TRIUM Exam Question
Professor Edward Altman
June 2019
In August 2017, Tesla Motors raised $1.8 billion in corporate bonds,
priced at 5.3%, 8-year notes. The bonds were subordinated to more senior
debt and received a B- rating from S&P and a B3 rating from Moody’s.
(1) Using the credit analytics discussed in our class, and traditional
metrics, what does your group think should be Tesla’s bond rating
before and just after the new bond issue?
(2) Would your answer change if the firm raised an additional $3 billion in
bonds to meet production objectives?
(3) What is your estimate of the Bond Rating Equivalent (BRE) as of the
most recent (Q1-2019) financials and latest (June 26, 2019) stock
price?
(4) Given your answers #1 and #2 above, what are your expected
cumulative PD (Probability of Default) and LGD (Loss Given Default)
for Tesla for one-year and five (5) years?
(5) What are the bonds issued in 2017 now (June 26, 2019) selling at
and what is the bond’s yield to maturity?
(6) Which of the two Z-Score models (Z or Z”) is most applicable to a firm
like Tesla? Why?
(7) What are the main differences between Z and Z”? List up to four
differences.
Please work on this question with no less than two (2) members and no
more than three (3). Do not confer with other groups with respect to any
aspect of this question. Limit your answer to no more than three (3) pages,
(double-spaced) plus Exhibits (no Appendices). Please make sure the
typeface is clear and large enough so it is easily legible.
dZ/hD'>K>D
ůĂƐƐŽĨϮϬϮϬ
^ƚƌĂƚĞŐLJĂŶĚ&ŝŶĂŶĐĞĨŽƌ'ůŽďĂůdžĞĐƵƚŝǀĞƐ
EĞǁzŽƌŬ
^ƵŶĚĂLJϭϲʹ&ƌŝĚĂLJϮϴ:ƵŶĞϮϬϭϵ
ƌĞĚŝƚZŝƐŬŽƵƌƐĞWĂĐŬ
WƌĞͲDŽĚƵůĞZĞƋƵŝƌĞĚZĞĂĚŝŶŐ
TRIUM Class of 2020: Module 4 − Pack 3: Credit
Risk
TRIUM Global Executive MBA
THIS PRINT COURSEPACK AND ITS ELECTRONIC COUNTERPART (IF ANY) ARE INTENDED SOLELY FOR THE PERSONAL USE
OF PURCHASER. ALL OTHER USE IS STRICTLY PROHIBITED.
XanEdu™ publications may contain copyrighted materials of XanEdu, Inc. and/or its licensors.
The original copyright holders retain sole ownership of their materials. Copyright permissions
from third parties have been granted for materials for this publication only. Further reproduction
and distribution of the materials contained herein is prohibited.
WARNING: COPYRIGHT INFRINGEMENT IS AGAINST THE LAW AND WILL RESULT IN
PROSECUTION TO THE FULLEST EXTENT OF THE LAW.
THIS COURSE PACK CANNOT BE RESOLD, COPIED
OR OTHERWISE REPRODUCED.
XanEdu Publishing, Inc. does not exert editorial control over materials that are included in this
course pack. The user hereby releases XanEdu Publishing, Inc. from any and all liability for any
claims or damages, which result from any use or exposure to the materials of this course pack.
Items are available in both online and in print, unless marked with icons.
− Print only
− Online only
TRIUM Class of 2020: Module 4 − Pack 3: Credit Risk
Table of Contents
“Toward a Botton−Up Approach to Assessing Sovereign Default Risk: An
Update” by Altman, Edward I; Rijken, Herbert
1
“The Fate of the Euro: It is Still Italia!” by Altman, Edward I
31
“Defaults and Returns in the High−Yield Bond and Distressed Debt Market:
The Year 2018 in Review and Outlook” by Altman, Edward I; Kuehne,
Brenda J
37
“The Investment Performance and Market Dynamics of Defaulted and
Distressed Corporate Bonds and Bank Loans: 2018 Review and 2019
Outlook” by Altman, Edward I; Kuehne, Brenda J
116
“A 50−Year Retrospective on Credit Risk Models, the Altman Z−Score
Family of Models and Their Applications to Financial Markets and
Managerial Strategies” by Altman, Edward I
169
i
Toward A Bottom-Up Approach to Assessing Sovereign Default Risk: An Update*
Edward I. Altman, New York University Stern School of Business, Herbert Rijken, Vrije
University **
Abstract
We propose a totally new approach toward assessing sovereign risk
by examining rigorously the health and aggregate default risk of a nation’s
private corporate sector. Models such as our new Z-Metrics™ approach can
be utilized to measure the median probability of default of the non-financial
sector cumulatively for five years, both as an absolute measure of corporate
risk vulnerability and a relative measure compared to other sovereigns and
to the market’s assessment via the now liquid credit-default-swap market.
Specifically, we measure the default probabilities of listed corporate entities
in nine European countries, and the U.S.A., as of 2009 and 2010. These
periods coincide with the significant rise in concern with sovereign default
risk in the Euro country sphere. We conclude that our corporate health
index of the private sector measured at periods prior to the explicit
recognition by most credit professionals, not only gave an effective early
warning indicator but provided a mostly appropriate hierarchy of relative
sovereign risk. Policy officials should, we believe, nurture, not penalize, the
tax revenue paying and jobs generating private sector when considering
austerity measures of distressed sovereigns.
=========================================================
Key Words: Sovereign Risk, Financial Crisis, Default Probability, Z-Metrics
JEL classification: F34, F36
*
This is an updated version of the article originally published in The Journal of Applied Corporate Finance, vol.23,
No. 3, Winter, 2011.
**The authors would like to thank Dan Balan and Matthew Watt of RiskMetrics Group, a subsidiary of MSCI, Inc.,
for computational assistance, and Brenda Kuehne of the NYU Salomon Center for her research assistance.
1
1
During the past four years, bank executives, government officials, and many others have been
sharply criticized for failing to anticipate the global financial crisis. The speed and depth of the
market declines shocked the public. And no one seemed more surprised than the credit rating
agencies that assess the default risk of sovereign governments as well as corporate issuers
operating within their borders.
Although the developed world had suffered numerous recessions in the past 150 years,
this most recent international crisis raised grave doubts about the ability of major banks and even
sovereign governments to honor their obligations. Several large financial institutions in the U.S.
and Europe required massive state assistance to remain solvent, and venerable banks like
Lehman Brothers even went bankrupt. The cost to the U.S. and other sovereign governments of
rescuing financial institutions believed to pose “systemic” risk was so great as to result in a
dramatic increase in their own borrowings.
The general public in the U.S. and Europe found these events particularly troubling
because they had assumed that elected officials and regulators were well-informed about
financial risks and capable of limiting serious threats to their investments, savings, and pensions.
High-ranking officials, central bankers, financial regulators, ratings agencies, and senior bank
executives all seemed to fail to sense the looming financial danger.
This failure seemed even more puzzling because it occurred years after the widespread
adoption of advanced risk management tools. Banks and portfolio managers had long been using
quantitative risk management tools such as Value at Risk (“VaR”). And they should also have
benefited from the additional information about credit risk made publicly available by the new
market for credit default swaps (“CDS”).
2
2
But, as financial market observers have pointed out, VaR calculations are no more
reliable than the assumptions underlying them. Although such assumptions tend to be informed
by statistical histories, critical variables such as price volatilities and correlations are far from
constant and thus difficult to capture in a model. The market prices of options—or of CDS
contracts, which have options “embedded” within them—can provide useful market estimates of
volatility and risk. And economists have found that CDS prices on certain kinds of debt
securities increase substantially before financial crises become full-blown. But because there is
so little time between the sharp increase in CDS prices and the subsequent crisis, policy makers
and financial managers typically have little opportunity to change course.1
Most popular tools for assessing sovereign risk are effectively forms of “top-down”
analysis. For example, in evaluating particular sovereigns, most academic and professional
analysts use macroeconomic indicators such as GDP growth, national debt-to-GDP ratios, and
trade and budget deficits as gauges of a country’s economic strength and well-being. But, as the
recent Euro debt crisis has made clear, such “macro” approaches, while useful in some settings
and circumstances, have clear limitations
In this paper, we present a totally new method for assessing sovereign risk, a type of
“bottom-up” approach that focuses on the financial condition and profitability of an economy’s
private sector. The assumption underlying this approach is that the fundamental source of
national wealth, and of the financial health of sovereigns, is the economic output and
productivity of their companies. To the extent we are correct, such an approach could provide
financial professionals and policy makers with a more effective means of anticipating financial
1
See, for example, Hekran Neziri’s “Can Credit Default Swaps predict Financial Crises?” in the Spring 2009
Journal of Applied Economic Sciences, Volume IV/Issue 1(7). Neziri found that CDS prices had real predictive
power for equity markets, but that the lead time was generally on the order of one month.
3
3
trouble, thereby enabling them to understand the sources of problems before they become
unmanageable.
In the pages that follow, we introduce Z-Metrics™, as a practical and effective tool for
estimating sovereign risk. Developed in collaboration with the Risk Metrics Group, now a
subsidiary of MSCI, Inc., Z-Metrics is a logical extension of the Altman Z-Score technique that
was introduced in 1968 and has since achieved considerable scholarly and commercial success.
Of course, no method is infallible, or represents the best fit for all circumstances. But by
focusing on the financial health of private enterprises in different countries, our system promises
at the very least to provide a valuable complement to, or reality check on, standard “macro”
approaches.
But before we delve into the details of Z-Metrics, we start by briefly reviewing the
record of financial crises to provide some historical perspective. Next we attempt to summarize
the main findings of the extensive academic and practitioner literature on sovereign risk,
particularly those studies designed to test the predictability of sovereign defaults and crises.
With that as background, we then present our new Z-Metrics system for estimating the
probability of default for individual (non-financial) companies and show how that system might
have been used to anticipate many developments during the current EU debt crisis. In so doing,
we make use of the most recent (2009 and 2010) publicly available corporate data for nine
European countries, both to illustrate our model’s promise for assessing sovereign risk and to
identify scope of reforms that troubled governments must consider not only to qualify for
bailouts and subsidies from other countries and international bodies, but to stimulate growth in
their economies.
4
4
More specifically, we examine the effectiveness of calculating the median company fiveyear probability of default of the sovereign’s non-financial corporate sector, both as an absolute
measure of corporate risk vulnerability and a relative health index comparison among a number
of European sovereigns, and including the U.S. as well. Our analysis shows that this health
index, measured at periods prior to the explicit recognition of the crisis by market professionals,
not only gave a distinct early warning of impending sovereign default in some cases, but also
provided a sensible hierarchy of relative sovereign risk. We also show that, during the current
European crisis, our measures not only compared favorably to standard sovereign risk measures,
notably credit ratings, but performed well even when compared to the implied default rates built
into market pricing indicators such as CDS spreads (while avoiding the well-known volatility of
the latter).
Our aim here is not to present a “beauty contest” of different methods for assessing
sovereign risk in which one method emerges as the clear winner. What we are suggesting is that
a novel, bottom-up approach that emphasizes the financial condition and profitability of a
nation’s private sector can be effectively combined with standard analytical techniques and
market pricing to better understand and predict sovereign health. And our analysis has one clear
implication for policy makers: that the reforms now being contemplated should be designed, as
far as possible, to preserve the efficiency and value of a nation’s private enterprises.
