Altman (2020)

Journal of Credit Risk 16(2), 1–28
DOI: 10.21314/JCR.2020.262
Copyright Infopro Digital Limited 2020. All rights reserved. You may share
using our article tools. This article may be printed for the sole use of the
Authorised User (named subscriber), as outlined in our terms and condi-
tions. https://www.infopro-insight.com/termsconditions/insight-subscriptions
Research Paper
Covid-19 and the credit cycle

Edward I. Altman
Max L. Heine Professor, Emeritus, Salomon Center, NYU Stern School of Business,
44 West 4th Street, Suite 9-160, New York, NY 10012, USA;
email: [email protected]
(Received April 6, 2020; revised May 17, 24, 2020; accepted May 28, 2020)
ABSTRACT
The Covid-19 health crisis has dramatically affected just about every aspect of the
economy, including the transition from a record long benign credit cycle to a stressed
one, with still uncertain dimensions. This paper seeks to assess the credit climate
from just before the unexpected global health crisis catalyst to its immediate and
extended impact. We analyze the performance of several key indicators of the nature
of credit cycles: default and recovery rates on high-yield bonds, and the number of
large firm bankruptcies that we expect over the next twelve months and beyond;
yield spreads and distress ratios; and liquidity. Our focus is primarily on the nonfi-
nancial corporate debt market in the United States, which reached a record percent-
age of gross domestic product at the end of 2019 as firms increased their debt to take
advantage of record low interest rates, and investor appetite grew for higher promised
yields on risky fixed-income assets. We also examine the leveraged loan and collater-
alized loan obligation markets, as well as the increasingly large and important BBB
tranche of the corporate bond market. Specifically, we discuss the latter’s vulnerabil-
ity to downgrades over the expected downturn in the real economy and this vulnera-
bility’s potential impact on expected default rates by “crowding out” low-quality debt
of other firms (some of which we believe are “zombies”). Using Z-scores for a sam-
ple of BBB companies between 2007 and 2019, we analyze this largest component
of the corporate bond market to provide some evidence on the controversial debate
Print ISSN 1744-6619 j Online ISSN 1755-9723

c 2020 Infopro Digital Risk (IP) Limited
1
2 E. I. Altman
as to whether there has been ratings inflation or, perhaps, persistent overvaluation of
the nonfinancial corporate debt market since the last financial crisis.
Keywords: Covid-19; credit cycle; high-yield bonds; default rates and recovery rates; rating
inflation; zombies.
1 INTRODUCTION
With global health concerns about the coronavirus dominating the news, this paper
gauges the financial health of the credit market both before and after the onset of the
Covid-19 pandemic. We begin with an examination of where we were in the credit
cycle during the pre-pandemic period, culminating at the end of 2019; at the end of
2019, the credit cycle was apparently in a benign state, albeit with some unmistakable
storm clouds on the horizon. By our definition, benign credit cycles are periods when
most, if not all, of four particular market conditions are incentivizing major growth
in the supply and demand for credit. That means three or more of the following:
(1) low and below-average default rates and forecasted rates;
(2) high and above-average recovery rates on actual defaults;
(3) low and below-average yields and spreads required from issuers by investors;
(4) highly liquid markets in which the riskiest credits can issue considerable debt
at low interest rates.
At the end of 2019, at least three of these signals indicated we were still in a benign
credit cycle, one that, assuming 2016 was an energy industry anomaly, was well into
its eleventh year. That is the longest benign cycle by far in the history of modern
finance. As of the end of 2019, the benign indicators were: a corporate high-yield
(HY) bond default rate of 2.87% (US dollar-denominated) compared with the his-
torical average of 3.3%; yield spreads required by investors in HY bonds and lever-
aged loans about 100 basis points (bps) below the historical average; and extremely
high liquidity in the risky debt segment. The recovery rate on defaulted bonds was
43.5%, slightly below the historical average of 46.0%. Indeed, during the first two
months of 2020, new HY issues were at record levels on top of new HY bond issues
that totaled US$250 billion in 2019, and newly issued leveraged loans amounted to
almost US$500 billion (Altman and Kuehne 2020).
Further, the stock market had just had an extremely profitable year, rising by
about 30% in 2019, with HY bond investors enjoying a return of about 14%. The
US economy was growing at a reasonably high rate, certainly at a level above most
Journal of Credit Risk www.risk.net/journals

