample (sovereign and domestic) was divided in three maturity categories and three leverage (debt to
reserves, in the sovereign case) categories. The cutoff values for each category were determined using the 33rd and 66th centile to ensure an approximately equal number of
observations in each bin. Each observation was assigned to a category. To compute the residuals, regressions were conducted in each bin. Then, for each bin, we average across
residuals. The bins are named dij and sij for i, j=1, 2, 3, where d stands for domestic and s stands for sovereign, i for maturity category (1 = shot-term, 2 = medium-term, 3 =
long-term), and j stand for leverage (debt to reserves) category (1 = low, 2 = medium, 3 = high). For example, d23 refers to a domestic, medium-term, high-leverage bin. Each
sovereign bin contains 66 observations, while each domestic bin contains 148 observations.

Table 7. Correlation structure of residuals.

 Eigenvalue
8.0
Panel A: Sovereign bins 7.0
Component Eigenvalue Difference Proportion Cumulative 6.0
1 6.8483 5.0152 0.7609 0.7609 5.0
2 1.8331 1.6574 0.2037 0.9646 4.0
3 0.1757 0.1047 0.0195 0.9841 3.0
4 0.0710 0.0406 0.0079 0.9920 2.0
1.0
0.0
0 1 2 3 4 5
Component

Eigenvalue
10.0
Panel B: Domestic bins 8.0
Component Eigenvalue Difference Proportion Cumulative
1 7.7606 6.9927 0.8623 0.8623 6.0
2 0.7679 0.5152 0.0853 0.9476 4.0
3 0.2527 0.1142 0.0281 0.9757
4 0.1385 0.0932 0.0154 0.9911 2.0

0.0
0 1 2 3 4 5
Component

Eigenvalue
8.0
7.0
Panel C: Domestic and Sovereign bins 6.0
Component Eigenvalue Difference Proportion Cumulative 5.0
1 7.5705 1.6083 0.4206 0.4206 4.0
2 5.9621 2.9935 0.3312 0.7518 3.0
3 2.9687 2.3462 0.1649 0.9167 2.0
4 0.6225 0.2166 0.0346 0.9513 1.0
0.0
0 1 2 3 4 5
Component

This table presents the results of applying principal components analysis to the residuals. Each sample (sovereign and
domestic) was divided in three maturity categories and three leverage (debt to reserves, in the sovereign case) categories.
Each observation was assigned to a category. For each bond in each sovereign bin we estimated the following equation:
∆Spreadi,t = Constant + β1*∆Debt to foreign reserves ratioi,t + β2*∆Country risk measurei,t + β3*∆U.S. Treasury yield
curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*∆Local volatilityi,t-1 + β6*Local returnt-1 + β7*∆Years to maturityi,t
+ εi,t. For bonds in the domestic bins the following equation was estimated: ∆Spreadi,t = Constant + β1*∆Leverage ratioi,t +
β2*∆Stock return volatilityi,t + β3*∆U.S. Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*S&P index
returni,t-1 + β6*∆Years to maturityi,t + εi,t. Residuals for each bond were computed and averaged across bins. For ease of
interpretation, only the first four components are shown for each panel.

Table 8. Principal component analysis of residuals.

112
 Sovereign bonds Domestic bonds
Overall Leverage Class Overall
∆Spread over U.S. Treasury AAA to BBB- BB+ to B B- to C sample ∆Spread over U.S. Treasury Low Medium High sample
Constant 0.203 -0.063 -1.055 -0.067 Constant 0.026 -0.019 0.103 0.019
(0.65) (0.81) (0.35) (0.72) (1.14) (0.83) (1.06) (0.68)
∆Years to maturity 2.875 -0.892 -14.339 -1.213 ∆Years to maturity 0.175 -0.006 1.009 0.109
(0.77) (0.97) (0.40) (1.10) (0.64) (0.02) (0.88) (0.33)
∆Debt to foreign reserves 0.103 0.035 1.047 0.125 ∆Leverage -0.673 0.147 0.198 0.141
(0.86) (2.64)*** (6.80)*** (7.01)*** (2.78)*** (0.86) (0.56) (0.83)
∆Political risk 0.023 0.019 0.086 0.049 ∆leverage lagged 4.122 1.19 7.622 5.393
(2.06)** (2.44)** (2.24)** (5.77)*** (17.10)*** (7.08)*** (21.00)*** (31.64)***
∆Political risk lagged 0.031 0.022 -0.026 0.024 ∆Stock return volatility 0.936 -1.009 2.033 1.495
(2.76)*** (2.89)*** (0.67) (2.83)*** (2.29)** (1.86)* (2.26)** (3.57)***
∆Political risk 2nd lag -0.01 0.004 0.046 0.021 ∆Stock return volatility lagged 4.487 3.281 12.396 9.197
(0.92) (0.56) (1.19) (2.42)** (11.46)*** (6.39)*** (14.37)*** (22.91)***
∆U.S. Treasury level
∆Local volatility lagged -0.005 0.054 0.113 0.052 -0.085 -0.17 -0.26 -0.18
(0.72) (10.49)*** (3.95)*** (9.20)*** (3.65)*** (6.13)*** (3.25)*** (6.42)***
113

∆U.S. Treasury level ∆U.S. Treasury slope

-0.59 -0.495 -1.903 -0.59 -0.154 -0.021 -0.106 -0.091
(6.80)*** (6.33)*** (2.37)** (6.72)*** (4.26)*** (0.49) (0.87) (2.11)**
∆U.S. Treasury slope
-0.147 -0.207 2.163 -0.035 S&P return lagged -0.004 -0.002 -0.009 -0.005
(1.24) (1.83)* (2.25)** (0.28) (4.55)*** (1.81)* (3.07)*** (4.92)***
1st factor sovereign
Local return lagged -0.641 -3.645 -6.104 -3.601 0.006 0.004 0.006 0.005
(3.73)*** (24.99)*** (7.59)*** (22.78)*** (3.20)*** (1.68)* (2.84)** (2.35)**
1st factor domestic 2dn factor sovereign
0.026 -0.039 -0.161 -0.044 -0.001 -0.008 0.015 0.003
(3.56)*** (5.39)*** (2.72)*** (5.42)*** (0.48) (2.26)** (1.44) (0.89)
2dn factor domestic 1st factor sovereign lagged
-0.024 -0.021 0.141 0.024 0.004 0.001 -0.006 -0.001
(1.75)* (1.60) (1.60) (1.69)* (1.84)* (0.53) (0.77) (0.53)

continued...
 Table 9 continued
1st factor domestic lagged 2dn factor sovereign lagged
0.013 0.036 -0.192 0.002 -0.013 -0.015 -0.003 -0.004
(1.85)* (5.07)*** (2.93)*** 0.00 (3.23)*** (2.90)*** (0.17) (0.78)
2dn factor domestic lagged
-0.084 -0.13 0.492 -0.07
(4.04)*** (6.79)*** (2.74)*** (3.23)***
Observations 1405 3460 513 5504 Observations 6630 5735 5864 18229
R-squared 0.20 0.35 0.31 0.20 R-squared 0.11 0.08 0.16 0.12