Modern History Sovereign Crises
When thinking about the most recent financial crisis, it is important to keep in mind how
common sovereign debt crises have been during the last 150 years—and how frequently such
debacles have afflicted developed economies as well as emerging market countries. Figure 1
5
5
shows a partial list of financial crises (identified by the first year of the crisis) that have occurred
in “advanced” countries. Overall, Latin America seems to have had more recent bond and loan
defaults than any other region of the world (as can be seen in Figure 2). But if we had included a
number of now developed Asian countries among the “advanced” countries, the period 19971999 period would be much more prominent.
FIGURE 1
Financial Crises, Advanced Countries 1870-2010
Crisis events (first year)
Austria
Brazil
Canada
Czechoslovakia
China
Denmark
DEU
GBR
Greece
Italy
Japan
Netherlands
Norway
Russia
Spain
Sweden
USA
1893,
1898,
1873,
1870,
1921,
1877,
1880,
1890,
1870,
1887,
1942
1897,
1899,
1918,
1920,
1876,
1873,
1989
1902,
1906,
1910,
1939
1885,
1891,
1974,
1894,
1891,
1921,
1921,
1998
1924,
1897,
1884,
1914, 1931, 1939
1923, 1983
1931, 2008
1902,
1901,
1984,
1932,
1907,
1907,
1931,
1991,
2009
1931,
1921, 1931, 1987
2008
2007
1930, 1935, 1990
1939
1931, 1988
1931, 1978, 2008
1907, 1922, 1931, 1991
1893, 1907, 1929, 1984, 2008
Source: IMF Global Financial Stability Report (2010), Reinhart and Rogoff (2010), and various other
sources, such as S&P’s economic reports.
6
6
Source: Compilation by Ingo Walter, NYU Stern School of Business
The clear lesson from Figures 1 and 2 is that sovereign economic conditions appear to
spiral out of control with almost predictable regularity and then require massive debt
restructurings and/or bailouts accompanied by painful austerity programs. Recent examples
include several Latin American countries in the 1980s, Southeast Asian nations in the late 1990s,
Russia in 1998, and Argentina in 2000. In most of those cases, major problems originating in
individual countries not only imposed hardships on their own people and markets, but had major
financial consequences well beyond their borders. We are seeing such effects now as financial
problems in Greece and other southern European countries not only affect their neighbors, but
threaten the very existence of the European Union.
Such financial crises have generally come as a surprise to most people, including even
those specialists charged with rating the default risk of sovereigns and the enterprises operating
7
7
in these suddenly threatened nations. For example, it was not long ago that Greek debt was
investment grade, and Spain was rated Aaa as recently as June 2010.2 And this pattern has been
seen many times before. To cite just one more case, South Korea was viewed in 1996 as an
“Asian Tiger” with a decade-long record of remarkable growth and an AA- rating. Within a year
however, the country was downgraded to BB-, a “junk” rating, and the county’s government
avoided default only through a $50 billion bailout by the IMF. And it was not just the rating
agencies that were fooled; most of the economists at the brokerage houses also failed to see the
problems looming in Korea.
What Do We Know about Predicting Sovereign Defaults?
There is a large and growing body of studies on the default probability of sovereigns, by
practitioners as well as academics.3 A large number of studies, starting with Frank and Cline’s
1971 classic, have attempted to predict sovereign defaults or rescheduling using statistical
classification and predicting methods like discriminant analysis as well as similar econometric
techniques.4 And in a more recent development, some credit analysts have begun using the
“contingent claim” approach5 to measure, analyze, and manage sovereign risk based on Robert
Merton’s classic “structural” approach (1974). But because of its heavy reliance on market
2
On April 27, 2010, Standard & Poor’s Ratings Services lowered its long- and short-term credit ratings on the
Hellenic Republic (Greece) to non-investment grade BB+; and on June 14, 2010, Moody’s downgraded Greece debt
to Ba1 from A2 (4 notches), while Spain was still Aaa and Portugal was A1. Both of the latter were recently
downgraded. S&P gave similar ratings.
3
One excellent primer on sovereign risk is Babbel’s (1996) study, which includes an excellent annotated
bibliography by S. Bertozzi on external debt capacity that describes many of these studies. Babbel lists 69
potentially helpful explanatory factors for assessing sovereign risk, all dealing with either economic, financial,
political, or social variables. Except for the political and social variables, all others are macroeconomic data and this
has been the standard until the last few years. Other work worth citing include two practitioner reports—Chambers
(1997) and Beers et al (2002)—and two academic studies—Smith and Walter (2003), and Frenkel, Karmann and
Scholtens (2004). Full citations of all studies can be found in References section at the end of the article.
4
Including Grinols (1976), Sargen (1977), Feder and Just (1977), Feder, Just and Ross (1981), Cline (1983),
Schmidt (1984), and Morgan (1986).
5
Gray, Merton and Bodie (2006, 2007)
8
8
indicators, this approach to predicting sovereign risk and credit spreads has the drawback of
producing large—and potentially self-fulfilling—swings in assessed risk that are attributable
solely to market volatility.
A number of recent studies have sought to identify global or regional common risk
factors that largely determine the level of sovereign risk in the world, or in a region such as
Europe. Some studies have shown that changes in both the risk factor of individual sovereigns
and in a common time-varying global factor affect the market’s repricing of sovereign risk.6
Other studies, however, suggest that sovereign credit spreads are more related to global
aggregate market indexes, including U.S. stock and high-yield bond market indexes, and global
capital flows than to their own local economic measures.7 Such evidence has been used to
justify an approach to quantifying sovereign risk that uses the local stock market index as a
proxy for the equity value of the country.8 Finally, several very recent papers focus on the
importance of macro variables such as debt service relative to tax receipts and the volatility of
trade deficits in explaining sovereign risk premiums and spreads.9
A number of studies have also attempted to evaluate the effectiveness of published credit
ratings in predicting defaults and expected losses, with most concluding that sovereign ratings,
especially in emerging markets, provide an improved understanding of country risks for
6
See Baek, Bandopadhyaya and Chan (2005). Gerlach, Schulz and Wolff (2010) observe that aggregate risk factors
drive banking and sovereign market risk spreads in the Euro area; and in a related finding, Sgherri and Zoli (2009)
suggest that Euro area sovereign risk premium differentials tend to move together over time and are driven mainly
by a common time-varying factor.
7
See Longstaff, Pan, Pedersen and Singleton (2007).
8
See Oshino and Saruwatari (2005).
9
These include Haugh, Ollivaud and Turner’s (2009) discussion of debt service relative to tax receipts in the Euro
area; Hilscher and Nobusch (2010) emphasis on the volatility of terms of trade; and Segoviano, Caceres and
Guzzo’s (2010) analysis of debt sustainability and the management of a sovereign’s balance sheet.
9
9
investment analytics.10 Nevertheless, the recent EU debt crisis would appear to contradict such
findings by taking place at a time when all the rating agencies and, it would seem, all available
models for estimating sovereign risk indicated that Greece and Spain—and others now
recognized as high-risk countries—were still classified as investment grade.11 What’s more,
although most all of the studies cited above have been fairly optimistic about the ability of their
concepts to provide early warnings of major financial problems, their findings have either been
ignored or have proven ineffective in forecasting most economic and financial crises.
In addition to these studies, a handful or researchers have taken a somewhat different
“bottom-up” approach by emphasizing the health of the private sectors supporting the
sovereigns. For example, a 1998 World Bank study of the 1997 East Asian crisis12 used the
average Z-Score of listed (non-financial) companies to assess the “financial fragility” of eight
Asian countries and, for comparison purposes, three developed countries and Latin America.
Surprising many observers, the average Z-Score for South Korea at the end of 1996 suggested
that it was the most financially vulnerable Asian country, followed by Thailand, Japan, and
Indonesia. As noted earlier, Korea’s sovereign bond rating in 1996 was AA- (S&P). But within
10
For example, Remolona, Scatigna and Wu (2008) reach this conclusion after using sovereign credit ratings and
historical default rates provided by rating agencies to construct a measure of ratings implied expected loss.
11
To be fair, S&P in a Reuter’s article dated January 14, 2009 warned Greece, Spain and Ireland that their ratings
could be downgraded further as economic conditions deteriorated. At that time, Greece was rated A1 by Moody’s
and A- by S&P. Interestingly, it was almost a full year later on December 22, 2009 that Greece was actually
downgraded by Moody’s to A2 (still highly rated), followed by further downgrades on April 23, 2010 (to A3) and
finally to “junk” status (Ba1) on June 14, 2010. As noted earlier, S&P downgraded Greece to “junk” status about
three months earlier.
12
See Pomerleano (1998), which is based on a longer article by the author (1997). Taking a somewhat similar
approach, many policy makers and theorists have recently focused on the so-called “shadow banking system.” For
example, Gennaioli, Martin and Rossi (2010) argued that the financial strength of governments depends on private
financial markets and its ability to attract foreign capital. They concluded that strong financial institutions not only
attract more capital but their presence also helps encourage their governments to repay their debt.
Chambers of S&P (1997) also mentions the idea of a “bottom-up” approach but not to the assessment of
sovereign risk, but to a corporate issuer located in a particular country. He advocates first an evaluation of an
issuer’s underlying creditworthiness to arrive at its credit rating and then considers the economic, business and
social environment in which the entity operates. These latter factors, such as the size and growth and the volatility
of the economy, exchange rates, inflation, regulatory environment, taxation, infrastructure and labor market
conditions are factored in on top of the micro variables to arrive at a final rating of the issuer.
10
10
a year, Korea’s rating dropped to BB-; and if not for the IMF bailout of $50 billion, the sovereign
would almost certainly have defaulted on its external, non-local currency debt. A traditional
macroeconomic measure like GDP growth would not have predicted such trouble since, at the
end of 1996, South Korea had been growing at double-digit rates for nearly a decade.13
The Z-Metrics™ Approach14
In 2009, we partnered with RiskMetrics Group with the aim, at least initially, of creating
a new and better way of assessing the credit risk of companies. The result was our new ZMetrics approach. This methodology might be called a new generation of the original Z-Score
model of 1968. Our objective was to develop up-to-date credit scoring and probability of default
metrics for both large and small, public and private, enterprises on a global basis.
In building our models, we used multivariate logistic regressions and data from a large
sample of both public and private U.S. and Canadian non-financial sector companies during the
20-year period 1989-2008.15 We analyzed over 50 fundamental financial statement variables,
including measures (with trends as well as point estimates) of solvency, leverage, size,
profitability, interest coverage, liquidity, asset quality, investment, dividend payout, and
financing results. In addition to such operating (or “fundamental”) variables, we also included
equity market price and return variables and their patterns of volatility. Such market variables
13
Afterwards, the World Bank and other economists such as Paul Krugman concluded that that crony capitalism and
the associated implicit public guarantees for politically influential enterprises coupled with poor banking regulation
were responsible for the crisis. The excesses of corporate leverage and permissive banking were addressed
successfully in the case of Korea and its economy was effectively restructured after the bailout.
14
For more details, see Altman, et al, 2010 “The Z-Metrics™ Methodology for Estimating Company Credit Ratings
and Default Risk Probabilities,” RiskMetrics Group, continuously updated, available from http://riskmetrics.com/ZMetrics.
15
Our first model’s original sample consisted of over 1,000 U.S. or Canadian non-financial firms that suffered a
credit event and a control sample of thousands of firms that did not suffer a credit event, roughly a ratio of 1:15.
After removing those firms with insufficient data, the credit event sample was reduced to 638 firms for our public
firm sample and 802 observations for our private firm sample.
11
11
have typically been used in the “structural distance-to-default measures” that are at the core of
the KMV model16 now owned by Moody’s.