Covid-19 and the credit cycle 3
of the developed world. The outlook for 2020 was still fairly rosy, despite the coron-
avirus ravaging the Wuhan area in China and some other countries. Indeed, the Fed’s
forecast for gross domestic product (GDP) growth in the United States was in the
2.0–2.2% range (Board of Governors of the Federal Reserve System 2019).
However, was this upbeat scenario indicative of a benign credit cycle or of a credit
bubble that created a false sense of security, even as risk built up in the system? Even
before the pandemic struck, there were signs of excess associated with a credit bubble
or a “risk-on” attitude in credit markets. For example, the amount of corporate bonds,
of both investment and noninvestment grades, in the United States had doubled from
2009 levels to more than US$9 trillion at the end of 2019. The largest growth in US
dollar amount was in the BBB-rating class, with almost US$3 trillion in marginally
investment-grade bond issues. And, with a similar growth in leveraged loans to over
US$1.2 trillion, most without any meaningful protective covenants for investors, and
with historically low interest rates and even the lowest quality, CCC-rated issues eas-
ily refinanced with ample new issues of at least US$20 billion each year from 2014
to 2019 (Altman and Kuehne 2020), most indicators were of a “risk-on”, low default
rate scenario. In addition, nonbank lending to commercial borrowers, mostly lever-
aged buyout companies, exploded to an estimated 42% of all commercial lending,
amounting to almost US$1 trillion (Bank of America Merrill Lynch (2019) estimate).
In short, we believe that corporate and government debt was increasing enormously
to perhaps dangerous levels, almost without pause as of the end of 2019 and into
the first two months of 2020; this, despite the number of Chapter 11 and Chapter 7
bankruptcy filings with liabilities above US$100 million spiking in 2019 to a total of
ninety-eight, the highest amount since 2009 (except for 2016). Further, the number of
billion dollar bankruptcies increased from twenty-one in 2018 to twenty-six, which
was almost double the median (fourteen) over the thirty-year period from 1989 to
2019. The continuing issuance of high risk debt despite these ominous signs points
to a debt bubble.
Skeptics of the debt bubble theory assert that if the levels of corporate debt were
considered relative to equity measured in terms of market values instead of book
values, the corporate debt level was actually lower than it was ten years ago. While
that is true, Figure 1 shows that if you simulate the debt/equity ratio with a 20–40%
decline in market equity values, the levels in 2019 would be the highest in modern
cycles, with the exception at the height of the global financial crisis (GFC) in 2008.
A related storm cloud over the pre-pandemic horizon involved the US levels of
nonfinancial corporate debt (NFCD) as a percentage of GDP. Figure 2 shows this
percentage from 1987 to 2019, with three peaks in that ratio over this sample period
(1990–91 (43%), 2001–2 (45%) and 2008–9 (45.2%)). Also shown in Figure 2 are
the levels of HY bond default rates over the same sample period. Note that peaks in
the NFCD/GDP ratio were followed within twelve months or less by peaks in the
www.risk.net/journals Journal of Credit Risk

4 E. I. Altman
FIGURE 1 US total nonfinancial corporate debt as a proportion of GDP and market cap
of equity.
Total corporate debt*/total equity market cap Total corporate debt*/GDP

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
Scenario 1
Scenario 2
Scenario 3
Scenarios 1–3 assume equity market cap lower by 20%, 30%, 40%, respectively. Debt and equity do not include
financial firms. Market cap includes New York Stock Exchange and Nasdaq companies. Sources: Bank of America,
Federal Reserve Economic Data, World Federation of Exchanges, Volatility and Risk Institute, NYU Stern.
default rate three times during the period from 1987 to 2019. In 2019, we observed
a new peak at the highest ever level (47%) of NFCD/GDP. Yet the default rate at the
end of 2019 was still below average. Would that low-risk default rate continue into
2020 and beyond? Figure 3 suggests that without a recession the benign credit cycle
may have continued; that is, without the Covid-19 economic downturn, we could still
have been in an environment of low default rates, high recovery rates, low yields and
liquid debt markets. Was the benign credit cycle another victim of the coronavirus?
Our analysis shows that the benign credit cycle was dying even before the pandemic
levied the death blow.
2 CREDIT AND ECONOMIC CONDITIONS DURING THE GLOBAL