This table shows estimates from an OLS regression model with Newey-West adjusted errors. We estimated the following equation to each sovereign bond observation: ∆Spreadi,t = Constant + β1*∆Debt
to foreign reserves ratioi,t + β2*∆Country risk measurei,t + β3*∆U.S. Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*∆Local volatilityi,t-1 + β6*Local returnt-1 + β7*∆Years to
maturityi,t+ β8*1st factor domestict + β9*2nd factor domestict + εi,t. For each domestic bond. we estimated the following equation: ∆Spreadi,t = Constant + β1*∆Leverage ratioi,t + β2*∆Stock return
volatilityi,t + β3*∆U.S. Treasury yield curve levelt + β4*∆U.S. Treasury yield curve slopet + β5*S&P index returni,t-1 + β6*∆Years to maturityi,t + β8*1st factor sovereignt + β9*2nd factor sovereignt +
εi,t. Years to maturity is the remaining life of a bond expressed in years, debt to reserves is the ratio of total debt outstanding (bank loans, Brady and Eurobond issues) denominated in U.S. dollars divided
by the total amount of international reserves also denominated in U.S. dollars. Political risk is the value of The Economist Intelligence Unit's country index. Local stock market volatility is the standard
volatility computed each month from daily stock market returns in U.S. dollars. The U.S. Treasury yield level is the yield of the 10 year U.S. Treasury note. The U.S. Treasury slope is computed as the
difference between the yield of the 10 year and the 2 year U.S. Treasury notes. The S&P stock return is the log return of Datastream's S&PCOMP total return index. Stock return volatility is the standard
volatility of each firm's stock return computed each month from daily log returns in U.S. dollars. The S&P index return is the log return of Datastream's S&P 500 total return index. Absolute value of t
statistics are in parentheses; *, **, *** denote significance at the 10%; 5%; and 1% level respectively.
Table 9. Sovereign and domestic regressions including the common factors.
114
 R-squared
No exog. All
Equation Obs vars. vars.
1st factor domestic 59 0.3684 0.4496
1st factor sovereign 59 0.0586 0.3769
2nd factor domestic 59 0.2029 0.4978
2nd factor sovereign 59 0.1490 0.6732

1st factor domestic Coeff. Std. Error z P>z 2nd factor domestic Coeff. Std. Error z P>z
1st factor domestic 1st factor domestic
Lag 1 -0.060 0.120 -0.50 0.614 Lag 1 -0.064 0.060 -1.06 0.290
Lag 2 0.014 0.115 0.12 0.902 Lag 2 0.076 0.058 1.31 0.189
1st factor sovereign 1st factor sovereign
Lag 1 -0.072 0.098 -0.73 0.467 Lag 1 -0.091 0.050 -1.82 0.068*
Lag 2 0.424 0.108 3.94 0.000*** Lag 2 0.079 0.054 1.45 0.146
2nd factor domestic 2nd factor domestic
Lag 1 -0.226 0.298 -0.76 0.448 Lag 1 0.244 0.151 1.62 0.106
Lag 2 -0.500 0.282 -0.77 0.427 Lag 2 -0.368 0.143 -2.58 0.010***
2nd factor sovereign 2nd factor sovereign
Lag 1 0.188 0.219 0.86 0.389 Lag 1 0.187 0.111 1.69 0.092*
Lag 2 0.241 0.222 1.09 0.278 Lag 2 0.123 0.112 1.1 0.273
∆flowstocks -0.032 0.017 -1.880 0.060* ∆flowstocks 0.017 0.008 1.980 0.048**
∆flowbonds -0.135 0.056 -2.420 0.015** ∆flowbonds 0.043 0.028 1.770 0.076*
∆netreserves 6.786 5.752 1.18 0.238 ∆netreserves 13.740 2.909 4.72 0.000***
∆onoff 9.751 9.013 1.08 0.279 ∆onoff -0.083 4.558 -0.02 0.986
Constant 0.309 0.222 1.39 0.164 Constant -0.104 0.112 -0.92 0.356

1st factor sovereign Coeff. Std. Error z P>z 2nd factor sovereign Coeff. Std. Error z P>z
1st factor domestic 1st factor domestic
Lag 1 0.225 0.136 1.65 0.099* Lag 1 -0.024 0.061 -0.39 0.696
Lag 2 -0.179 0.131 -1.36 0.174 Lag 2 0.081 0.058 1.38 0.168
1st factor sovereign 1st factor sovereign
Lag 1 0.128 0.112 1.14 0.256 Lag 1 0.002 0.050 0.04 0.967
Lag 2 -0.119 0.123 -0.97 0.333 Lag 2 0.112 0.055 2.06 0.040**
2nd factor domestic 2nd factor domestic
Lag 1 -0.464 0.340 -1.36 0.172 Lag 1 0.326 0.152 2.15 0.032**
Lag 2 0.469 0.322 1.46 0.145 Lag 2 -0.381 0.143 -2.66 0.008***
2nd factor sovereign 2nd factor sovereign
Lag 1 0.388 0.250 1.55 0.120 Lag 1 0.112 0.111 1.01 0.312
Lag 2 -0.113 0.253 -0.45 0.654 Lag 2 0.287 0.113 2.54 0.011**
∆flowstocks -0.052 0.019 -2.710 0.007* ∆flowstocks 0.054 0.009 6.360 0.000***
∆flowbonds -0.312 0.064 -4.910 0.000*** ∆flowbonds -0.112 0.028 -3.950 0.000***
∆netreserves 18.746 6.564 2.86 0.004*** ∆netreserves 13.553 2.924 4.64 0.000***
∆onoff 18.549 10.286 1.8 0.071* ∆onoff -0.731 4.582 -0.16 0.873
Constant -0.055 0.253 -0.22 0.827 Constant 0.028 0.113 0.25 0.801
This table shows estimates a vector autoregression model of the following form:
k k
FacSovt = a1 + ∑ β1 j FacSovt − j + ∑ γ 1 j FacDomt − j + δ 1 X t +ε1
j =1 j =1

k k
FacDomt = a2 + ∑ β 2 j FacSovt − j +∑γ 2 j FacDomt − j + δ 2 X t +ε 2
j =1 j =1

The first and second domestic factors are the factors extracted from the principal component analysis of the residuals of equation (2)
applied to the domestic bins. The first and second sovereign factors are extracted from the principal component analysis of the
residuals of equation (1) applied to the sovereign bins. Net borrowed reserves is computed as total borrowing minus extended credit
minus excess reserves, divided by total reserves. Onoff is the difference between the on-the-run thirty year U.S. Treasury bond and
the most recent off-the-run bond. Flowsstocks is from the IFC’s statistics and is the amount of money flowing into equity mutual
funds. Flowbonds is from the same source and represents the flows into bond funds; *, **, *** denote significance at the 10%; 5%;
and 1% level respectively.
Table 10. Vector autoregression model with exogenous variables.