In addition to these firm-specific, or micro, variables, we also tested a number of macroeconomic variables that are often used to estimate sovereign default probabilities, including GDP
growth, unemployment, credit spreads, and inflation. Since most companies have a higher
probability of default during periods of economic stress—for example, at the end of 2008—we
wanted to use such macro variables to capture the heightened or lower probabilities associated
with general economic conditions.17
The final model, which consists of 13 fundamental, market value, and macroeconomic
variables, is used to produce a credit score for each public company. (And as discussed later,
although our primary emphasis was on applying Z-Metrics to publicly traded companies, we also
created a private firm model by using data from public companies and replacing market value
with book value of equity.)
The next step was to use a logit specification of the model (described in the Appendix)
that we used to convert the credit scores into probabilities of default (PDs) over both one-year
and five-year horizons. The one-year model is based on data from financial statements and
market data approximately one year prior to the credit event, and the five-year model includes up
to five annual financial statements prior to the event.
To test the predictive power of the model and the resulting PDs, we segregated all the
companies in our sample into “cohorts” according to whether they experience “credit events”
16
Developed by Crosbie in 1998 and adapted for sovereigns by Gray in 2007.
In all cases, we carefully examined the complete distribution of variable values, especially in the credit-event
sample. This enabled us to devise transformations on the variables to either capture the nature of their distributions
or to reduce the influence of outliers. These transformations included logarithmic functions, first differences and
dummy variables if the trends or levels of the absolute measures were positive/negative.
17
12
12
that include either formal default or bankruptcy (whichever comes first). All companies that
experienced a credit event within either one year or five years were assigned to the “distressed”
or “credit event” group (with all others assigned to the non-distressed group).
Our test results show considerable success in predicting defaults across the entire credit
spectrum from the lowest to the highest default risk categories. Where possible, we compared
our output with that of publicly available credit ratings and existing models. The so-called
“accuracy ratio” measures how well our model predicts which companies do or do not go
bankrupt on the basis of data available before bankruptcy. The objective can be framed in two
ways: (1) maximizing correct predictions of defaulting and non-defaulting companies (which
statisticians refer to as Type I accuracy) and (2) minimizing wrong predictions (Type II
accuracy).
As can be seen in Figure 3, our results, which include tests on actual defaults during the
period 1989-2009, show much higher Type I accuracy levels for the Z-Metrics model than for
either the bond rating agencies or established models (including an older version of Z-Scores).
At the same time, our tests show equivalent Type II accuracies at all cutoff levels of scores.18
18
We assessed the stability of the Z-Metrics models by observing the accuracy ratios for our tests in the in-sample
and out-of-sample periods and also by observing the size, signs and significance of the coefficients for individual
variables. The accuracy ratios were very similar between the two sample periods and the coefficients and
significance tests were extremely close.
13
13
FIGURE 3
Type I error for Agency ratings, Z”-score, and Z-Metrics agency equivalent (AE ratings (19892008): one year prediction horizon for publicly owned firms
type I error rate (defaulters classified as non-defaulters / total defaulters)
1
AE rating: Z" score
type I error rate
0.8
Agency rating
AE rating: Z-Metrics public one year
0.6
0.4
0.2
C
CC
BB
B-
BB
BB
-
B+
B
B-
BB
+
/C
C/
C
0
rating class
(cutoff score = score at upper boundary of rating class N)
Perhaps the most reliable test of credit scoring models is how well they predict critical
events based on samples of companies that were not used to build the model, particularly if the
events took place after the period during which the model was built (after 2008, in this case).
With that in mind, we tested the model against actual bankruptcies occurring in 2009, or what we
refer to as our “out-of-sample” data. As with the full test sample results shown in Figure 3, our
Z-Metrics results for the “out of sample” bankruptcies of 2009 outperformed the agency ratings
and the 1968 Z-score and 1995 Z”-score models using both one-year and five-year horizons.
14
14
A “Bottom-Up” Approach for Sovereign Risk Assessment
Having established the predictive power of our updated Z-score methodology, our next
step was to use that model (which, again, was created using large publicly traded U.S.
companies) to evaluate the default risk of European companies. And after assuring ourselves
that the model was transferable in that sense, we then attempted to assess the overall
creditworthiness of sovereign governments by aggregating our Z-Metrics default probabilities for
individual companies and then estimating both a median default probability and credit rating for
different countries.
In conducting this experiment, we examined nine key European countries over three time
periods, end of 2008, 2009 and 2010 (Figure 4) and again at the end of 2010 (Figure 5), when the
crisis was well known. People clearly recognized the crisis and concern for the viability of the
European Union in June 2010, when Greece’s debt was downgraded to non-investment grade
and both Spain and Portugal were also downgraded. Credit markets, particularly CDS markets,
had already recognized the Greek and Irish problems before June 2010. Market prices during the
first half of 2010 reflected high implied probabilities of default for Greece and Ireland, but were
considerably less pessimistic in 2009. By contrast, as can be seen in Figure 4, which shows our
Z-Metric median PD estimates alongside sovereign CDS spreads over both periods,19 our PD
estimates were uniformly higher (more risky) in 2009 than early in 2010, even if the world was
more focused on Europe’s problems in the latter year. In this sense, our Z metrics PD might be
viewed as providing a leading indicator of possible distress. It should be noted that the statistics
19
The median CDS spread is based on the daily observations in the six/four-month periods. The median Z-Metrics
PD is based on the median company PDs each day and then we calculated the median for the period. The results are
very similar to simply averaging the median PDs as of the beginning and ending of each sample period.
15
15
in Figure 4 report only on the non-financial private sector, while those in Figure 5 include results
from our banking credit risk model, as well.
For the first four months of 2010, our Z-Metrics’ five-year PDs for European corporate
default risk placed Greece (10.60%) and Portugal (9.36%) in the highest risk categories (ZCratings), followed by Italy (7.99%), Ireland (6.45%) and Spain (6.44%), all in the ZC category.
Then came Germany and France (both about 5.5% - ZC+), with the U.K. (3.62%) and the
Netherlands (3.33%) at the lowest risk levels (ZB– and ZB). The U.S.A. looked comparatively
strong, at 3.93% (ZB-).
For the most part, these results are consistent with how traditional analysts now rank
sovereign risks. Nevertheless, there were a few surprises. The U.K. had a fairly healthy private
sector, and Germany and France were perhaps not as healthy as one might have thought. The
U.K.’s relatively strong showing might have resulted from the fact that our risk measure at this
time did not include financial sector firms, which comprised about 35% of the market values of
listed U.K. corporates and were in poor financial condition. And several very large, healthy
multinational entities in the U.K. index might have skewed results a bit. The CDS/5-year
market’s assessment of U.K. risk was harsher than that of our Z-Metrics index in 2010, with the
median of the daily CDS spreads during the first four months implying a 6.52% probability of
default, about double our Z-Metrics median level. Greece also had a much higher CDS implied
PD at 24.10%, as compared to 10.60% for Z-Metrics. (And, of course, our choice of the median
Z-Metrics PD is arbitrary, implying as it does that fully 50% of the listed companies have PDs
higher than 10.60%.)
We also observed that several countries had relatively high standard deviations of ZMetrics PDs, indicating a longer tail of very risky companies. These countries included Ireland,
16
16
Greece and, surprisingly, Germany, based on 2010 data. So, while almost everyone considers
Germany to be the benchmark-low risk country in Europe (for example, its 5-year CDS spread
was just 2.67% in 2010, even lower than the Netherlands (2.83%), we are more cautious based
on our broad measure of private sector corporate health.
2010 Results
Figure 5 shows the weighted-average median PDs for 11 (including now Sweden and
Belgium) European countries and the U.S. as of the end of 2010. Note that we now are able to
include PDs for the banking sectors (listed firms only) for these countries, an important addition,
especially for countries like Greece, Ireland and the U.K. The results show the large difference
between Greece (16.45%) and all the rest, but also that the “big-five PIIGS” stand out as the
clear higher risk domains. Indeed, we feel that Italy could be the ‘fulcrum” country to decide the
ultimate fate of the Euro (see our “Insight” piece in the Financial Times, June 21, 2011).
CDS Implied PDs
Figure 6 shows the implied PDs for the “Big-Five” European high-risk countries from the
start of 2009 to mid-July 2011, just after the European Union’s comprehensive rescue plan was
announced (July 21, 2011) for Greece and a contingent plan for other countries. Note that while
the PDs, based on CDS spreads and assuming a 40% recovery rate, all came down from their
highs, all still imply a considerable default risk.
2010 vs. 2009
As noted earlier from Figure 4, our PD estimates for 2009 were uniformly higher (more
risky) than those for early 2010. One important reason for the higher PDs in 2009 is the
17
17
significant impact of the stock market, which is a powerful variable in the Z-Metrics model—and
in many other default probability models (notably, Moody’s KMV). Recall that the stock
markets were at very low levels at the end of 2008 and into the early months of 2009, while there
was a major recovery later in 2009 and in early 2010.
FIGURE 4
Financial Health of the Corporate, Non-Financial Sector: Selected European Countries
and U.S.A. in 2008-2010
Z-Metrics PD Estimates: Five-Year Public Model
Five-Year Implied PD
from CDS Spread*
Listed
Companies
Y/E 2010
Median PD
Y/E 2009
Median PD
Y/E 2008
Median PD
2010
2009
2008
85
3.56%
3.33%
5.62%
2.03%
2.83%
6.06%
U.S.A.
2226
3.65%
3.93%
6.97%
3.79%
3.28%
4.47%
Sweden
245
3.71%
5.31%
6.74%
2.25%
4.60%
6.33%
Ireland
29
3.72%
6.45%
7.46%
41.44%
12.20%
17.00%
Belgium
69
3.85%
5.90%
5.89%
11.12%
4.58%
5.53%
U.K.
507
4.28%
3.62%
5.75%
4.73%
6.52%
8.13%
France
351
4.36%
5.51%
7.22%
4.51%
3.75%
4.05%
Germany
348
4.63%
5.54%
7.34%
2.50%
2.67%
3.66%
Italy
174
7.29%
7.99%
10.51%
9.16%
8.69%
11.20%
Spain
91
7.39%
6.44%
7.39%
14.80%
9.39%
8.07%
Portugal
33
10.67%
9.36%
12.07%
41.00%
10.90%
7.39%
Greece
93
15.28%
10.60%
11.57%
70.66%
24.10%
13.22%
Country
Netherlands
*Assuming a 40% recovery rate (R); based on the median CDS spread (s). PD computed as 1-e(-5*s/(1-R)).
Sources: RiskMetrics Group (MSCI), Markit, Compustat.
18
18
FIGURE 5
Weighted Average Median Five-Year (PD) for Listed Non-Financial* and Banking Firms**
(Europe and U.S.), 2010
Non-Financial Firms
Banking Firms
CDS
Weighted
Spread
Average
(%)
PD (%)*** Rank
Country
PD (%)
Weight
PD (%)
Weight
Rank
Netherlands
3.56
0.977
11.1
0.023
3.73
1
2.03
1
Sweden
3.71
0.984
17.3
0.016
3.93
2
2.25
2
Belgium
3.85
0.972
12.4
0.028
4.21
3
11.12
8
France
4.36
0.986
14.0
0.014
4.49
4
4.51
5
U.K.
4.28
0.977
15.5
0.023
4.54
5
4.73
6
Germany
4.63
0.983
13.1
0.017
4.77
6
2.50
3
U.S.A.
3.65
0.837
13.8
0.163
5.30
7
3.79
4
Spain
7.39
0.948
10.9
0.052
7.57
8
14.80
9
Italy
7.29
0.906
20.0
0.094
8.48
9
9.16
7
Ireland
3.72
0.906
77.6
0.094
10.65
10
41.44
11
Portugal
10.67
0.971
12.1
0.029
10.71
11
41.00
10
Greece
15.28
0.921
30.1
0.079
16.45
12
70.66
12
*Based on the Z-Metrics Probability Model. **Based on Altman-Rijken Model (Preliminary). ***PD based on
the CDS Spread as of 4/26/11.