PANDEMIC
We will never know what would have happened to the credit cycle if the pandemic
had not triggered economic collapse. At the time of writing (May 2020), most
economists were predicting that the Covid-19 crisis would likely result in unem-
ployment in the United States reaching perhaps 20–25% in Q2, with an economic
recession in the United States, and globally, by the end of 2020 Q3. Indeed, both
Morgan Stanley (predicting a 30% drop in GDP in Q2) and Goldman Sachs (24%

FIGURE 2 US nonfinancial corporate debt (credit market instruments) to GDP: com-

parison with four-quarter moving average default rate, January 1, 1987 to December 31,
2019.
%NFCD to GDP Four-quarter moving average default rate 16

48
47
% NFCD to GDP (quarterly)
14
average default rate (%)

46
Four-quarter moving
45 12
44 10
43
8
42
41 6
40 4
39
2
38
37 0
Jan 1987
Jan 1988
Jan 1989
Jan 1990
Jan 1991
Jan 1992
Jan 1993
Jan 1994
Jan 1995
Jan 1996
Jan 1997
Jan 1998
Jan 1999
Jan 2000
Jan 2001
Jan 2002
Jan 2003
Jan 2004
Jan 2005
Jan 2006
Jan 2007
Jan 2008
Jan 2009
Jan 2010
Jan 2011
Jan 2012
Jan 2013
Jan 2014
Jan 2015
Jan 2016
Jan 2017
Jan 2018
Jan 2019
Jan 2020
Sources: Federal Reserve Economic Data, Federal Reserve Bank of St Louis and Kroll Bond Rating Agency/Altman
HY default rate data.
drop) forecasted in March an annualized recession for several more quarters, if not
years, with the highest unemployment rate since the Great Depression. JP Morgan’s
forecast in March for a recession in 2020 was 55% (JP Morgan 2020). This large
expected downturn was not just in the United States. Most foreign analysts expected
China to have reduced growth in 2020 even before the awareness of Covid-19, and
many parts of Europe were already in a recession, including those that depend on
China to buy their goods, such as Germany.
However, the buoyant but fragile credit markets that we observed at 2019 year-end
continued their “risk-on” market confidence until early March 2020, despite ominous
warnings about the virus in China and some other countries. Perhaps the catalyst
for a change in the credit cycle, from benign to distressed, could have come as a
result of China’s GDP decline even before the coronavirus crisis became evident.
However, the proliferation of the virus on a global scale was clearly the catalyst
for a major shift in the market environment. Yield spreads that were 100 bps below
average as of year-end 2019 had spiked by more than 150 bps by March 6, 2020.
In the fortnight that followed, spreads doubled to over 1000 bps. New issues in the
leveraged finance market that were setting monthly records in early 2020 essentially
dried up in mid-March, with firms postponing new debt issues due to much higher
investor-required interest rates. The distress ratio (HY bonds trading at more than

6 E. I. Altman
FIGURE 3 Historical default rates, benign credit cycles and recession periods in the US
HY bond market (1972–2019).
14
5 yr 7 yr 7 yr 4 yr 10+ yr
12
10
8
%
6
0
1972
1974
1976
1978
1980
1982
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
Periods of recession: November 1973 to March 1975, January 1980 to July 1980, July 1981 to November 1982,
July 1990 to March 1991, April 2001 to December 2001, and December 2007 to June 2009. Benign credit cycles
are approximated. All rates annual. Sources: National Bureau of Economic Research and author data.
1000 bps over Treasuries) jumped from 8.2% as of the end of 2019 to almost double
that level by early March, approaching the historical average. Indeed, the distress
ratio actually reached 40% in late March. Returns on HY bonds went from C1:5%
in the first months of 2020 to 14% in late March. These declines were only the
beginning of the negative trends, which became much more severe by the end of
March 2020. Further, the stock market’s enormous decline caused the declaration of
a bear market (20% decline).
The stock market and most (but not all) risky debt markets rebounded strongly
in April 2020, mainly due to the Fed and the US government’s enormous support
and the hope for an early end to the health crisis. However, the post-realization of
Covid-19 has produced a bifurcation in the markets for leveraged loans versus cor-
porate bonds. Both markets had immediate large declines in prices and consequent
increases in yield spreads in March, but thanks to unprecedented Fed support for the
corporate bond market, the HY bond market regained much of its losses by mid-May
2020. HY bond spreads settled at about 750–800 bps, about 250 bps above its histor-
ical average, and liquidity resumed its early-2020 record levels of new issues, albeit
at higher interest rates than before Covid-19. Leveraged loans, on the other hand,
have continued to languish since the declines of March 2020, and yield spreads have
remained high as the bank loan and collateralized loan obligation (CLO) markets