115
 Argentina Brazil Chile
Local Local Local
index ADRs index ADRs index ADRs

Mean 0.00022 0.00056 0.00030 -0.00407 0.00067 0.00074

Maximum 0.12014 0.25784 0.15302 0.40547 0.17549 0.16866
Minimum -0.13380 -0.51544 -0.12137 -0.69315 -0.14989 -0.15095
Std. Dev. 0.01905 0.03211 0.02190 0.03040 0.01559 0.01602

Observations 1934 1934 1694 1694 2870 2870

Unconditional
correlation with the 0.42392 0.17064 0.37893 0.17120 0.15943 0.16059
S&P 500

Colombia Mexico Venezuela S&P 500

Local Local Local
index ADRs index ADRs index ADRs

Mean -0.00029 -0.00037 0.00052 0.00002 -0.00027 0.00033 0.00052

Maximum 0.08753 0.15127 0.19022 0.19088 0.29747 0.28549 0.04928
Minimum -0.15875 -0.15944 -0.25528 -0.27844 -0.51028 -0.50828 -0.07022
Std. Dev. 0.01231 0.01488 0.01954 0.01841 0.02891 0.03060 0.00928

Observations 2298 2298 2870 2870 2870 2870 2870

Unconditional
correlation with the 0.03345 -0.00292 0.40539 0.34982 0.05350 0.03456 1.00000
S&P 500

This table provides with basic statistics for the data. Log returns in dollars were calculated using daily data
from Datastream. In each case, the local index corresponds to the country index return available in
Datastream. ADRs is an equally weighted index of live ADRs of each country in the U.S. Argentina, Brazil
and Colombia have less observations because of data conflicts. Specifically, Argentina and Brazil, have
problems when backfilling series because of changes in their legal currencies and monetary regimes.

Table 11. Summary statistics.

116
 Panel A: Extreme correlations for pairs formed by S&P and local stock market returns.

Number of left tail exceedances Number of right tail exceedances

25 50 100 200 300 300 200 100 50 25
Argentina 0.0563 0.0572 0.1393 0.1805 0.1902 0.1433 0.0917 0.0429 0.0835 NA
Brazil 0.3190 0.2001 0.3525 0.3471 0.4936 0.4643 0.3955 0.3511 0.2134 0.1093
Chile 0.1107 0.1923 0.1816 0.2165 0.2281 0.2132 0.1456 0.0983 0.0566 0.0566
Colombia NA 0.0285 0.0979 0.1263 0.2085 0.2490 0.1576 0.0426 0.0284 NA
Mexico 0.3772 0.3812 0.3709 0.4407 0.4691 0.3731 0.3170 0.3029 0.3416 0.2194
Venezuela 0.0563 0.0572 0.1393 0.1805 0.1902 0.1433 0.0917 0.0429 0.0835 NA

Panel B: Extreme correlations for pairs formed by S&P and en equally weighted basket of ADRs.

Number of left tail exceedances Number of right tail exceedances

25 50 100 200 300 300 200 100 50 25
Argentina 0.2348 0.3324 0.3705 0.3850 0.4785 0.3927 0.3476 0.2505 0.2485 NA
Brazil 0.3203 0.3029 0.4138 0.4194 0.4880 0.4613 0.3858 0.3325 0.2659 0.2119
Chile 0.2126 0.1947 0.1697 0.2168 0.2493 0.2040 0.1449 0.0841 0.0285 NA
Colombia 0.0566 0.0288 0.0990 0.1398 0.1708 0.1772 0.1243 0.0286 0.0284 NA
Mexico 0.2290 0.2809 0.3201 0.4263 0.4378 0.3250 0.2634 0.2487 0.2198 0.1659
Venezuela NA 0.0566 0.0853 0.1402 0.1759 0.1436 0.0849 0.0568 0.0562 NA

Panel C: Extreme correlations for Mexico before and after the 1995 Mexican crisis.

Number of left tail exceedances Number of right tail exceedances

25 50 100 200 300 300 200 100 50 25
Local before 0.1149 0.2457 0.3235 0.3816 0.4681 0.4348 0.3546 0.2516 0.1741 0.1780
Local after 0.3944 0.3695 0.4586 0.5570 0.6159 0.5301 0.4340 0.3397 0.3213 0.2673

ADRs before 0.1144 0.2946 0.2881 0.3737 0.4337 0.3732 0.2715 0.1571 0.1437 0.1703
ADRS after 0.3390 0.2868 0.4070 0.5022 0.5973 0.4909 0.3949 0.2988 0.2711 0.2192
This table shows how correlations change as we move further into the tails. NA stands for not available, since in those
cases the parameters of the dependence function could not be estimated. All data was obtained form Datastream. All
figures are in dollars. Cutoff point for panel C is 12/29/1994. The equally weighted basket of ADRs was calculated
considering only the live issues.

Table 12. Extreme correlations using different number of tail exceedances.

117
 Date of Maximum Minimum Number of
Country
initial rating rating rating changes

Argentina 8/25/1993 BB SD 9
Brazil 11/30/1994 BB- B 6
Bulgaria 11/23/1998 BB B 4
Chile 12/7/1992 A- BBB 3
China 12/7/1992 BBB+ BBB 3
Colombia 6/21/1993 BBB- BB 3
Czech 7/28/1993 A- BBB 4
Egypt 1/15/1997 BBB- BB+ 2
Greece 12/7/1992 A BBB- 4
Hungary 4/20/1992 A- BB+ 5
Indonesia 12/7/1992 BBB SD 16
Jamaica 11/9/1999 B+ B 2
Kazakhstan 11/5/1996 BB BB- 4
Korea 10/1/1988 AA- B+ 11
Lebanon 2/26/1997 BB- B- 4
Malaysia 9/13/1990 A+ BBB- 8
Mexico 7/29/1992 BBB- BB 4
Pakistan 12/21/1994 B+ SD 8
Panama 1/22/1997 BB+ BB 2
Peru 12/18/1997 BB BB- 2
Philippines 6/30/1993 BB+ BB- 3
Poland 6/1/1995 BBB+ BB 4
Portugal 12/7/1992 A+ AA 3
Qatar 2/1/1996 A- BBB 3
Russia 10/7/1996 BB SD 11
Slovakia 2/15/1994 BBB BB- 6
South Africa 10/3/1994 BBB- BB 3
Thailand 6/14/1989 A BBB- 5
Turkey 4/28/1996 B+ B- 5
Ukraine 12/21/2001 B B 1
Uruguay 2/14/1994 BBB- CC 8
Venezuela 7/24/1991 BB CCC+ 8
This table shows a breakdown of sovereign rating changes issued by Standard and Poor’s.
S&P uses 24 different categories, going from D (lowest) to AAA (highest), with a “+” or “-“
to denote issuers above or below the mean in each category. Ratings above BBB- (inclusive)
are considered as investment grade. SD stands for 'Selective Default'. Date of initial rating is
the date when the country was first rated by Standard and Poor's. Max and min ratings are the
highest and lowest qualifications issued to that country from the date of initial rating to the
end of 2003.
Table 13. Sovereign rating changes by Standard & Poor's.