FIGURE 6
Five-Year Implied Probabilities of Default (PD)* from Capital Market CDS Spreads, Jan 2009 –
Jul 22, 2011**
90
Default Probability (As %)
80
Greece
74.52
70
60
Portugal
53.38
50
Ireland
51.57
40
Spain
22.70
30
20
Italy
19.01
10
Spain
Italy
Portugal
4-Jul-11
4-Jun-11
4-Apr-11
4-May-11
4-Jan-11
4-Feb-11
4-Mar-11
4-Dec-10
4-Oct-10
4-Nov-10
4-Sep-10
4-Jul-10
4-Aug-10
4-Jun-10
4-Apr-10
Greece
4-May-10
4-Jan-10
4-Feb-10
4-Mar-10
4-Dec-09
4-Oct-09
4-Nov-09
4-Sep-09
4-Jul-09
4-Aug-09
4-Jun-09
4-Apr-09
4-May-09
4-Jan-09
4-Feb-09
4-Mar-09
0
Ireland
* Assumes 40% Recovery Rate. PD computed as 1-e(-5*s/(1-R)). ** On July 19, 2011, PDs for all countries
peaked as follows: Greece 88.22, Portugal 64.74, Ireland 64.23, Spain 27.54, and Italy 23.74.
Sources: Bloomberg and NYU Salomon Center.
19
19
Figure 7 shows, for each of our nine European countries and the U.S., the percentage
increases in median stock market index levels and sovereign PD levels between the first six
months of 2009 and the first six months of 2010. As can be seen in the figure, most countries
enjoyed increases of greater than 20%. Only Greece had a relatively low increase (5.5%),
consistent with its modest improvement in its Z-Metrics PD (-8.4%). Figure 6 shows the
percentage improvement (lower risk) in sovereigns’ PDs in 2010, which are largely consistent
with the increases in stock market index values. Note that Ireland stands out in that while its
stock market index value increased by 26.2%, its corporate sector experienced only a modest
improvement (-7.4%) in its Z-Metrics’ median PD. This may be attributable to the earlier
austerity measures taken in Ireland, as compared to those in other distressed European nations.
But likely more important were changes in the many other variables in the Z-Metrics model that
are not affected by stock prices, particularly the fundamental measures of corporate health.
FIGURE 7
Median Percentage Change in Various Country Stock Market Index Values and Z-Metrics’ PDs
Between the First Six Months of 2010 vs. 2009
Country
Index
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
UK
USA
CAC40
DAX
ASE
ISEQ
FTSEMIB
AEX
PSI-20
IBEX35
FTSE100
S&P500
Median Percent
Median Z-Metrics Percent
Change (2010 vs. 2009)*
Change (2010 vs. 2009)
24.1%
31.8%
5.5%
26.2%
18.2%
34.4%
17.8%
20.9%
27.8%
31.9%
-23.6%
-24.5%
- 8.4%
- 7.4%
-24.0%
- 25.3%
-22.4%
-12.9%
-37.6%
-43.6%
*Median of the various trading day stock index values and PDs, first six months of 2009 vs. First six
months of 2010.
Sources: Z-Metrics Model calculations from RiskMetrics (MSCI) Group, Bloomberg for stock index values.
20
20
Comparing PD Results Based on Privately Owned vs. Publicly Owned Firm Models
As shown in Figures 4 and 5, the improvement (reduction) in Z-Metrics PDs for most countries
in 2010—a period in which most EU sovereigns appeared to be getting riskier—looks
attributable in large part to the stock market increases in almost all countries. But to the extent
such increases could conceal a deterioration of a sovereign’s credit condition, some credit
analysts might prefer to have PD estimates that do not make use of stock market data.
With this in mind, we applied our private firm Z-Metrics model to evaluate the same nine
European countries and the U.S. The private and public firm models are the same except for the
substitution of equity book values (and volatility of book values) for market values. This
adjustment is expected to remove the capital market influence from our credit risk measure.
Figure 8 summarizes the results of our public vs. private firm Z-Metrics models
comparative PD (delta) results for 2010 and 2009. For eight of the ten countries, use of the
private firm model showed smaller reductions in PDs when moving from 2009 to 2010 than use
of the public model. Whereas the overall average improvement in PDs for the public firm model
was a drop of 1.91 percentage points, the drop was 0.79% for our private firm model. These
results are largely the effect of the positive stock market performance in late 2009 and into 2010.
But improvements in general macro conditions, along with their effects on traditional corporate
performance measures, also helped improve (reduce) the PDs. Moreover, in two of these eight
countries—the U.K. and France—not only did the public firm model show an improved (lower)
PD, but the private firm model’s PD actually got worse (increased) in 2010 (as indicated by the
positive delta in the last column of Figure 8).
21
21
FIGURE 8
Private Vs. Public Firm Model PDs in 2010 and 2099
Country
No. Listed
Companies
2010 2009
Public-Firm Z-Metrics Model Private-Firm Z-Metrics Model
PDs
PDs
PDs
PDs
2010
2009
Delta*
2010
2009
Delta*
Netherlands
61
60
3.33% 5.62% -2.29%
5.25% 6.00% -0.75%
U.K.
442 433
3.62% 5.75% -2.13%
6.48% 5.97% +0.49%
U.S.A.
2226
2171 3.93% 6.97% -3.04%
4.28% 4.80% -0.52%
France
297 294
5.51% 7.22% -1.71%
7.33% 7.19% +0.14%
Germany
289 286
5.54% 7.34% -1.80%
6.29% 7.56% -1.27%
Spain
82
78
6.44% 7.39% -0.95%
8.06% 9.32% -1.26%
Ireland
28
26
6.45% 7.46% -1.01%
6.31% 6.36% -0.05%
Italy
155 154
7.99% 10.51% -2.52%
8.14% 9.07% -0.89%
Portugal
30
30
9.36% 12.07% -2.71%
8.73% 9.62% -0.89%
Greece
79
77
10.60% 11.57% -0.97% 11.03% 13.93% -2.90%
Average
6.28% 8.19% -1.91%
7.19% 7.98% -0.79%
_________________________________________________________________________
*Negative sign means improved credit risk.
Sources: Figure 4 and Riskmetrics (MSCI).
Correlation of Sovereign PDs: Recent Evidence on Z-Metrics vs. Implied CDS PDs
As a final test of the predictive of our approach, we compared our Z-Metrics five-year
median PDs for our sample of nine European countries (both on a contemporary basis and for
2009) with the PDs implied by CDS spreads in 2010. The contemporary PD correlation during
the first third of 2010 was remarkably high, with an R2 of 0.82. This was a period when it was
becoming quite evident that certain European countries were in serious financial trouble and the
likelihood of default was not trivial. But if we go back to the first half of 2009, the correlation
drops to an R2 of 0.36 (although it would be considerably higher, at 0.62, if we excluded the case
of Ireland). Ireland’s CDS implied PD was considerably higher in 2009 than 2010 (17.0% vs.
12.0%), while the Z-Metrics PD was relatively stable in the two years (7.5% and 6.5%
22
22
respectively).20 In 2010, whether we calculate the correlation with or without Ireland, the results
are essentially the same (0.82 and 0.83).
Given the predictive success of Z-metrics in the tests already described, we were curious
to find out whether it could be used to predict capital market (i.e., CDS) prices. So, we regressed
our public firm model’s 2009 Z-Metrics median, non-financial sector PDs against implied CDS
PDs one year later in 2010. Admittedly, this sample was quite small (10 countries) and the
analysis is for only a single time-series comparison (2009 vs. 2010). Nevertheless, these two
years spanned a crucial and highly visible sovereign debt crisis, whereas the PDs implied by
prior years’ Z-Metrics and CDS showed remarkably little volatility.21
As can be seen in Figure 9, the correlation between our Z-Metrics PDs and those implied
by CDS one year later proved to be remarkably strong, with an r of 0.69 and R2 of 0.48. In sum,
the corporate health index for our European countries (plus the U.S.) in 2009 explained roughly
half of the variation in the CDS results one year later.22
A potential shortcoming of our approach is that we are limited in our private sector
corporate health assessments to data from listed, publicly held firms. This is especially true for
relatively small countries like Ireland (with just 28 listed companies), Portugal (with 30), Greece
(79), Netherlands (61), and Spain (82). Since the private, non-listed segment is much larger in
20
No doubt the CDS market was reacting quite strongly to the severe problems in the Irish banking sector in 2009,
while Z-Metrics PDs were not impacted by the banks. This implies a potential strength of the CDS measure,
although the lower CDS implied PD in early 2010 was not impressive in predicting the renewed problems of Irish
banks and its economy in the fall of 2010.
21
The last time an entire region and its many countries had a sovereign debt crisis was in Asia in 1997-1998.
Unfortunately, CDS prices were not prominent and the CDS market was illiquid at that time.
22
Several other non-linear structures (i.e., power and exponential functions) for our 2009 Z-Metrics vs. 2010 CDS
implied PDs showed similar results. In all cases, we are assuming a recovery rate of 40% on defaults in calculation
of implied sovereign PDs.
23
23
all of the countries, we are not clearly assessing the health of the vast majority of its firms and
our sovereign health index measure is incomplete.23
But if the size of the listed firm population is clearly a limitation in our calculations, there
does not seem to be a systematic bias in our results. To be sure, the very small listings in
Ireland, Portugal, and Greece appear heavily correlated with their high PDs, but the country with
the lowest PD (the Netherlands) also has a very small listed population. Another potentially
important factor is that the listed population in countries like the U.K. and the Netherlands is
represented quite heavily by multinational corporations that derive most of their income from
outside their borders.24
23
We suggest that complete firm financial statement repositories, such as those that usually are available in the
sovereign’s central bank be used to monitor the performance of the entire private sector.
24
Results showing the percentage of “home-grown” revenues for listed firms across our European country sample
were inclusive, however, as to their influence on relative PDs.
24
24
25
CDS Implied PD
0.00%
0.00%
5.00%
10.00%
15.00%
20.00%
25.00%
30.00%
Figure 9
2.00%
4.00%
Spain
Ireland
25
6.00%
8.00%
Z-metrics PD
France
Netherlands
USA Germany
UK
R-Square = 48%
y = 1.9367x - 0.0743
12.00%
Portugal
Greece
Italy
10.00%
2009 Z-metrics PD vs. 2010 CDS Implied PD
Figure 11
14.00%
Conclusion and Implications
As the price for bailing out distressed sovereigns, today’s foreign creditors, especially the
stronger European nations, are demanding a heavy dose of austerity. Several governments,
including those of Greece, Ireland, Spain, Portugal, Italy, and the U.K., have already enacted
some painful measures. Others, such as France and Hungary, have either resisted austerity
measures or faced significant social unrest when austerity measures have been proposed. These
measures typically involve substantial cuts in cash benefits paid to public workers, increases in
retirement age, and other reduced infrastructure costs, as well as increased taxes for companies
and individuals. The objective is to reduce deficits relative to GDP and enhance the sovereigns’
ability to repay their foreign debt and balance their budgets.
While recognizing the necessity of requiring difficult changes for governments to qualify
for bailouts and subsidies, we caution that such measures should be designed to inflict as little
damage as possible on the health and productivity of the private enterprises that ultimately fund
the sovereign. The goal should be to enable all private enterprises with clear going concern
value to pay their bills, expand (or at least maintain) their workforces, and return value to their
shareholders and creditors (while those businesses that show no promise of ever making a profit
should be either reorganized or liquidated). For this reason, raising taxes and imposing other
burdens on corporate entities is likely to weaken the long-run financial condition of sovereigns.