lagged other asset classes. Despite the enormous support of the Fed and the US Trea-
sury extended to the corporate bond market, new issues of leveraged loans were very
low during April and May 2020, falling to perhaps one-third of the normal monthly
amounts over the past five years (LCD News 2020).
One interpretation of the cause of differences in the performance of leveraged
loans versus HY bonds is that the CLO market is constrained in its ability to add
new leveraged loans to existing structures due to increased downgrades to CCC
and defaults in the existing pools of collateral loans. Since CLOs had purchased
as much as 70% of new leveraged loans from banks in recent years, this constraint
caused banks to reduce their own lending, especially to high risk borrowers. Thus, the
“risk-on” bubble conditions that were prevalent before the coronavirus outbreak have
contributed to the depressed credit conditions in the leveraged loan market during
2020.
A key question is whether the corporate bond market or the leveraged finance mar-
ket provides the better early warning of economic conditions on the horizon. Prior
research suggests that the loan market provides an earlier and clearer forecast of
economic recovery than does the bond market. Altman et al (2010) analyzed loan
prices versus bond prices in the secondary market for a large sample of defaulted
companies that had both sources of debt outstanding. The evidence was conclusive
that loan price movements declined significantly earlier than bond prices. Further,
the LCD News (2020) report noted above stated that “leveraged loan performance
has been a leading indicator of when an economic recovery has begun, much more
so than equity prices”. Finally, the dramatic recovery of corporate bond prices can be
attributed to Fed intervention rather than indicators of fundamental economic condi-
tions. Thus, analysis of current credit market indicators suggests that the US econ-
omy will remain in a deep recession for longer than the HY bond and equity markets
are indicating.
3 FORECASTING COVID-19 DEFAULT AND RECOVERY RATES

AND BANKRUPTCIES
Against this backdrop, we update our forecasts of credit conditions over the next
twelve months. We estimate default rates on US dollar-denominated North American
HY bonds using three different methods: the mortality rate approach, the required
yield spread demanded by market investors and the distressed ratio method.
Mortality rate analytics (based on Altman (1989)) have been maintained and
updated annually for thirty years. This actuarial technique records the frequency of
default of newly issued bonds from every major rating category, including investment
and noninvestment grades, for one to ten years after issuance. Our latest estimates
cover 3578 corporate bond defaulting issues from 1971 to 2019 (Table 1). A similar

8
E. I. Altman
TABLE 1 Mortality rates (%) by original rating: all rated corporate bonds, 1971–2019.
Journal of Credit Risk