118
 Date of Maximum Minimum Number of
Country
initial rating rating rating changes

Argentina 11/18/1986 Ba3 Caa3 19

Brazil 11/18/1986 Ba1 B2 9
Bulgaria 9/27/1996 B1 B3 3
Chile 5/25/1999 Baa1 1
China 5/23/1988 A3 Baa1 6
Colombia 8/4/1993 Ba1 Ba2 7
Czech 6/22/1998 A1 Baa1 2
Egypt 7/6/2001 Ba1 1
Greece 5/24/1994 A1 Baa3 7
Hungary 12/27/1993 A1 Ba1 9
Indonesia 3/14/1994 Baa3 B3 4
Jamaica 3/30/1998 Ba3 1
Kazakhstan 11/11/1996 Baa3 B1 6
Korea 4/9/1998 A3 Ba1 6
Lebanon 2/26/1997 B1 B2 2
Malaysia 11/18/1986 A1 Baa3 14
Mexico 2/20/1991 Baa2 Ba2 8
Pakistan 11/23/1994 Ba3 Caa1 7
Panama 1/22/1997 Ba1 1
Peru 7/20/1999 Ba3 1
Philippines 7/1/1993 Ba1 Ba3 4
Poland 6/1/1995 A2 Baa3 3
Portugal 11/18/1986 Aa2 A1 5
Qatar 9/22/1999 A3 Baa2 2
Russia 11/22/1996 Ba2 B3 10
Slovakia 5/15/1995 A3 Ba1 4
South Africa 10/3/1994 Baa2 Baa3 3
Thailand 8/1/1989 A2 Ba1 8
Turkey 5/5/1992 Baa3 B1 9
Ukraine 2/6/1998 B2 Caa1 8
Uruguay 10/15/1993 Baa3 B3 7
Venezuela 6/3/1987 Ba1 Caa1 9
This table shows a breakdown of sovereign rating changes issued by Moody’s. Moody’s has
27 categories, ranging from a lowest rating of C to a highest rating of Aaa. Moody's adds the
number 1, 2, or 3 after the rating to signal an issuer above the mean, on the mean or below the
mean in each category, respectively. All ratings above Baa3 (inclusive) are considered as
investment grade rating. Date of initial rating is the date when the country was first rated by
Moody's. Max and min ratings are the highest and lowest qualifications issued to that country
from the date of initial rating to the end of 2003.

Table 14. Sovereign rating changes by Moody's.

119
 Panel A. Sovereign rating downgrades Panel B. Sovereign rating upgrades
Cumulative
Abnormal Abnormal Cumulative
Event day abnormal t-stat Event day t-stat
return return abnormal return
return
-5 -0.0010 -0.0010 -0.2721 -5 -0.0001 -0.0001 -0.0342
-4 -0.0031 -0.0040 -0.8707 -4 0.0000 -0.0001 0.0007
-3 -0.0095 -0.0136 -2.6985 -3 0.0030 0.0029 1.0155
-2 -0.0112 -0.0248 -3.1701 -2 -0.0034 -0.0005 -1.1558
-1 -0.0102 -0.0350 -2.8933 -1 -0.0026 -0.0031 -0.8640
0 -0.0072 -0.0422 -2.0330 0 0.0023 -0.0008 0.7572
1 -0.0007 -0.0429 -0.1850 1 -0.0014 -0.0023 -0.4817
2 0.0067 -0.0362 1.8822 2 0.0027 0.0005 0.9225
3 0.0048 -0.0314 1.3562 3 0.0007 0.0012 0.2435
120

4 0.0033 -0.0282 0.9219 4 -0.0029 -0.0016 -0.9571

5 0.0003 -0.0279 0.0793 5 -0.0016 -0.0033 -0.5410
Total number of events: 81 Total number of events: 57

Cumulative abnormal returns Cumulative abnormal returns

(-5, +5) (-1, +1) (-5, +5) (-1, +1)
CAR -0.0279 -0.0181 CAR -0.0033 -0.0018
t-stat -2.3768 -2.9510 t-stat -0.3300 -0.3398
This table presents results from an event study analysis conducted on datastream's local stock market indices. Returns were computed
using the market model with datastream's world market (TOTMKWD) as the market return and a (t-244, t-6) estimation window. The
t-statistic is computed as the ratio of the mean abnormal return to the estimated standard deviation from the time series of mean
abnormal returns. All returns are in U.S. dollars.

Table 15. Stock index results using Standard and Poor's ratings.
 Panel A. Sovereign rating downgrades Panel B. Sovereign rating upgrades
Cumulative
Abnormal Abnormal Cumulative
Event day abnormal t-stat Event day t-stat
return return abnormal return
return
-5 0.0038 0.0038 1.0324 -5 -0.0056 -0.0056 -1.2315
-4 -0.0143 -0.0104 -3.8333 -4 0.0024 -0.0032 0.5359
-3 0.0002 -0.0102 0.0521 -3 0.0057 0.0026 1.2607
-2 -0.0030 -0.0132 -0.8008 -2 0.0010 0.0036 0.2302
-1 -0.0082 -0.0214 -2.1876 -1 -0.0010 0.0026 -0.2304
0 -0.0002 -0.0216 -0.0652 0 0.0035 0.0061 0.7661
1 0.0016 -0.0200 0.4326 1 -0.0038 0.0023 -0.8270
2 0.0052 -0.0149 1.3825 2 -0.0036 -0.0013 -0.7957
121

3 -0.0010 -0.0159 -0.2772 3 -0.0025 -0.0038 -0.5419

4 -0.0003 -0.0162 -0.0776 4 -0.0013 -0.0051 -0.2764
5 0.0068 -0.0094 1.8262 5 -0.0048 -0.0099 -1.0575
Total number of events: 53 Total number of events: 45

Cumulative abnormal returns Cumulative abnormal returns

(-5, +5) (-1, +1) (-5, +5) (-1, +1)
CAR -0.0094 -0.0068 CAR -0.0099 -0.0013
t-stat -0.7586 -1.0508 t-stat -0.6535 -0.1681
This table presents results from an event study analysis conducted on datastream's local stock market indices. Returns were computed
using the market model with datastream's world market (TOTMKWD) as the market return and a (t-244, t-6) estimation window. The
t-statistic is computed as the ratio of the mean abnormal return to the estimated standard deviation from the time series of mean
abnormal returns. All returns are in U.S. dollars.

Table 16. Stock index results using Moody's ratings.

 Abnormal
Event day Cumulative abnormal return t-stat
return
-5 -0.0025 -0.0025 -0.4332
-4 -0.0021 -0.0046 -0.3671
-3 0.0022 -0.0024 0.3878
-2 0.0003 -0.0021 0.0484
-1 -0.0068 -0.0089 -1.1842
0 -0.0050 -0.0139 -0.8701
1 0.0078 -0.0061 1.3625
2 0.0045 -0.0016 0.7824
3 0.0037 0.0021 0.6429
4 0.0052 0.0073 0.8970
5 0.0027 0.0100 0.4699
Total number of events: 17

Cumulative abnormal returns

(-5, +5) (-1, +1)
CAR 0.0100 -0.0040
t-stat 0.5235 -0.3994
This table presents results from an event study analysis conducted on datastream's
local stock market indices. The event is the date when a country was rated for the first
time by either agency. We have 10 instances in which S&P was the initial rating
agency and 7 where Moody's was. Returns were computed using the market model
with datastream's world market (TOTMKWD) as the market return and a (t-244, t-6)
estimation window. The t-statistic is computed as the ratio of the mean abnormal
return to the estimated standard deviation from the time series of mean abnormal
returns. All returns are in U.S. dollars.