To better estimate sovereigns’ risk of default, we propose that traditional measures of
macroeconomic performance be combined with more modern techniques, such as the contingent
claims analysis pioneered by Robert Merton and the bottom-up approach presented in these
pages. Along with the intuitive appeal of such an approach and our encouraging empirical
results, the probabilities of sovereign default provided by aggregating our Z-Metrics across a
26
26
national economy can be seen, at the very least, as a useful complement to existing methods and
market indicators—one that is not subject to government manipulation of publicly released
statistics. Using our approach, the credit and regulatory communities could track the
performance of publicly held companies and the economies in which they reside—and by
making some adjustments, unlisted entities as well. And if sovereigns were also willing to
provide independently audited statistics on a regular basis, so much the better.
Edward Altman is the Max L. Heine Professor of Finance, NYU Stern School of Business,
ealtman@stern.nyu.edu.
Herbert Rijken is Professor of Finance, Vrije University, Amsterdam, the Netherlands,
hrijken@feweb.vu.nl.
27
27
APPENDIX: Logit Model Estimation of Default Probabilities
We estimated our credit scoring model based on a standard logit-regression functional form
whereby:
CSi,t D 6% j &i,t H i,t (1)
CSi,t
ZMetrics credit score of companyi at timet
Bj
variable parameters(or weights)
X i ,t
set of fundamenta
l , marketbased and macroecono
mic variables for
firm i quarterobservations
H i,t
error terms (assumedto be identically and independently distributed )
CS i ,t is transformed int o a probability of defaultby PDi ,t
x
1
1 exp (CS i ,t )
We compare Z-Metrics results with issuer ratings. To ensure a fair comparison, credit scores are
converted to agency equivalent (AE) ratings by ranking credit scores and by matching exactly
the actual Agency rating distribution with the AE rating distribution at any point in time.
x
We also compare our Z-Metrics results to the well established Altman Z”-score (1995) model.25
25
Altman’s original Z-score model (1968) is well-known to practitioners and scholars alike. It was built, however,
over 40 years ago and is primarily applicable to publicly-held manufacturing firms. A more generally applicable
Z”-score variation was popularized later (Altman, Hartzell and Peck, 1995) as a means to assess the default risk of
non-manufacturers as well as manufacturers, and was first applied to emerging market credits. Both models are
discussed in Altman and Hotchkiss (2006) and will be compared in several tests to our new Z-Metrics model.
Further, the Altman Z-score models do not translate easily into a probability of default rating system, as does the ZMetrics system. Of course, entities that do not have access to the newer Z-Metrics system can still use the classic Zscore frameworks, although accuracy levels will not be as high and firm PDs not as readily available.
28
28
References
Abassi, B. and R. J. Taffler, 1982, “Country Risk: A Model of Economic Performance Related to
Debt Servicing Capacity,” WP #36, City University Business School, London.
Altman, E. I., 1968, “Financial Ratios Discriminant Analysis and the Prediction of Corporate
Bankruptcy,” Journal of Finance, v. 23, 4, September, 189.
Altman, E. I. and E. Hotchkiss, 2006, Corporate Financial Distress and Bankruptcy, 3rd edition,
John Wiley & Sons, NY and NJ.
Altman, E. I., et. al., 2010, “The Z-Metrics™ Methodology for Estimating Company Credit
Ratings and Default Risk Probabilities,” RiskMetrics Group, NY, June, available from
www.riskmetrics.com/Z-Metrics.
Babbel, D. F., 1996, “Insuring Sovereign Debt against Default,” World Bank Discussion Papers,
#328.
Baek, I. A. Bandopadhyaya and C. Du, 2005, “Determinants of Market-Assessed Sovereign Risk:
Economic Fundamentals or Market Risk Appetite?” Journal of International Money and
Finance, Vol. 24 (4), pp. 533-48.
Beers, D., M. Cavanaugh and O. Takahira, 2002, “Sovereign Credit Ratings: A Primer,” Standard
& Poor’s Corp., NY, April.
Bertozi, S., 1996, “An Annotated Bibliography on External Debt Capacity,” in D. Babbel’s
“Insuring Sovereign Debt Against Default,” World Bank Discussion Papers #328.
Caouette, J., E. Altman, P. Narayanan and R. Nimmo, 2008, Managing Credit Risk, 2nd edition,
John Wiley & Sons, NY.
Chambers, W.J., 1997, “Understanding Sovereign Risk,” Credit Week, Standard & Poor’s January 1.
Cline, W., 1983, “A Logit Model of Debt Restructuring, 1963-1982,” Institute for International
Economics, WP, June.
Feder, G. and R. E. Just, 1977, “A Study of Debt Servicing Capacity Applying Logit Analysis,”
Journal of Development Economics,” 4 (1).
Feder, G. R. E. Just and K. Ross, 1981, “Projecting Debt Capacity of Developing Countries,”
Journal of Financial & Qualitative Analysis, 16 (5).
Flynn, D., 2009, “S&P Cuts Greek Debt Rating as Global Crisis Bites,” Reuters, January 14.
Frank, C. R. and W. R. Cline, 1971, “Measurement of Debt Servicing Capacity: An
Application of Discriminant Analysis,” Journal of International Economics, 1.
Frenkel, M., A. Karmann and B. Scholtens, eds., 2004, “Sovereign Risk and Financial Crises,”
Heidelberg and New York, Springer, xii, 258.
Gennaioli, N., A. Martin and S. Rossi, 2010, “Sovereign Default, Domestic Banks and Financial
Institutions,” Working Paper, Imperial College, London, July.
Gerlach, S., A. Schultz and G. Wolff, 2010, “Banking and Sovereign Risk in the Euro Area,”
Deutsche Bundesbank, Research Centre, Discussion Paper Series 1: Economic Studies: 2010.
Gray, D. F., R. Merton and Z. Bodie, 2006, “A New Framework for Analyzing and Managing
Macrofinancial Risk of an Economy,” IMF Working Paper, October.
Gray, D. F., R. Merton and Z. Bodie, 2007, “Contingent Claims Approach to Measuring and
Managing Sovereign Credit Risk,” Journal of Investment Management, vol. 5, No. 4, p.1.
Grinols, E., 1976, “International Debt Rescheduling and Discrimination Using Financial
Variables,” U.S. Treasury Dept., Washington, D.C.
29
29
Haugh, D., P. Ollivaud and D. Turner, 2009, “What Drives Sovereign Risk Premiums?: An
Analysis of Recent Evidence from the Euro Areas,” OECD, Economics Department,
Working Paper, 718.
Hilscher, J. and Y. Nosbusch, 2010, “Determinants of Sovereign Risk: Macroeconomic
Fundamentals and the Pricing of Sovereign Debt,” Review of Finance, Vol. 14 (2), pp. 23562. IMF, 2010, “Global Financial Stability Report,” Washington, D.C.
KMV Corporation, 1999, “Modeling Default Risk,” KMV Corporation, R. Crosbie.
Krugman, P., 1989, “Financing vs. Forgiving a Debt Overhang: Some Analytical Notes,”
Journal of International Business Studies, 17.
Longstaff, F., J. Pan, L. Pedersen and K. Singleton, 2007, “How Sovereign is Sovereign Credit
Risk?,” National Bureau of Economic Research, Inc. , NBER Working Paper: 13658.
Merton, R. C., 1974, “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,”
Journal of Finance, 29, May, 449.
Neziri, H., 2009, “Can Credit Default Swaps predict Financial Crises?,” Journal of Applied
Economic Sciences, Volume IV/Issue 1(7)
Oshiro, N., Y. Saruwatari, 2005, “Quantification of Sovereign Risk: Using the Information in
Equity Market Prices,” Emerging Markets Review, Vol. 6 (4), pp. 346-62.
Pomerleano, M., 1998, “Corporate Finance Lessons from the East Asian Crisis,” Viewpoint, The
World Bank Group, Note #155, October.
Pomerleano, M., 1999, “The East-Asia Crisis and Corporate Finance – The Untold Micro
Study,” Emerging Markets Quarterly.
Reinhart, M. and K. Rogoff, 2010, “This Time is Different,” Princeton University Press,
Princeton, NJ.
Remolona, E. M. Scatigna and E. Wu, 2008, “A Ratings-Based Approach to Measuring
Sovereign Risk,” International Journal of Finance and Economics, Vol. 13 (1), pp. 26-39.
Saini, K. and P. Bates, 1978, “Statistical Techniques for Determining Debt Servicing Capacity
for Developing Countries: Analytical Review of the Literature and Further Empirical
Results,” Federal Reserve Bank of New York Research Paper, #7818.
Sargen, H., 1977, “Economics Indicators and Country Risk Appraisal,” Federal Reserve Bank of
San Francisco, Economic Review, Fall.
Schmidt, R., 1984, “Early Warning of Debt Rescheduling,” Journal of Banking and Finance, 8.
Segoviano, B., A. Miguel, C. Caceres and V. Guzzo, 2010, “Sovereign Spreads: Global Risk
Aversion, Contagion or Fundamentals?,” IMF Working Paper: 10/120, p. 29.
Sgherri, S. and E. Zoli, 2009, “Euro Area Sovereign Risk During the Crisis,” International
Monetary Fund, IMF Working Papers: 09/222.
Smith, R. and I. Walter, 2003, Global Banking, Oxford University Press, London.
Trebesch, C., U. Das and M. Papaioannou, 2010, “Sovereign Default Risk and Private Sector
Access to Capital in Emerging Markets,” IMP Working Paper: October.
30
30
The Fate of the Euro: It is Still Italia!
Edward I. Altman1
One year ago, I wrote about the European debt crisis and predicted that Italy would be
either the hero or villain of the Euro. I felt then, as I do now, that the escalation of the crisis
would come down to whether one of the key Southern European countries would be able to
survive, without a bailout, the onslaught of a capital market “attack.” Based on the inherent
strengths of its fundamental competitive and wealth attributes, I concluded that Italy would be
the “fulcrum country,” with a 70% chance to emerge successfully, enabling the Euro itself to
survive. Since the end of 2010, unfortunately, Italy’s fundamentals have deteriorated
dramatically, its economy is in a double-dip recession, unemployment is over 10%, and even top
European politicians are now saying that the Euro’s survival is at the critical stage. My current
assessment is that Italy’s, and also the Euro’s, solvency is at best a 50-50 probability,
notwithstanding the announced change in EU policy toward pumping €130 billion of fiscal
stimuli into the most vulnerable European nations and steps toward a financial and political
union.
My increased pessimism is based on our “bottom-up” approach toward assessing the
health of any sovereign. Recently, Professor Herbert Rijken of the Free University of Amsterdam
and I began suggesting that financial and political analysts should not focus solely on the
traditional macroeconomic metrics like Debt/GDP or Deficits/GDP, but also to monitor the
health of the sovereign based on the condition of its private sector - - both its non-financial
corporate sector and its privately owned banks. After all, if the corporate sector is healthy, it can
pay more taxes from profits and hire more workers, as well as provide vital new investments. On
1
Max L. Heine Professor of Finance, NYU Stern School of Business, Director of Debt and Credit Markets,
NYU Salomon Center
1
31
the other hand, if a significant proportion of a sovereign’s private sector is on the verge of
financial distress and bankruptcy, or needs increased capital itself, it cannot hope to contribute
much, particularly if companies are asked to increase tax payments.