Years after issuance
‚ …„ ƒ
1 2 3 4 5 6 7 8 9 10
AAA Marginal 0.00 0.00 0.00 0.00 0.01 0.02 0.01 0.00 0.00 0.00
Cumulative 0.00 0.00 0.00 0.00 0.01 0.03 0.04 0.04 0.04 0.04
AA Marginal 0.00 0.00 0.16 0.04 0.02 0.01 0.03 0.03 0.03 0.04
Cumulative 0.00 0.00 0.16 0.20 0.22 0.23 0.26 0.29 0.32 0.36
A Marginal 0.01 0.02 0.08 0.09 0.07 0.03 0.02 0.21 0.05 0.02
Cumulative 0.01 0.03 0.11 0.20 0.27 0.30 0.32 0.53 0.58 0.60
BBB Marginal 0.28 2.23 1.19 0.94 0.47 0.19 0.20 0.20 0.18 0.30
Cumulative 0.28 2.50 3.66 4.57 5.02 5.20 5.39 5.58 5.75 6.03
BB Marginal 0.88 2.11 3.77 1.94 2.36 1.50 1.40 1.05 1.36 3.05
Cumulative 0.88 2.97 6.63 8.44 10.60 11.94 13.18 14.09 15.26 17.84
B Marginal 2.82 7.60 7.70 7.70 5.70 4.42 3.66 2.01 1.68 0.68
Cumulative 2.82 10.21 17.12 23.50 27.86 31.05 33.57 34.91 36.00 36.44
CCC Marginal 8.03 12.35 17.64 16.17 4.85 11.56 8.37 4.74 0.59 4.20
Cumulative 8.03 19.39 33.61 44.34 47.04 53.16 57.08 59.12 59.36 61.07
Rated by Standard & Poor’s at issuance. Based on 3578 issues. Sources: S&P Global Ratings and author’s compilation.
www.risk.net/journals
analysis and compilation is done for mortality losses and can be used to estimate loss
given default (LGD), which includes our observations of recovery rates on defaulting
issues. These mortality statistics can be used to forecast default rates or probability
of default (PD). The technique involves the impact of bond aging by adjusting the
base population over time for other nondefault bond disappearances, such as bonds
“called” by the issuer, maturities or merger-related activities. Thus, if we observe the
US dollar amount of new issues by rating category for the past ten years and apply
the marginal mortality rate estimates from Table 1, we can aggregate the amount of
defaults in a subsequent year and then divide that into the forecasted population of
HY bonds (as of the mid-year of the next twelve months) to obtain our first forecast
of the annual default rate over the next twelve months.
Using the above mortality methodology, our forecast for the next twelve months
as of December 31, 2019, was 5.75%. Note that we aggregate estimates based on all
initial ratings, even investment-grade bonds. Since the last fifty years of default and
new issuance data do not include a pandemic environment, and the number of crisis
years was only about six out of fifty, our forecast of 2020 defaults will probably be
on the low side. We prefer, however, not to ignore this actuarial method, especially
since our other two methods do incorporate expectations based on current market
conditions.
Our second and third techniques rely on the current yield spread in the market
compared with ten-year Treasury bonds and the distress ratio. We started using the
ten-year US Treasury bond benchmark before the market adopted a similar method,
called the option-adjusted spread (the results of the two are very similar). The yield
spread method observes the historical annual relationship between current (time-t)
yield to maturity spreads and a default rate on HY bonds in t C 1 (one year in the
future). We update results annually, and the latest regression estimate is based on data
from 1978–2018 yield spreads and 1979–2019 defaults, resulting in the following
default rate estimate equation:
default rate.t C 1/ D 3:15 C 1:28.yield spread.t //; adjusted R2 D 59:6%:
Plugging in the yield spread of 9.84%, as of March 26, 2020, results in a next-
twelve-month forecast default rate of 9.45%. Note that due to extreme market volatil-
ity in late March and April 2020, the yield to maturity spread fluctuated from as low
as 7.5% to about 11.0%, so our forecasted default rate is likewise volatile, depending
on when the data is accessed.
Our final technique is the so-called distress ratio method, a measure we developed
(Altman 1990) to assess the segment of the HY market that is most likely to default
should either specific firms’ conditions and/or the real economy deteriorate signifi-
cantly. Under these circumstances, default rates, in general, increase. We originally

10 E. I. Altman
used the benchmark of 10% above the ten-year T-bond rate as our distress ratio crite-
rion, but we have now adopted the market standard of 1000 bps above the comparable
duration Treasury rate (the option-adjusted spread). Since 2000, this distress ratio’s
median annual rate has been 10.35%, with an average ratio of 16.38%. This ratio has
been as high as 81.2% (in December 2008) and as low as 1.62% (in December 2006).
Based on market data for 2000–18 for the distressed ratio and 2000–19 for default
rates, our linear regression estimated equation is
default rate.t C 1/ D 0:923 C 0:240.distress ratio.t //; adjusted R2 D 75%:
R2 was not increased by using nonlinear estimates.

Plugging in the distress ratio of 33% as of March 26, 2019 (the ratio reached as
high as 40.0% earlier in March and as low as 22% in early May), our PD estimate
for the next twelve months is 8.84%. Averaging the three methods, our preliminary
forecasted default rate for the twelve-month period as of March 2020 is 8.01%, which
is about 2.4 times the comparable rate in 2019. Summarizing the results of the three
models yields
mortality rate D 5:75%;

yield spread D 9:45%;
distress ratio D 8:84%;
average D 8:01%:
Finally, we have added a new element to our 2020 post-pandemic forecast, based on
the huge increase in triple-B-rated debt and the likely “crowding-out” effect caused
by the almost certain increase in downgrades to “junk” status, ie, fallen angels. These
downgrades, we posit, will have a negative impact on marginal firms’ ability to sur-
vive in a downturn (see further discussion in Section 3). This new factor, not consid-
ered in our historical time series models, will add perhaps an additional 1% to our
forecasted default rate, bringing the forecast for 2020 to about 9%. This implies that
the forecasted total US dollar amount of defaults would be US$135 billion in the
post-pandemic period. For comparison, defaults in 2009 totaled US$123 billion.
Further, the estimated population of HY bonds at the beginning of 2020 was
US$1.5 trillion. As of May 2020, the HY population had already swelled a fair
amount based on increases from fallen angels and a significant amount of new
issues, especially in January/February and April/May, even net of defaults. Admit-
tedly, the triple-B-related element added to the forecast default rate is somewhat
arbitrary, but to ignore it would, we feel, be an oversight. Forecasting an increase
in defaults due to the crowding-out effect of BBB downgrades is a tricky exercise.