Table 17. Stock index results using initial ratings.

122
 No. of pairs
S&P Moody's First rating
of events

4/2/1997
1 S&P
10/2/1997

3/26/2001
2 S&P
3/28/2001

5/8/2001
3 S&P
6/4/2001

7/12/2001
4 S&P
7/13/2001

10/9/2001
5 S&P
10/12/2001

11/6/2001
6 S&P
12/20/2001

11/30/1994
7 S&P
11/30/1994

9/3/1998
8 Moody's
1/14/1999

10/16/2000
9 Moody's
1/3/2001

7/2/2002
10 S&P
8/12/2002

11/7/2001
11 S&P
12/19/2001

This table has the sequence of events identified as pairs of rating

changes for Argentina. We define a pair of events as any two rating
changes by both agencies that took place within a six month period.
The earliest announcement within each pair is what we consider the
first rating.
Table 18. First ratings for Argentina.

123
Panel A. Downgrades
S&P's downgrades Moody's downgrades
Cumulative
Cumulative Event Abnormal
Event day Abnormal return t-stat abnormal t-stat
abnormal return day return
return
-5 -0.0099 -0.0099 -1.6574 -5 0.0079 0.0079 1.3997
-4 -0.0170 -0.0269 -2.8341 -4 -0.0189 -0.0110 -3.3419
-3 -0.0115 -0.0384 -1.9180 -3 0.0144 0.0034 2.5381
-2 -0.0159 -0.0543 -2.6508 -2 -0.0002 0.0032 -0.0346
-1 -0.0113 -0.0655 -1.8808 -1 0.0043 0.0075 0.7592
0 -0.0058 -0.0713 -0.9647 0 -0.0037 0.0037 -0.6588
1 0.0059 -0.0654 0.9902 1 -0.0054 -0.0017 -0.9618
2 0.0134 -0.0520 2.2295 2 -0.0010 -0.0027 -0.1792
3 0.0039 -0.0482 0.6472 3 0.0042 0.0015 0.7488
4 -0.0179 -0.0660 -2.9800 4 0.0102 0.0117 1.8042
5 0.0060 -0.0600 1.0030 5 -0.0101 0.0016 -1.7839
Total number of events: 21 Total number of events: 19
Cumulative abnormal returns Cumulative abnormal returns
(-5, +5) (-1, +1) (-5, +5) (-1, +1)
CAR -0.0600 -0.0111 CAR 0.0016 -0.0049
t-stat -3.0199 -1.6711 t-stat 0.0874 -0.4973
Panel B. Upgrades
S&P's upgrades Moody's upgrades
Cumulative
Cumulative Event Abnormal
Event day Abnormal return t-stat abnormal t-stat
abnormal return day return
return
-5 -0.0038 -0.0038 -0.5489 -5 -0.0013 -0.0013 -0.2074
-4 -0.0037 -0.0075 -0.5444 -4 0.0071 0.0058 1.1435
-3 0.0038 -0.0037 0.5484 -3 0.0004 0.0062 0.0575
-2 -0.0117 -0.0154 -1.7052 -2 0.0029 0.0091 0.4663
-1 -0.0086 -0.0240 -1.2560 -1 0.0015 0.0106 0.2418
0 0.0043 -0.0197 0.6301 0 -0.0001 0.0105 -0.0148
1 0.0062 -0.0135 0.9032 1 0.0003 0.0108 0.0457
2 0.0071 -0.0064 1.0338 2 -0.0055 0.0052 -0.8889
3 0.0043 -0.0021 0.6272 3 0.0009 0.0062 0.1516
4 -0.0069 -0.0090 -1.0068 4 0.0003 0.0065 0.0477
5 0.0003 -0.0087 0.0447 5 -0.0001 0.0064 -0.0109
Total number of events: 17 Total number of events: 21
Cumulative abnormal returns Cumulative abnormal returns
(-5, +5) (-1, +1) (-5, +5) (-1, +1)
CAR -0.0087 0.0019 CAR 0.0064 0.0017
t-stat -0.3841 0.1601 t-stat 0.3112 0.1574
This table uses the events we identified as pairs of rating changes for all countries in our sample. We define a pair of
events as any two rating changes by both agencies that took place within a six month period. The earliest announcement
within each pair is what we consider the first rating. Results are from an event study analysis using stock market indices.
Returns were computed using the market model with datastream's world market (TOTMKWD) as the market return and a
(t-244, t-6) estimation window. The t-statistic is computed as the ratio of the mean abnormal return to the estimated
standard deviation from the time series of mean abnormal returns. All returns are in U.S. dollars.
Table19. Stock market reaction to the first rating by either agency.

124
Panel A: Sovereign credit downgrades Panel B: Sovereign credit upgrades

All firms All firms

CAR (-5, +5) -0.0830 CAR (-5, +5) -0.0030
z-stat (4.2709) z-stat (0.2798)
CAR (-1, +1) -0.0475 CAR (-1, +1) -0.0001
z-stat (4.6847) z-stat (0.0169)

No. of events 2,523 No. of events 2,804

ADRs No ADRs ADRs No ADRs

CAR (-5, +5) -0.0808 -0.0833 CAR (-5, +5) -0.0073 -0.0022
z-stat (4.1059) (4.0653) z-stat (0.5128) (0.2075)
CAR (-1, +1) -0.0466 -0.0477 CAR (-1, +1) 0.0031 -0.0006
z-stat (4.5315) (4.4516) z-stat 0.4166 (0.1021)

No. of events 337 2,186 No. of events 384 2,420

No international No international
International debt International debt
debt debt
CAR (-5, +5) -0.1223 -0.0718 CAR (-5, +5) -0.0122 -0.0013
z-stat (4.4765) (3.8425) z-stat (0.5378) (0.1526)
CAR (-1, +1) -0.0798 -0.0382 CAR (-1, +1) 0.0028 -0.0011
z-stat (5.5943) (3.9128) z-stat (0.2383) (0.2414)

No. of events 567 1,956 No. of events 616 2,188

No ADR and No ADR and

ADR and ADR and
international international
International debt International debt
debt debt

CAR (-5, +5) -0.1058 -0.0814 CAR (-5, +5) -0.0181 -0.0023
z-stat (4.0382) (4.1356) z-stat (0.7640) (0.2213)
CAR (-1, +1) -0.0793 -0.0456 CAR (-1, +1) 0.0034 -0.0003
z-stat (5.7954) (4.4323) z-stat (0.2711) (0.0599)

No. of events 144 2,379 No. of events 170 2,634

This table presents CARs for two splits of the original sample. First, we split our sample in firms that have American
Deposit Receipts (ADR) trading abroad and those who do not. ADR data is from Citibank and The Bank of New York.
We require firms to have an active ADR of level 1, 2 or 3 trading abroad at the time of the sovereign downgrade. The
second split is by firms that have internationally traded debt and those who do not. International debt data is from
Datastream. We require firms to have at least one outstanding foreign bond or eurobond at the time of the sovereign
downgrade. Returns were computed using the market model with datastream's world market (TOTMKWD) as the
market return. All returns are in U.S. dollars. Absolute value of z-statistics in parentheses, z-stats are computed using
the time series standard deviation of the market model residuals multiplied times the square root of the number of days
in the event window as the estimate for the abnormal returns' standard deviation.