We developed an index of individual firm probability of default, based on an updated
version of my somewhat ancient, but still relevant and accurate Z-Score Approach (now
available in a new App called “AltmanZscoreplus.com”). We then observed the median (50th
percentile) and 75th percentile firm’s probability of default (PD) for the non-financial, listedfirm, population in each of nine European countries, as well as in the U.S.A., for the years 20082010 (see Figure 1). The results were extremely revealing with the highest risk countries, Greece
(16.7%) and Portugal (16.6%), followed by Italy (11.3%), and Spain (8.6%), showing the most
troubling corporate PDs as of year-end 2008, even before the financial world showed much
concern with these sovereigns and before the Greek PM informed the world in October 2009,
that its budget deficit was 12.7% (double the previous estimate).
Our updated bottom-up results through the end of 2011 are startling and highly indicative
of the profound deterioration of all European nations in just one year, with Italy and France
“leading the way” down. For example, the 75th percentile listed non-financial company in Italy
(e.g., indicative of the 25% most risky companies), as of year-end 2010, had a PD of 14.1% over
a five-year horizon (note that 25% of the private sector’s PDs was greater than 14.1%). In just
one year, that figure spiked to 26.4%, (second only in Europe to Greece’s astounding 50.5%), a
deterioration of 87% (see Figure 2), the largest drop in Europe. It is worth noting that these are
the largest, and arguably the most solvent Italian enterprises. A weakened banking sector, with
its own profitability and capital challenges, will be hard pressed to support such a problematic
corporate sector. While Italy’s global competitiveness is far stronger than that of Greece,
2
32
Portugal and Spain, the problem is that unlike these smaller vulnerable countries, Italy is “too big
to save.” Thus, while we focus on the private sector, the world’s markets, as well as European
politicians, are traumatized by Italy’s escalating cost of new debt financing that now exceeds the
thought to be unsustainable 6.0% level for 10-year bonds. Furthermore, the implied 5-year PD
from CDS spreads for Italy is above 35% and close to 40% for Spain, (Figure 3). No wonder
Italy’s PM, Mario Monti, has conceded that the survival of the Euro is now at stake. His country
will be the last bulwark before the possible bursting of the “Euro-dyke” that EU leaders are so
desperately trying to keep intact - - with good reason.
A surprising recent finding in our sovereign risk assessment is that the country with the
second largest percentage deterioration in the risk profile of its private, non-financial in sector in
2011 is France, whose 75th percentile-firm has a PD of 14.8%, up from 8.5% one year ago - - a
74% slide! Among the countries we analyzed in 2011, the Netherlands’s 75th percentile was the
lowest (8.7%). Incidentally, the 75% percentile U.S. company’s PD is slightly higher than the
Dutch at 11.7% - - a respectable figure similar to that of Germany. One can wonder if France,
with its enormously leveraged and problematic banking system and with an increasingly
vulnerable private sector, is indeed worthy of its AA+ (S&P) and AAA (Fitch), Aaa (Moody’s)
ratings. I certainly question those lofty assessments. The same could be said for Italy’s A3/Arating.
So, will Italy make it? Will the Euro survive? We should know the answer within 12
months, perhaps sooner, but the odds against this happy ending are increasing with each piece of
bad news about Italy’s economy and its fundamental components.
3
33
Figure 1: Financial Health of the Corporate, Non-Financial Sector: Selected European Countries and
U.S.A. in 2008 – 2011, Z-Metrics Median PD Estimates
Z-Metrics PD Estimates*: Five-Year Public Model
Country
Listed
Companies
(2010)
Y/E 2011
Y/E 2010
Y/E 2009
Y/E 2008
Sweden
245
2.7%
2.6%
3.1%
6.7%
Netherlands
85
3.1%
2.5%
2.7%
5.0%
Germany
348
4.6%
3.9%
4.5%
7.6%
U.K.
507
4.6%
3.7%
4.5%
7.3%
U.S.A.
2226
4.8%
3.8%
3.3%
4.5%
France
351
6.6%
4.0%
4.6%
7.2%
Spain
91
10.6%
7.1%
5.9%
8.6%
Italy
174
11.9%
7.7%
7.7%
11.3%
Portugal
33
15.1%
9.9%
8.2%
16.6%
Greece
93
26.7%
18.7%
11.9%
16.7%
Median PD
*
Since the Z-Metrics Model is not practically available for most analysts, we could substitute the Z”-Score method (available from
).
Sources: RiskMetrics Group (MSCI), Markit, Compustat.
4
34
Figure 2: Financial Health of the Corporate, Non-Financial Sector: Selected European Countries and
U.S.A. in 2008 – 2011, Z-Metrics 75th Percentile PD Estimates
Z-Metrics PD Estimates*: Five-Year Public Model
Listed
Companies
(2010)
Y/E 2011
Y/E 2010
Y/E 2009
Y/E 2008
Netherlands
85
8.7%
5.7%
6.7%
15.7%
Sweden
245
9.6%
6.8%
8.0%
13.5%
U.K.
507
9.7%
5.7%
9.3%
16.6%
Germany
348
11.2%
9.7%
11.9%
22.2%
U.S.A.
2226
11.7%
8.0%
11.5%
19.5%
France
351
14.8%
8.5%
10.3%
19.2%
Spain
91
20.1%
13.2%
12.7%
18.4%
Portugal
33
24.9%
20.1%
12.3%
26.6%
Italy
174
26.4%
14.1%
18.1%
27.1%
Greece
93
50.5%
40.1%
27.6%
31.0%
Country
75th Percentile PD
*
Since the Z-Metrics Model is not practically available for most analysts, we could substitute the Z”-Score method (available from
).
Sources: RiskMetrics Group (MSCI), Markit, Compustat.
5
35
Figure 3: Five Year Implied Probabilities of Default (PD) from Capital Market CDS Spreads*, Jan. 2009 –
June 25, 2012
100
Greece (9/16/11)
94.75
90
Default Probability (As %)
80
Portugal
49.77
70
60
50
Ireland
40.27
40
Spain
38.42
30
Italy
35.65
20
10
Spain
Italy
Greece
Portugal
4-May-12
4-Mar-12
4-Jan-12
4-Nov-11
4-Sep-11
4-Jul-11
4-May-11
4-Mar-11
4-Jan-11
4-Nov-10
4-Sep-10
4-Jul-10
4-May-10
4-Mar-10
4-Jan-10
4-Nov-09
4-Sep-09
4-Jul-09
4-May-09
4-Mar-09
4-Jan-09
0
Ireland
*
Assuming a 40% recovery rate (R); based on the median CDS spread (s). PD Computed as 1-e (-5*s/(1-R)).
Source: Bloomberg
6
36
New York University Salomon Center
Leonard N. Stern School of Business
Special Report on
Defaults and Returns in the High-Yield Bond and Distressed Debt
Market:
The Year 2018 in Review and Outlook
By
Edward I. Altman
and
Brenda J. Kuehne
February 21, 2019
37
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Contents
Executive Summary ....................................................................................................................................... 3
Defaults, Default Rates, and Recoveries ........................................................................................................ 5
Bankruptcies in 2018 and Trends in Bankruptcy Filings ............................................................................. 10
Industry Defaults.......................................................................................................................................... 16
Age of Defaults ............................................................................................................................................ 19
Fallen Angel Defaults .................................................................................................................................. 21
Default Losses and Recoveries .................................................................................................................... 23
Recovery Forecast Versus Actual ................................................................................................................ 27
Distressed Exchanges and Recoveries in 2018 ............................................................................................ 28
Related Recovery Statistics .......................................................................................................................... 31
Mortality Rates and Losses .......................................................................................................................... 35
Returns, Yields and Spreads ........................................................................................................................ 37
A Continuing Investment Dilemma ............................................................................................................. 39
Credit Markets: The Benign Credit Cycle Continues .................................................................................. 42
Global Debt Concerns .................................................................................................................................. 44
New Issues and Other Changes in Size of the High-Yield Market .............................................................. 53
Proportion and Size of the Distressed and Defaulted Public and Private Debt Markets .............................. 53
Forecasting Default Rates and Recoveries ................................................................................................... 57
Performance of Defaulted Debt Securities ................................................................................................... 63
Appendix A: Quarterly Default Rate Comparison (1989 -- 2018) ............................................................... 64
Appendix B: Defaulted Corporate Straight Bond Issues ............................................................................. 67
Appendix C: Distressed Exchanges ............................................................................................................. 69
Appendix D: Leveraged Loan Defaults ....................................................................................................... 70
Appendix E: Chapter 11 Filings with Liabilities ≥$1 billion (2008 -- 2018) ............................................... 71
Appendix F: Chapter 11 Filings by Liability Size in 2018 .......................................................................... 75
Appendix G: Defaults by Industry ............................................................................................................... 77
Appendix H: Emergences from Bankruptcy ................................................................................................ 78
Acknowledgments
Dr. Altman is the Max L. Heine Professor of Finance, Emeritus, and Director of the Credit and
Debt Markets Research Program at the NYU Salomon Center, Leonard N. Stern School of
Business. Brenda Kuehne is a Credit and Debt Markets Research Specialist at the NYU Salomon
Center. We appreciate the assistance of the several market makers who provided us with price
quotations. We offer a special thanks to the various rating agencies, S&P, Ben Schlafman and
Kerry Mastroianni of New Generation Research, and the members of the Credit Strategy
Investment team of Paulson & Co.
2
38
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Despite a continuation of extremely low default and relatively high recovery
rates, the year 2018 proved to be a challenging one for investors and issuers of
high-yield bonds from both a new issuance and performance perspective; the
annual amount of new issuance was the lowest recorded since 2009 and annual
returns, on both an absolute and relative basis, were negative. Additionally,
9.9% of outstanding high-yield bonds were classified as distressed by year-end,
compared to only 6.1% one year earlier.
The default rate declined to 1.74% in 2018, 6bp lower than last year, and was
the lowest annual rate since 2013. Our forecast for 2018 was for a default rate
of 2.53% (2.88% if we equally weight our two main forecasting models). The
fourth-quarter 2018 default rate was 0.38%, slightly lower than one year
(0.45%), but higher than one quarter (0.08%), prior. This marks the third
consecutive quarter in which quarterly default rates were lower than 0.50%.
The S&P/LSTA 12-month, dollar-denominated default rate on leveraged loans
decreased from 2.05% at the end of 2017 to 1.63% by year-end 2018, while the
issuer default rate fell from 1.72% to 1.56%. In the majority of months since
March 2013, the dollar-denominated loan default rate exceeded the issuerdenominated rate due to relatively large defaults by only a few issuers.
Typically, during benign credit cycles, however, the issuer denominated default
rate on loans and bonds exceeds the dollar denominated rate.
Default losses on high-yield bonds were 0.91%, based on a weighted average
recovery rate of 52.1% just after default, a level notably higher than the
historical arithmetic average of 46.0%, but slightly lower than last year’s rate of
56.7%, the highest rate since 2014. The weighted average recovery on
bankruptcy and payment defaults in 2018 was lower at 48.6%, with the
recovery on distressed exchange defaults considerably higher at 66.4%.
Returns on high-yield bonds were negative in 2018, ending the year at -2.13%
(FTSE Index).1 The excess return versus 10-yr US Treasuries was -2.11%,
inferior compared to the 2.72% historic average, and was well below the 4.92%
experienced in 2017. Yield-to-maturity (YTM) spreads versus 10-yr US
Treasuries increased to 5.47% by year-end 2018, a level 153bp higher than
year-end 2017, and exceeded the historical average of 5.19% for the first time
since 2015. Defaulted bonds and bank loans experienced similar losses in 2018,
with a combined annual return of -3.25%.