We are cognizant that both the numerator and denominator in the default rate cal-
culation will be increased by fallen angels. Already, between March and mid-May
2020, about 18 BBB issuers were downgraded to junk-bond status, amounting to
about US$150 billion face value. So, while there will no doubt be an increase in
default amounts due to any crowding-out effects, it is not clear if the default rate will
increase; it depends on the timing of the downgrades and the consequent increase
in the HY bond population compared with the increase in new defaults. We clearly
expect that the default rate for the first half of 2020 will increase, since the HY pop-
ulation base will only be impacted by first-half-year fallen angels. For all of 2020,
however, the mid-year population base, which is our annual default rate benchmark,
will be more widely affected. In summary, we are comfortable with our fallen angel
default rate estimated adjustment of C1% over the next twelve months (until May
2021).
During the pandemic, financial markets have witnessed unprecedented short-term
volatility in both equity and debt markets. A related consequence is that default
rate forecasts which partially depend on market conditions will have a good deal of
volatility, even from month to month. To indicate this, our recent next-twelve-month
default rate forecasts based on an average of our three methods plus the adjustment
factor have fluctuated from 4.6% as of the end of 2019 to 9.0% as of March 26, 2020,
and to 8.6% as of May 15, 2020. As of the end of May, the preliminary calculation
of default rates, year to date (YTD), is about 2.33%, based on about US$35 billion
of defaults.
3.1 Forecasting Covid-19 recovery rates

Another forecast of credit conditions relates to the recovery rate, which is based
on the price of the bond issue just after default. Utilizing regression estimates on
the concurrent relationship between default and recovery rates, we can estimate the
recovery rate on defaults implied by our default rate forecasts. Recovery rates are
extremely important for many reasons, including estimates for LGD, now neces-
sary to meet Bank for International Settlements (BIS) bank capital requirements, for
prices of distressed investor strategies, credit default swap (CDS) prices and lender
decisions, among others. Our measure of recovery rates (Altman et al 2005) is based
on the weighted (by amount outstanding) average of the prices on defaulting issue
from just after default. This price reflects approximately what existing creditors of
the debt could sell their holdings for and what distressed investors would have to pay.
As such, it provides a market clearing estimate based on supply and demand condi-
tions at the time of default as well as the present value of expected future values of the
reorganized debt at the end of the restructuring period, usually when emerging from
Chapter 11. Our recovery rate measure and alternatives, primarily those calculated

12 E. I. Altman
FIGURE 4 Recovery rate/default rate association: US dollar weighted average recovery

rates to US dollar weighted average default rates, 1982–2019.
70
2007
2006 2014
1987
60 2011 1993 20122005
1983 2017
1997 2004
2013
1996 2018
Recovery rate (%)
50 1984 1992
2010 1985 2003
2019 1988 2008
40 1995
1994 1982 1989 1991
2009
1998 1986
2015
30 2016
1999
2000 2001 2002
1990
20
10
0 2 4 6 8 10 12 14
Default rate (%)
Linear (blue line): y D 2.7129x C 0.5483, R2 D 0.4803. Log linear (red line): y D 0.117ln.x/ C 0.0196,
R2 D 0.5863. Quadratic (green line): y D 39.526x 2 7.4502x C 0.625. R2 D 0.5748. Exponential (black line):
y D 0.5514 exp. 6.756x/. R2 D 0.5303. Source: Altman et al (2005).
by rating agencies or the thirty-day, post-default auction price used in the CDS mar-
ket, are discussed in Altman et al (2004, 2019). Our default recovery rate estimates
are based on any one of the four regression structures shown in Figure 4: linear,
log linear, quadratic and exponential associations between default and recovery rates
on an annual basis for all corporate bond defaults.1
Updating the model through 2019, Figure 4 (from Altman et al (2005)) shows that
our approximate 9% forecast of the default rate in 2020 implies an average recovery
rate on these defaults of 30%, about 16 percentage points below the weighted aver-
age historical recovery rate (46%). Note that almost all of the regression constructs
forecast a 30% recovery for 2020. If this materializes, the weighted average recovery
rate in 2020 will be lower than the rate in the GFC. Results through April 2020 gave
recovery rates of 30.2%, which is about what we forecast for all of 2020.2
1Color figures are available in the online version of this paper.