Table 20. Cumulative Abnormal Returns (CAR) for Stocks with International Financing.

125
 Panel A: 3-day CAR (-1, +1) regressions for all stocks, sovereign downgrades, dollar returns
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Constant YES YES YES YES YES YES YES YES YES YES YES

ADR dummy 0.0099 -0.0657 -0.0821 -0.1004 -0.1585 -0.1649 -0.1838

(1.4876) (0.9943) (1.2439) (1.4888) (1.8107)* (1.9032)* (1.9701)**
International debt dummy -0.0107 0.0084 0.0535 0.0324 -0.0295 0.0322 0.0419
(1.9484)* (0.1581) (1.0386) (0.6001) (0.4085) (0.4498) (0.5429)
ADR * Size 0.0053 0.0064 0.0078 0.0120 0.0125 0.0139
(1.1792) (1.4207) (1.6818)* (1.9826)** (2.0761)** (2.1411)**
International debt * Size -0.0011 -0.0044 -0.0028 0.0018 -0.0027 -0.0034
(0.2919) (1.2461) (0.7533) (0.3539) (0.5328) (0.6373)
Log (GDP per capita)
0.4216 0.4664 0.4701 0.3343 0.3263 0.3664 0.3294
126

(6.0513)*** (5.4981)*** (5.5395)*** (2.7927)*** (2.7221)*** (3.5247)*** (2.7491)***

Stock market cap / GDP 0.0015 0.0017 0.0017 0.0011 0.0011 0.0013 0.0011
(5.3021)*** (5.1709)*** (5.2224)*** (2.4618)** (2.3959)** (3.3914)*** (2.4412)**
Size -0.0032 -0.0009 -0.0026 -0.0086 -0.0055 -0.0084 -0.0080
(1.7704)* (0.4802) (1.3318) (1.8790)* (1.2139) (1.9410)* (1.7214)*
Intangibles / Total assets -0.0162 -0.0190 -0.0129 -0.0165
(0.3324) (0.3895) (0.2755) (0.3386)
Revenues / Total assets 0.0053 0.0052 0.0054 0.0055
(1.4219) (1.3845) (1.5429) (1.4566)
Inventories / Total assets -0.0034 -0.0060 -0.0051 -0.0030
(0.1056) (0.1896) (0.1728) (0.0935)
Debt / Total assets 0.0239 0.0219 0.0145 0.0248
(1.5470) (1.4132) (1.0509) (1.5966)
continued...
 Cash / Assets 0.0352 0.0326 0.0341
(1.3339) (1.2321) (1.2904)
Observations 2,523 2,523 2,161 2,161 2,523 2,161 2,161 1,487 1,487 1,613 1,487
R-squared 24.70% 24.86% 26.11% 26.01% 25.96% 27.33% 27.39% 27.89% 27.59% 27.80% 27.96%

Panel B: 11 day CAR (-5, +5) regressions for all stocks, sovereign downgrades, dollar returns
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Constant YES YES YES YES YES YES YES YES YES YES YES

ADR dummy 0.0151 -0.1049 -0.1006 -0.1311 -0.1747 -0.2232 -0.2266

(1.5086) (1.0680) (1.0211) (1.3027) (1.3464) (1.7147)* (1.6399)
International debt dummy -0.0127 -0.0111 0.0584 0.0230 -0.0014 0.0850 0.0957
(1.5469) (0.1409) (0.7589) (0.2865) (0.0129) (0.7913) (0.8359)
ADR * Size
0.0085 0.0081 0.0103 0.0138 0.0173 0.0176
127

(1.2667) (1.1928) (1.4948) (1.5355) (1.9026)* (1.8344)*

International debt * Size 0.0007 -0.0045 -0.0018 -0.0004 -0.0066 -0.0075
(0.1233) (0.8546) (0.3320) (0.0505) (0.8832) (0.9377)
Log (GDP per capita) 0.4994 0.4813 0.4875 0.0503 0.0409 0.3748 0.0416
(4.7786)*** (3.8018)*** (3.8493)*** (0.2834) (0.2301) (2.4009)** (0.2344)
Stock market cap / GDP 0.0003 0.0001 0.0001 -0.0022 -0.0022 -0.0001 -0.0022
(0.8165) (0.1940) (0.2554) (3.2657)*** (3.3085)*** (0.2429) (3.2858)***
Size -0.0048 -0.0025 -0.0044 -0.0034 0.0014 -0.0038 -0.0023
(1.771299)* (0.8654) (1.4891) (0.5060) (0.2068) (0.5805) (0.3335)
Intangibles / Total assets -0.0206 -0.0262 -0.0134 -0.0219
(0.2857) (0.3619) (0.1904) (0.3030)
Revenues / Total assets -0.0007 -0.0008 0.0003 -0.0003
(0.1188) (0.1490) (0.0524) (0.0600)
Inventories / Total assets 0.0492 0.0438 0.0412 0.0495
(1.0454) (0.9317) (0.9352) (1.0513)
continued...
 Table 21 continued
Debt / Total assets 0.0314 0.0283 0.0138 0.0325
(1.3685) (1.2321) (0.6669) (1.4151)
Cash / Assets 0.0500 0.0466 0.0481
(1.2779) (1.1876) (1.2266)
Observations 2,523 2,523 2,161 2,161 2,523 2,161 2,161 1,487 1,487 1,613 1,487
R-squared 15.29% 15.42% 16.84% 16.66% 16.20% 17.66% 17.75% 21.56% 21.27% 20.05% 21.66%

The dependent variable is the cumulative abnormal return (CAR) computed using market model adjusted returns á la Brown and Warner (1985). All returns are in U.S. dollars. All
figures used on the right-hand side of the regressions are converted to U.S. dollars using then prevailing exchange rates collected from Datastream. All items except GDP per capita and
stock market cap / GDP are from Worldscope. GDP per capita and stock market cap / GDP are from the World Bank's Economic Indicators database. Size is Total assets from
Worldscope. The following are the Worldscope codes for our variables: Intangible assets (02649), Total assets (02999), Tota debt (03255), Total revenues (07240), Cash plus short term
investments (02001). ADR dummy is a dummy variable that takes a value of one when the firm has an ADR issue level 1, 2 or 3. International debt dummy is a dummy variable that
takes a value of one when the firm has at least a foreign bond or an eurobond issue outstanding. We control for country fixed effects using country dummies (not reported). Absolute
value of t-statistics in parentheses. We correct for heteroskedasticity using White's (1980) covariance estimator. * significant at 10%; ** significant at 5%; *** significant at 1%.