US high-yield bond issuance in 2018 decreased considerably from last year
($240.7 billion), reaching a total of $143.6 billion (BoAML statistics). The
proportion of high-yield bonds issued at B- or lower was slightly above the
22.7% average of the last five years, with 23.3% issued in these rating
categories in 2018. Similarly, new issuance of leveraged loans (secured, noninvestment grade loans) decreased in 2018, with S&P estimating that $620
billion of new, U.S. dollar-denominated leveraged loans (institutional and “pro
rata” syndicated by banks)) were issued during the year. This amount was down
approximately 5% from the $651 billion issued in 2017.
1
In 2017 FTSE took over the calculation of the high-yield bond index from our theretofore usual Citi source.
39
3
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
The distress ratio of bonds yielding more than 1,000bp over comparable
duration treasuries, measured by number of issues, increased significantly to
9.9% as of the end of 2018, from 3.9% three months earlier, but a little less so
from 6.1% at year-end 2018. The distress ratio at year-end was below the
historic year-end average of 16.8%, but was at the historical median of 9.9%.
We estimate that the face value of the distressed and defaulted debt markets
rose to $847.0 billion as of December 31, 2018, up 13% from $746.8 billion one
year earlier, primarily due to the increase in the distress ratio. The market value
estimate increased, as well, to approximately $477 billion from $414 billion one
year earlier.
Based on three different methodologies, our 2019 default rate forecasts range
from 2.28% (distress ratio method) to 4.20% (mortality rate method), with an
average rate of 3.46%, a level above the historic average of 3.3%. If we equally
weight our mortality rate and market-level measures, the forecasted default rate
rises slightly to 3.65%.
4
40
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Defaults, Default Rates, and Recoveries
High-yield bond default rates on U.S., Canadian and Mexican high-yield bonds
decreased slightly in 2018, remaining below historic averages. The rate decreased
from 1.81% at year-end 2017 to 1.74% for all of 2018. Defaults include straight
corporate bonds whose firms went bankrupt, missed an interest payment and did not
cure it within the grace or forbearance period, or completed a distressed exchange.
The 2018 rate is based upon a mid-year market size of $1.66 trillion, up by $42
billion from a year earlier. In all, $29.0 billion of defaults were recorded in 2018.
Note in Figure 1 that the historical weighted-average annual default rate is 3.27%
over the 48-year period (1971-2018). This weighted-average rate is down compared
to 3.38% at the end of 2017. The last year that the historical weighted average
default rate was above 4.0% was 2010. Our weights are based on the par value of
high-yield bonds outstanding in each year. The historical arithmetic unweighted
annual average default rate fell to 3.08% from 3.10% one year earlier. Of interest, the
2018 default rate was about the same level as the median annual rate (1.77%) over
the last 48 years.
The fourth-quarter 2018 default rate was 0.38%, slightly lower than one year (0.45%),
but higher than one quarter (0.08%), prior. This marks the third consecutive quarter in
which quarterly default rates were lower than 0.50%. Between 1Q 2010 and 4Q 2014,
quarterly default rates were below 0.50% in seventeen out of twenty quarters. In fact,
during this period, defaults remained below 0.50% for seven consecutive quarters –
from 1Q 2010-3Q 2011. Since 1989, there has been one longer, consecutive quarterly
period of default rates also below 0.5% -- nine from 1Q 2006 to 1Q 2008 (Figure 2 and
Appendix A).
Twelve issuers defaulted in the fourth-quarter 2018 on 28 issues, representing 29% of
all defaulting issuers and 26% of all issues defaulting in 2018. In all, 106 issues from 41
issuers defaulted in 2018 (Appendix B), compared to 108 issues and 60 issuers in 2017.
The average dollar amount of defaulting bonds per defaulting issuer in 2018 was $707
million, compared to $448 million in 2017, and $773 million in 2016. The most sizeable
total bond defaults during the year were those of iHeart Communications, Inc., with
defaults of $9.1 billion, Claire’s Stores, Inc., with defaults of $1.8 billion, Sears Holding
Corp., with defaults of $1.3 billion, Windstream Services, with defaults of $1.0 billion,
Southeastern Grocers, LLC., with defaults of $946 million, Cenveo Corp., with defaults
of $935 million and Ultra Resources Inc. with defaults of $780 million. The remaining
defaulted issuers in 2018 each had total bond defaults of less than $750 million. Energy
sector defaults had the largest number (14 of the 41) of defaulting issuers of any sector.
In our default statistics, we include those bonds from distressed exchanges actually
tendered. For example, in the Ultra Resources exchange, $780 million of bonds were
exchanged of the $1.2 billion outstanding and subject to the exchange offer. In 2018,
there were 20 distressed exchanges, involving 14 companies, comprising $5.62 billion
of defaults (19.4% of the total). See Appendix C for the list of 2018 distressed
exchanges and later our discussion of these restructurings.
In 2018, S&P and Moody’s issuer-denominated default rates were 2.4%
(preliminary) and 2.82%, respectively. Moody’s 2.16% dollar-denominated default
41
5
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
rate was lower than its issuer-denominated rate. As we have shown in past research2
and indicated in past reports, when the credit markets are in a strong and benign
state, dollar-denominated bond default rates typically fall to lower levels than do
issuer rates. The reverse happens in highly stressed conditions, when the dollar rate
usually exceeds issuer rates. Hence, dollar default rates are more volatile, less easily
modeled, but still quite relevant, especially to investors who concentrate on
investments of specific sizes that are not average for the market. Due to several, very
large, defaulting entities in 2014 and 2015, namely Energy Future Holdings (EFH)
and Caesars Entertainment Operating (CZR), neither high-yield bond nor leveraged
loan default rates in those years behaved in the manner typical of a benign period, as
dollar default rates were higher than those calculated on an issuer basis. This
reversed in 2016, however, when issuer default rates exceeded those of dollardenominated rates. This more typical comparison continued through the end of 2018.
Fitch’s dollar-denominated default rate for 2017 was 2.39%. The rating agencies’
default rate calculations relate only to those bonds rated by their respective agencies,
which is most, but not all, issues.
The S&P/LSTA 12-month, dollar-denominated default rate on leveraged loans
decreased from 2.05% at the end of 2017 to 1.63% by year-end 2018, while the
issuer default rate fell from 1.72% to 1.56% (Figures 3a and 3b). Typically, during
benign credit cycles, the issuer denominated default rate on loans and bonds exceeds
the dollar denominated rate. In the majority of months since March 2013, however,
the dollar-denominated loan default rate has exceeded the issuer-denominated rate
due to relatively large defaults by only a few issuers. However, the reverse was true
for all of 2016. Seventeen leveraged loans issuers defaulted in 2018 compared to 25
one year earlier.
Leveraged loans, according to S&P, are institutional and “pro rata” (syndicated to
banks and finance companies), secured loans issued by non-investment grade
companies. As such, their index does not include the rare case of a fallen-angel
secured loan unless that loan became secured in a distressed exchange, but was
unsecured when originally issued.
According to our comparison between high-yield bond defaults (Appendix B) and
leveraged loan defaults (Appendix D), eight firms, American Tire Distributors, Inc.,
Claire’s Stores, Inc., David’s Bridal, Inc., iHeart Media, Inc., Nine West Holdings,
Inc., Remington Outdoor Co., Sears Holdings Corp. and Westmoreland Coal Co.
defaulted on both bonds and institutional leveraged loans in 2018.
See Figure 4 for the association between dollar-denominated bond default rates and
economic recessions in the U.S. since the early 1970’s, including the most recent
recession that ended in mid-2009. As usual, we see the default rate peaking at or near
the end of the recession, although we observed the peak before it was confirmed that
the recession had indeed ended in June 2009.
Our forecast for 2019 is for a high-yield bond default rate of 3.46%, higher than
our forecasts for the prior two years, and above the historic average of 3.3%. If
we equally weight our mortality rate and market-level measures, the forecasted
2
High Yield Bonds: Default and Loss Rate Comparison – Mid-Cap Versus Large-Cap Issuers, M. Verde, P. Mancuso and E. Altman
November 11, 2005, Fitch.
6
42
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
default rate rises slightly to 3.65%. This forecast assumes the U.S. economy will
not be in a recession in 2019. However, more than 50% of economists currently
believe that the U.S. will be in a recession by the end of 2020; thus, our forecast
a year from now may include an additional measure for forecasting 2020’s
default rate. There are several important risks on the horizon associated with this
forecast (see discussion later on the benign credit cycle) which have not been
reflected in higher-than-average yield spreads by high-yield bond investors in
some time, excluding a two-week period from December 20, 2018 through
January 04, 2019, but we have many concerns about the intermediate term
(greater than one year) horizon. We will discuss this later in our report.
43
7
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 1. Historical Default Rates — Straight Bonds Only, Not Including Defaulted Issues
In Par Value Outstanding, 1971–2018 (Dollars in Millions)
Year
2018
2017
2016
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
1979
1978
1977
1976
1975
1974
1973
1972
1971
Par Value
Outstandinga ($)
Par Value
Defaults ($)
1,664,166
1,622,365
1,656,176
1,595,839
1,496,814
1,392,212
1,212,362
1,354,649
1,221,569
1,152,952
1,091,000
1,075,400
993,600
1,073,000
933,100
825,000
757,000
649,000
597,200
567,400
465,500
335,400
271,000
240,000
235,000
206,907
163,000
183,600
181,000
189,258
148,187
129,557
90,243
58,088
40,939
27,492
18,109
17,115
14,935
10,356
8,946
8,157
7,735
7,471
10,894
7,824
6,928
6,602
28,994
29,301
68,066
45,122
31,589
14,539
19,647
17,963
13,809
123,878
50,763
5,473
7,559
36,209
11,657
38,451
96,858
63,609
30,295
23,532
7,464
4,200
3,336
4,551
3,418
2,287
5,545
18,862
18,354
8,110
3,944
7,486
3,156
992
344
301
577
27
224
20
119
381
30
204
123
49
193
82
Default
Rates(%)
a As of midyear. b Weighted by par value of amount outstanding for each year.
Source: NYU Salomon Center.
8
Historical Averages (%)
Std. Dev. (%)
1.742 Arithmetic Average Default Rate
1.806
1971 to 2018
3.076
4.110
1978 to 2018
3.270
2.827
1985 to 2018
3.699
2.110
Weighted Average Default Rat273
1.044
1971 to 2018
3.273
1.621
1978 to 2018
3.276
1.326
1985 to 2018
3.287
1.130
Median Annual Default Rate
10.744
1978 to 2018
1.774
4.653
0.509
0.761
3.375
1.249
4.661
12.795
9.801
5.073
4.147
1.603
1.252
1.231
1.896
1.454
1.105
3.402
10.273
10.140
4.285
2.662
5.778
3.497
1.708
0.840
1.095
3.186
0.158
1.500
0.193
1.330
4.671
0.388
2.731
1.129
0.626
2.786
1.242
44
2.981
3.131
3.249
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 2.
1989–2018
Quarterly and the Four-Quarter Moving Average Default Rate
6.0%
16.0%
14.0%
5.0%
12.0%
4 - Quarter Moving Average
4.0%
Quarterly Default Rate
10.0%
3.0%
8.0%
6.0%
2.0%
4.0%
1.0%
2.0%
0.0%
0.0%
Quarterly
Moving
Source: NYU Salomon Center.