2Note that through May 2020 the actual weighted average recovery rate on US$35.7 billion YTD
defaults was about 30.2%, almost exactly as our regression estimates predict.

3.2 Summary of default rate forecasts at different points in time in

2020
As a result of the pandemic, financial markets have witnessed enormous increases in
short-term volatility in both equity and debt markets. A related consequence is that
default rate forecasts that partially depend on market conditions will also have a good
deal of volatility, even from month to month. Our recent next-twelve-month default
rate forecasts, as of several recent dates, based on an average of our three methods
plus the adjustment factor, are as follows.
End of 2019: 4.6%.
March 26, 2020: 9.0%.
May 19, 2020: 8.6%.

Note that Table 2 shows the next twelve months’ (or end of 2020) default rate
estimates on HY bonds by various rating agencies and investment banks.
Our estimate (8.6%) for the next-twelve-month (April 2021) default rate is lower
than most of the other forecasters, with the exception of Fitch Ratings (2020).
Because of the uncertainty surrounding Covid-19 we do not forecast two-year default
rates, but others do (see the summary in Table 2).
3.3 Bankruptcies in the Covid-19 crisis

As noted above, our default statistics include
(1) missed interest payments not cured within the grace period,
(2) out-of-court distressed exchanges in which bondholders receive less than par
value in the exchange, and
(3) the most dramatic indication of insolvency: bankruptcy filings.

It is relevant now to discuss all bankruptcy filings, even those that do not involve
HY bond-issuing companies. Indeed, one of the current questions in the Covid-19
crisis is whether the number and size of corporate bankruptcies will overwhelm the
bankruptcy court system and challenge the ability of firms to successfully restructure
in Chapter 11 reorganization; see Ellias et al (2020) for a proposal to support the
bankruptcy system during the crisis period.
For more than thirty years, the NYU Salomon Center has been compiling and
monitoring Chapter 11 filings for relatively large firms with liabilities at the time
of filing above US$100 million and so-called mega-bankruptcies above US$1 bil-
lion (see Altman and Kuehne 2020). By far, most of the bankruptcy system’s cred-
itor exposures are in these large firm categories. This is not to mean that small and

14 E. I. Altman
TABLE 2 Default rate estimates on HY bonds (%).
Cumulative
through
2020 2021
Barclay’s 9–10 20
Bank of America 9.6 22
Deutsche Bank 9.5 —
S&P Global 10.0–13 20
Moody’s 6.8–16.1 —
Fitch 5–7 13–17

Twelve months. Twelve months, base and severe recession estimates.
medium firm bankruptcies are not important for the impact of Covid-19, but these
are not the focus of our analysis.
Even more than bond and loan defaults, the number of corporate bankruptcies
so far in 2020 has increased dramatically. As of May 19, 2020, there were sixty-
six Chapter 11 filings with liabilities greater than US$100 million, twenty-three
of which had liabilities over US$1 billion (mega-bankruptcies; data sourced from
BankruptcyData.com (2020)). Since we do not know of any statistical forecasting
models specifically developed for Chapter 11 filings, we simply extrapolate current
totals. As the statistics in Table 3 show, if we extrapolate these numbers for the rest
of 2020, the number of over US$100 million Chapter 11 liability firms will be second
only to 2009 and the mega-bankruptcies will easily break the all-time annual record,
again recorded in 2009. The extrapolated 2020 total exceeds the record, even when
adjusting the 2019 liabilities for inflation. Further, there is reason to believe that the
numbers for 2020 will be even greater than the extrapolated amounts, since the YTD
totals as of May 19, 2020 include over two months when the credit cycle was still
benign (Table 3).
In summary, we firmly expect that the US bankruptcy reorganization system
will be severely challenged in 2020, and could perhaps benefit from additional
Congressional support.
4 THE BBB DEBT MARKET AND FUTURE CREDIT CONDITIONS