Table 21. Cumulative Abnormal Returns (CAR) for all stocks following a sovereign rating downgrade
128
 Panel A: 3-day CAR (-1, +1) regressions for all stocks, sovereign upgrades, dollar returns
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Constant YES YES YES YES YES YES YES YES YES YES YES

ADR dummy -0.0026 0.0023 0.0133 0.0068 0.0213 0.0271 0.0311

(0.8060) (0.0765) (0.4412) (0.2222) (0.5990) (0.7216) (0.8537)
International debt dummy 0.0040 0.0001 0.0093 0.0017 -0.0164 -0.0293 -0.0213
(1.4296) (0.0060) (0.4190) (0.0752) (0.5905) (1.0266) (0.7517)
ADR * Size -0.0001 -0.0009 -0.0004 -0.0014 -0.0016 -0.0021
(0.0514) (0.4353) (0.1985) (0.5607) (0.6418) (0.8467)
International debt * Size 0.0002 -0.0005 0.0001 0.0015 0.0022 0.0018
(0.1309) (0.3252) (0.0523) (0.8020) (1.1512) (0.9620)
129

Log (GDP per capita) 0.0428 0.0445 0.0461 -0.0462 -0.0460 0.0473 -0.0461
(2.5468)** (2.4579)** (2.5379)** (1.8766)* (1.8731)* (2.2399)** (1.8742)*
Stock market cap / GDP -0.0004 -0.0004 -0.0004 0.0003 0.0004 -0.0005 0.0003
(7.9173)*** (7.7554)*** (7.7179)*** (2.1044)** (2.1960)** (8.1374)*** (2.1494)**
Size -0.0008 -0.0011 -0.0010 0.0027 0.0019 -0.0011 0.0021
(0.9591) (1.2226) (1.0835) (1.1843) (0.8274) (0.4643) (0.9100)
Intangibles / Total assets -0.0087 -0.0073 -0.0075 -0.0074
(0.4910) (0.4098) (0.4123) (0.4175)
Revenues / Total assets -0.0038 -0.0039 -0.0003 -0.0039
(1.8439)* (1.8938)* (0.1441) (1.8839)*
Inventories / Total assets 0.0178 0.0185 -0.0006 0.0182
(1.2028) (1.2530) (0.0450) (1.2321)
continued...
 Debt / Total assets -0.0056 -0.0057 -0.0087 -0.0058
(0.6757) (0.6829) (1.1201) (0.6977)
Cash / Assets -0.0023 -0.0012 -0.0010
(0.1792) (0.0948) (0.0789)
Observations 2,804 2,804 2,280 2,280 2,804 2,280 2,280 1,459 1,459 1,736 1,459
R-squared 8.15% 8.23% 10.01% 10.05% 10.24% 12.44% 12.49% 14.72% 14.89% 15.33% 14.94%

Panel B: 11 day CAR (-5, +5) regressions for all stocks, sovereign upgrades, dollar returns
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Constant YES YES YES YES YES YES YES YES YES YES YES

ADR dummy -0.0015 -0.0296 -0.0001 -0.0125 0.0281 0.0318 0.0450

(0.2728) (0.5882) (0.0016) (0.2480) (0.4743) (0.5111) (0.7426)
International debt dummy 0.0063 -0.0125 0.0033 -0.0112 -0.0414 -0.0321 -0.0480
(1.3791) (0.3290) (0.0910) (0.2932) (0.8972) (0.6767) (1.0181)
130

ADR * Size 0.0024 0.0002 0.0011 -0.0016 -0.0018 -0.0029

(0.7024) (0.0592) (0.3238) (0.4017) (0.4148) (0.6913)
International debt * Size 0.0013 -0.0001 0.0010 0.0032 0.0024 0.0036
(0.5032) (0.0253) (0.3965) (1.0330) (0.7474) (1.1441)
Log (GDP per capita) -0.1343 -0.1611 -0.1581 -0.2361 -0.2370 -0.1346 -0.2362
(4.9031)*** (5.4098)*** (5.2895)*** (5.7641)*** (5.7991)*** (3.8397)*** (5.7655)***
Stock market cap / GDP -0.0006 -0.0007 -0.0007 0.0002 0.0003 -0.0008 0.0002
(7.0936)*** (7.1314)*** (7.0891)*** (0.8759) (0.9498) (6.9547)*** (0.8944)
Size -0.0024 -0.0026 -0.0019 0.0071 0.0060 0.0001 0.0062
(1.7488)* (1.8166)* (1.2684) (1.8513)* (1.5758) (0.0245) (1.5930)
Intangibles / Total assets -0.0115 -0.0100 -0.0015 -0.0100
(0.3913) (0.3389) (0.0504) (0.3376)
Revenues / Total assets -0.0092 -0.0093 -0.0026 -0.0092
(2.6557)*** (2.6823)*** (0.7913) (2.6636)***
continued...
 Table 22 continued
Inventories / Total assets 0.0446 0.0459 0.0030 0.0455
(1.8091)* (1.8614)* (0.1280) (1.845929)*
Debt / Total assets 0.0296 0.0301 0.0083 0.0295
(2.1267)** (2.1661)** (0.6485) (2.1195)**
Cash / Assets 0.0247 0.0258 0.0259
(1.1590) (1.2067) (1.2141)
Observations 2,804 2,804 2,280 2,280 2,804 2,280 2,280 1,459 1,459 1,736 1,459
R-squared 6.26% 6.33% 7.59% 7.61% 9.10% 11.02% 11.08% 11.20% 11.29% 12.06% 11.34%

The dependent variable is the cumulative abnormal return (CAR) computed using market model adjusted returns á la Brown and Warner (1985). All returns are in U.S. dollars. All figures used on
the right-hand side of the regressions are converted to U.S. dollars using then prevailing exchange rates collected from Datastream. All items except GDP per capita and stock market cap / GDP
are from Worldscope. GDP per capita and stock market cap / GDP are from the World Bank's Economic Indicators database. Size is Total assets from Worldscope. The following are the
Worldscope codes for our variables: Intangible assets (02649), Total assets (02999), Tota debt (03255), Total revenues (07240), Cash plus short term investments (02001). ADR dummy is a
dummy variable that takes a value of one when the firm has an ADR issue level 1, 2 or 3. International debt dummy is a dummy variable that takes a value of one when the firm has at least a
foreign bond or an eurobond issue outstanding. We control for country fixed effects using country dummies (not reported). Absolute value of t-statistics in parentheses. We correct for
heteroskedasticity using White's (1980) covariance estimator. * significant at 10%; ** significant at 5%; *** significant at 1%.