Figure 3a. S&P Leveraged Loan Index 12-Month Moving Average Default
Rate 1998–2018 (Number of Issuers)
9%
8%
7%
6%
5%
4%
3%
2%
1%
Dec-18
Jun-18
Jun-17
Dec-17
Jun-16
Dec-16
Jun-15
Dec-15
Jun-14
Dec-14
Jun-13
Dec-13
Jun-12
Dec-12
Jun-11
Dec-11
Jun-10
Dec-10
Jun-09
Dec-09
Jun-08
Dec-08
Jun-07
Dec-07
Jun-06
Dec-06
Jun-05
Dec-05
Jun-04
Dec-04
Jun-03
Dec-03
Jun-02
Dec-02
Jun-01
Dec-01
Jun-00
Dec-00
Jun-99
Dec-99
0%
Dec-98
February 21, 2019
Source: S&P Global Market Intelligence.
45
9
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 3b. S&P Leveraged Loan Index 12-Month Moving Average Default
Rate 1998–2018 (Dollar Amount)
12%
10%
8%
6%
4%
2%
Jun-18
Dec-18
Jun-17
Dec-17
Jun-16
Dec-16
Jun-15
Dec-15
Jun-14
Dec-14
Jun-13
Dec-13
Jun-12
Dec-12
Jun-11
Dec-11
Jun-10
Dec-10
Jun-09
Dec-09
Jun-08
Dec-08
Jun-07
Dec-07
Jun-06
Dec-06
Jun-05
Dec-05
Jun-04
Dec-04
Jun-03
Dec-03
Jun-02
Dec-02
Jun-01
Dec-01
Jun-00
Dec-00
Jun-99
Dec-99
Dec-98
0%
Source: S&P Global Market Intelligence.
Figure 4. Historical Default Rates and Recession Periods in the US
High-Yield Bond Market, 1972–2018
14.0%
12.0%
10.0%
8.0%
6.0%
4.0%
2.0%
18
16
14
12
10
08
06
04
02
00
98
96
94
92
90
88
86
84
82
80
78
76
74
72
0.0%
Periods of Recession: 11/73–3/75, 1/80–7/80, 7/81–11/82, 7/90–3/91, 4/01–12/01, 12/07–
6/09.
Sources: Figure 1 of this report and National Bureau of Economic Research.
Bankruptcies in 2018 and Trends in Bankruptcy
As can be seen in Figure 5, the amount of total liabilities for Chapter 11 bankruptcies
in 2018 was $99.1 billion, based on 92 filings. This data, from the NYU Salomon
Center database, does not include filings with less than $100 million in liabilities, but
does include private firms. The total number of filings, for bankruptcies with
liabilities greater than or equal to $100 million, increased marginally from 91 at yearend 2017, and was higher than both the historical median annual number (76) from
1989-2018, and the median number since 1998 of 91 (Figure 6). These figures are
10
46
Altman-Kuehne High-Yield Bond Default and Return Report
not adjusted for inflation. iHeart Communications, Inc. was the largest bankruptcy
filing in 2018, with $20.3 billion in liabilities, followed by Sears Holding Corp.
($11.3 billion). Appendix F lists 2018’s Chapter 11 bankruptcies with liabilities
greater than $100 million.
Figure 5. Total Filings and Liabilitiesa of Public Companies Filing
for Chapter 11 Bankruptcy, 1989–2018
Pre- Petition Liabilities, in $ billions (left axis)
Median Liabilities
Number of Filings (right axis)
Median No. of Filings.
$800
280
$700
240
2017
$600
$ Billion
200
$500
160
$400
120
$300
80
$200
91 filings and
liabilities of
$121.1 billion
2018
92 filings and
liabilities of
$99.1 billion
40
$100
0
$0
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
February 21, 2019
a Minimum $100 million in liabilities.
Sources: Appendix F and the NYU Salomon Center Bankruptcy Filings Database.
The number of billion-dollar bankruptcies in 2018 decreased from 24 in the prior
year to 21, but was still 1.6 times the average over the 39-year period (1980-2018) of
13 (Figure 6). In 2018, billion-dollar bankruptcies represented 23% of all bankruptcy
filings with liabilities greater than $100 million, also higher than the median yearly
percentage of 21% from 1989-2018 (Figure 7). Appendix E lists all billion-dollar
bankruptcies from 2008-2018. Note that excluding Lehman Brothers, the average
amount of total liabilities at filing over this nine-year period for our billion-dollar
“babies” was $5.2 billion (median = $1.9 billion).
According to New Generation Research, the number of public companies filing for
bankruptcy in 2018 was 56. The average of total liabilities for our public and private
large (≥$100 million) company bankruptcy filings (92) in 2018 was $1.07 billion,
down slightly from $1.33 billion one year earlier, and below the historical average
since 1980 of $1.53 billion (Figure 6).
47
11
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 6.
Year
2018
2017
2016
2015
2014
2013
2012
2011
2010
2009
2008
2007
2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
1989
1988
1987
1986
1985
1984
1983
1982
1981
1980
Total
Historical Bankruptcy Filings 1980–2018
Total
Average
Total Filingsb
Liabilitiesb
Liabilitiesb
c
($ MN)
(>$100 Total Filings
($ MN)
Million) (≥$1 Billion) (>100 Million) (>100 Million)
Total Filingsa
56
71
99
79
54
71
87
86
106
211
138
78
66
86
93
176
229
265
187
145
122
83
86
85
70
86
91
123
115
135
122
112
149
149
121
89
84
74
62
4341
92
91
98
70
59
66
69
84
114
232
146
38
34
36
45
102
136
170
137
107
55
36
33
32
24
37
38
53
35
23
14
12
13
14
12
14
12
6
4
2393
21
24
37
20
14
11
14
7
14
49
24
8
4
11
11
26
41
39
23
19
6
5
1
7
1
5
14
12
10
10
2
1
3
2
0
3
3
1
0
503
99115.8
121079.4
125304.6
80291.3
91992.4
39480.2
71612.6
109118.9
56834.8
604269.9
724222.4
72338.4
22774.8
142950.3
40099.7
115171.8
338175.6
229860.9
99091.0
70516.0
31913.4
18865.9
11949.0
27153.0
8396.0
17701.1
64676.5
82423.6
41115.1
34516.1
6905.0
25421.0
9830.3
8605.2
3440.0
13674.0
7113.0
3960.0
746.0
3,672,704.9
1077.3
1330.5
1278.6
1147.0
1559.2
598.2
1037.9
1299.0
498.6
2604.6
4960.4
1903.6
669.8
3970.8
891.1
1129.1
2486.6
1352.1
723.3
659.0
580.2
524.1
362.1
848.5
349.8
478.4
1702.0
1555.2
1174.7
1500.7
493.2
2118.4
756.2
614.7
286.7
976.7
592.8
660.0
186.5
1,534.8
a Represents both Chapter 7 (8) and 11 (48) public company filings in 2018 (Source: New
Generation Research).
b Filings with Total Liabilities greater than $100 million. Includes some private company
filings (Source: NYU Salomon Center Bankruptcy Filings Database).
C Filings with Total Liabilities greater than $1 billion. Can include private company
filings (rare). (Source: NYU Salomon Center Bankruptcy Filings Database and New
Generation Research).
12
48
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 7.
Chapter 11 Filing Statisticsa
Year
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Mean No. of Filings,
1989-2018
Median No. of
Filings, 1989-2018
Median No. of
Filings, 1998-2018
Mean Liabilities,
1989-2018
Median Liabilities,
1989-2018
Number
of
Filings
23
35
53
38
37
24
32
33
36
55
107
137
170
136
102
45
36
34
38
146
232
114
84
69
66
59
70
98
91
92
Pre-Petition
Liabilities
($ millions)
34,516
41,115
82,424
64,677
17,701
8,396
27,153
11,949
18,866
31,913
70,516
99,091
229,861
338,176
115,172
40,100
142,950
22,775
72,338
724,222
604,270
56,835
109,119
71,613
39,480
91,992
80,291
125,305
121,079
99,116
Number of
Filings ≥
$1B
10
10
12
14
5
1
7
1
5
6
19
23
39
41
26
11
11
4
8
24
49
14
7
14
11
14
20
37
24
21
≥$1B/Total
Filings (%)
43
29
23
37
14
4
22
3
14
11
18
17
23
30
25
24
31
12
21
16
21
12
8
20
17
24
29
38
26
23
76
16
21
63
13
21
91
19
$119,767
$71,976
a Minimum $100 million in liabilities.
Source: NYU Salomon Center Bankruptcy Filings Database.
The issue of bankruptcy trends and their impact on the entire corporate bankruptcy
system is, of course, more complicated than simply the number and dollar value of
filings. For example, the time spent in reorganization from filing to emergence, the
number and impact of prepackaged Chapter 11 filings, the success, or not, of the
reorganization, as well as the role of senior creditors and D.I.P. lenders, are all
factors that need further analysis and commentary. Indeed, conclusions and
suggestions for revisions to the Bankruptcy Code can be observed in the American
Bankruptcy Institute report issued four years ago on December 8, 2014.3
One aspect of bankruptcy trends that disturbs us is the frequency of multiple-filings
by the same entity, i.e. the recidivism phenomenon. Figure 8 shows that from 1984-
3
After about two years of evaluation of the U.S. Bankruptcy Code, the ABI released its report on the Reform of Chapter 11 on
December 8, 2014. Please see for the complete document.
49
13
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
2018, 299 firms have filed twice (Chapter 22), 21 firms have filed three times
(Chapter 33), three firms filed four times (Chapter 44) and one filed Chapter 55!
Through 2018 (Figure 9), there were 324 multiple filings, 8% of total bankruptcy
filings for the same time period. Importantly, an estimated 21% of all firms emerging
from the bankruptcy process as a “going concern,” have subsequently refiled. We
believe this percentage is too high, and we proposed a partial remedy in a paper
published in early 2015.4
In Figure 10, we compare the date of default with the Chapter 11 filing date for firms
that defaulted on bonds and went bankrupt, going back to 1981. Based on 1,009
observations from the NYU Salomon Center Master Default and Bankruptcy
Databases, both events occurred on the same date in 495 instances (49%). In the
remaining 51% of the cases, the lag between the default date and bankruptcy date
varied considerably, with decreasing levels as the two dates became further separated
from each other. Of course, some defaulting issuers never formally file for
bankruptcy as their problems are settled out of court or the default is a result of a
distressed exchange (DE), and they do not file for bankruptcy in subsequent years,
(more than half). DEs do, however, subsequently file– in many cases – see our
discussion at a later point.
For more comprehensive statistics on bankruptcy trends in the U.S., please see our
2013 annual report5 as well as the publication of the “Hedge Funds in Bankruptcy”6
conference proceedings in the ABI Law Review (April 2014).
4
E. Altman and B. Branch, “The Bankruptcy System’s Chapter 22 Recidivism Problem: How Serious is it?”, The Financial Review,
Vol. 50, No. 1, 2015. Also see Professor Altman’s website.
5
E. Altman and B. Kuehne “Defaults and Returns in the High-Yield Bond and Distressed Debt Market: The Year 2013 in Review
and Outlook”, Paulson & Co., February 6, 2014, and the NYU Salomon Center Special Report, February, 7, 2014.
6
14
Held at St. John’s University Law School, October 04, 2013 (K. Sharfman, coordinator).
50
February 21, 2019
Altman-Kuehne High-Yield Bond Default and Return Report
Figure 8.
Year
Chapter 22’s, 33’s, 44’s and 55’s in the U.S. (1984–2018)
Chapter
22’s
1984-89
18
1990
10
1991
9
1992
6
1993
8
1994
5
1995
9
1996
12
1997
5
1998
2
1999
10
2000
12
2001
17
2002
11
2003
17
2004
...
Purchase answer to see full
attachment