Table 22. Cumulative Abnormal Returns (CAR) for all stocks following a sovereign rating upgrade
131
Table 23. Countries included in this
comparison

Argentina El Salvador Philippines

Brazil Greece Poland
Bulgaria Hungary Portugal
Chile Indonesia Qatar
China Jordan Russia
Colombia Korea Slovakia
South
Costa Rica Lebanon Africa
Croatia Malaysia Thailand
Czech Mexico Turkey
Dominican Pakistan Venezuela
Ecuador Panama Vietnam
Egypt Peru

132
Table 24. Coverage for sovereign bonds on Datastream and Warga databases.

Warga database Datastream

Naic303 Naic304 Naic99303 Naic99304

Total number of different

bonds for 37 emerging market 73 58 88 89 276
countries with valid ISIN data

Warga files only

Bonds included only in NAIC

33 23 33 33
files

Bonds included in Datastream 40 35 55 56

and NAIC

Observations with usable data 552 221 844 745

from BOTH databases

Price correlation 92.33% 97.24% 93.88% 94.70%

Holding period return

73.66% 89.83% 74.88% 75.86%
correlation

HPR one month or less 75.01% 93.64% 84.61% 64.21%

133
APPENDIX C

FIGURES

134
 -6
-5
-4
-3
-2
-1
0
1
2
3
-6
-4
-2
0
2
4
6
8
10
12
6/1/1997 6/1/1997

9/1/1997 9/1/1997

12/1/1997 12/1/1997

3/1/1998 3/1/1998

6/1/1998 6/1/1998

9/1/1998 9/1/1998

12/1/1998 12/1/1998

3/1/1999 3/1/1999

6/1/1999 6/1/1999

9/1/1999 9/1/1999

Sovereign

135
Sovereign
12/1/1999 12/1/1999

3/1/2000 3/1/2000

6/1/2000 6/1/2000
Domestic

Domestic
9/1/2000 9/1/2000
Figure 1. First common component

Figure 2. Second common component

12/1/2000 12/1/2000

3/1/2001 3/1/2001

6/1/2001 6/1/2001

9/1/2001 9/1/2001

12/1/2001 12/1/2001

3/1/2002 3/1/2002

6/1/2002 6/1/2002
Figure 3. Q-Q Plots for the left tail and the right tail of the dollar return of the Mexican
equity index

8
6
Exponential Quantiles
4
2

-0.15 -0.10 -0.05 0.0 0.05 0.10 0.15

Ordered Data
8
6
Exponential Quantiles

4
2

-0.15 -0.10 -0.05 0.0 0.05 0.10 0.15

Ordered Data

136
Figure 4. Q-Q Plots for the left tail and the right tail of the dollar return of the Mexican
ADR equally weighted portfolio

8
6
Exponential Quantiles
4
2

-0.2 -0.1 0.0 0.1

Ordered Data
8
6
Exponential Quantiles

4
2

-0.1 0.0 0.1 0.2

Ordered Data

137
Figure 5. Q-Q Plots for the left tail and the right tail of the dollar return of the S&P 500
equity index

8
6
Exponential Quantiles

4
2

-0.10 -0.05 0.0 0.05 0.10 0.15

Ordered Data
8
6
Exponential Quantiles

4
2

-0.15 -0.10 -0.05 0.0 0.05 0.10

Ordered Data

138
Figure 6. Excess mean graphs for the left tail and the right tail of the dollar return of the
Mexican equity index

0.25
0.20
Mean Excess

0.15
0.10
0.05

-0.2 -0.1 0.0 0.1

Threshold
0.15
Mean Excess
0.10
0.05

-0.15 -0.10 -0.05 0.0 0.05 0.10

Threshold

139
Figure 7. Excess mean graphs for the left tail and the right tail of the dollar return of the
Mexican ADR equally weighted portfolio.

0.25
0.20
Mean Excess

0.15
0.10
0.05

-0.2 -0.1 0.0 0.1

Threshold
0.15
Mean Excess

0.10
0.05

-0.15 -0.10 -0.05 0.0 0.05 0.10

Threshold

140
Figure 8. Excess mean graphs for the left tail and the right tail of the dollar return of the
S&P 500 equity index

0.14
0.12
0.10
Mean Excess
0.08
0.06
0.04
0.02

-0.10 -0.05 0.0 0.05

Threshold
0.15
Mean Excess

0.10
0.05

-0.15 -0.10 -0.05 0.0 0.05

Threshold

141
Figure 9. Correlation between S&P and the Mexican stock market index (solid line) and correlation between S&P
and Mexican ADRs (dashed line). The number of exceedances used to calculate each correlation is shown in the
horizontal axis.

Figure 10. Correlation between S&P and the Chilean stock market index (solid line) and correlation between S&P
and Chilean ADRs (dashed line). The number of exceedances used to calculate each correlation is shown in the
horizontal axis.

142
Figure 11. Correlation between S&P and the Venezuelan stock market index (solid line) and correlation between
S&P and Venezuelan ADRs (dashed line). The number of exceedances used to calculate each correlation is shown
in the horizontal axis.

Figure 12. Correlation between S&P and the Colombian stock market index (solid line) and correlation between
S&P and Colombian ADRs (dashed line). The number of exceedances used to calculate each correlation is shown
in the horizontal axis.

143
Figure 13. Correlation between S&P and the Brazilian stock market index (solid line) and correlation between
S&P and Brazilian ADRs (dashed line). The number of exceedances used to calculate each correlation is shown in
the horizontal axis.

Figure 14. Correlation between S&P and the Argentinean stock market index (solid line) and correlation between
S&P and Argentinean ADRs (dashed line). The number of exceedances used to calculate each correlation is
shown in the horizontal axis.

144
Figure 15. Correlation between S&P and the Mexican stock market index (solid line) and correlation between
S&P and Mexican ADRs (dashed line) before the 1995 Mexican crisis. The number of exceedances used to
calculate each correlation is shown in the horizontal axis.

Figure 16. Correlation between S&P and the Mexican stock market index (solid line) and correlation between
S&P and Mexican ADRs (dashed line) after the 1995 Mexican crisis. The number of exceedances used to
calculate each correlation is shown in the horizontal axis.

145
 Figure 17. Downgrades (S&P)
0.01

0
-5 -4 -3 -2 -1 0 1 2 3 4 5
-0.01

-0.02

-0.03

-0.04

-0.05

Abnormal return Cumulative abnormal return

Figure 18. Upgrades (S&P)

0.004

0.003

0.002

0.001

0
-5 -4 -3 -2 -1 0 1 2 3 4 5
-0.001

-0.002

-0.003

-0.004
Abnormal return Cumulative abnormal return

146
 Figure 19. Downgrades (Moody's)
0.01

0.005

0
-5 -4 -3 -2 -1 0 1 2 3 4 5
-0.005

-0.01

-0.015

-0.02

-0.025
Abnormal return Cumulative abnormal return

Figure 20. Upgrades (Moody's)

0.008
0.006
0.004
0.002
0
-0.002 -5 -4 -3 -2 -1 0 1 2 3 4 5
-0.004
-0.006
-0.008
-0.01
-0.012
Abnormal return Cumulative abnormal return

147