Even before the Covid-19 crisis realization in the United States, much discussion
in the financial press was focused on the huge increase in the amount of bonds and
loans outstanding that received a BBB rating from the credit rating agencies (CRAs)
(see, for example, Celik et al 2020). Indeed, Figure 5 shows that the amount of BBB
bonds issued exploded to more than US$2.5 trillion as of December 2019, amounting

TABLE 3 Large firm bankruptcies in the United States as of May 19, 2020, and
extrapolations for 2020.
(a) Chapter 11 filings of more than US$100 million
Number of Chapter 11 filings 66

Extrapolated number for 2020 173
Extrapolated all time ranking 2nd
All time highest year (2009) 232
Historic annual average (1989–2019) 78
Historic annual median (1989–2019) 66
(b) Chapter 11 filings of more than US$1 billion
Number of Chapter 11 filings 23

Extrapolated number for 2020 61
Extrapolated all time ranking 1st
Next highest year (2009) 49
Historic annual average year (1989–2019) 17
Historic annual median year (1989–2019) 20
to about 52% of all investment-grade debt. For comparison, in the 2007 credit bub-
ble year there was only about US$700 billion BBB-rated debt outstanding (36% of
investment-grade debt in 2007), representing about 28% of the BBB debt outstanding
in 2019. The enormous growth in corporate debt in general, and BBB-rated debt in
particular, has been carefully documented and commented upon in an Organization
for Economic Cooperation and Development study (Celik et al 2020). This study
emphasized the huge importance of BBB debt in the current market and the role of
CRAs in allowing companies to increase their leverage ratios and still maintain their
investment-grade BBB rating. This growth raises the question of inflation of ratings,
which we will return to shortly.
As we are now in a new crisis period, we address the important question of the
percentage and dollar amount of the enormous total of BBB-rated bonds likely to be
downgraded to noninvestment-grade, HY bonds, thereby becoming so-called fallen
angels. Not only is this amount important for default rate estimates, as we have dis-
cussed, but it turned out that all fallen angels after March 23 were eligible for sub-
sequent purchase and for liquidity support by the Fed for the debt markets. When
asked, even before the Covid-19 pandemic threat was realized by the markets, what
amount of downgrades to below BBB could be expected if the credit markets were
to change to a distressed condition, all of the major CRAs and most analysts opined
that the impact of these fallen angels on the HY market would be minimal, and that

16 E. I. Altman
FIGURE 5 US BBB-rated corporate bonds outstanding, 2005–18.
BBB-rated bonds outstanding as a percentage

3.0 52
of US investment-grade corporate bonds

BBB-rated bonds outstanding (US$tn)
Bonds outstanding
50
Bonds outstanding as a percentage
2.5 of US investment-grade corporate bonds 48
46
2.0
44
42
1.5
40
38
1.0
36
0.5 34
32
0 30
2005
2006
2007
2008
2009
2011
2012
2013
2014
2015
2016
2017
2018
2010
Source: Bloomberg Barclays US Corporate Investment Grade Index.
the amount would probably not top 10%. This estimate implies about US$250 bil-
lion added to the US$1.5 trillion HY bond market over a two- to three-year downturn
period. Hence, the inference of a “crowding-out” of marginal, low-quality HY bond
issuers by the new fallen angel BBB firms would not be material. This also implies
that these marginal firms – including “zombie” companies kept alive somewhat arti-
ficially due to forbearance by banks and record low interest rates on new loans with
little or no protective covenants and a buoyant new issue market – would default in
relatively small numbers. The main basis for the CRA assertions was the observation
of what happened to BBBs in the GFC of 2008–9 and other past downturns.
Our assessment of the issue is different. We are concerned about a much larger
deterioration of ratings by the rating agencies and/or in the market’s perception of
BBB firms. It is likely, in our opinion, that a type of credit rationing will take place in
a post-2019 downturn when liquidity dries up and any equilibrium interest rates for
the lowest-quality credits, mostly CCCs, will not be observed. Hence, we posit that
the crowding-out effect will take place in the event of a massive downgrade market.
To be fair, we did not observe both a huge BBB downgrade and credit rationing in
past financial crises, since the BBB market was then much smaller. For example,
in 2007 the BBB market was only US$700 billion, about one-quarter of the current
BBB market’s size (Figure 5).
We therefore applied our Z-score models to manufacturing and nonmanufacturing
industrial firms as of the end of 2019, before the realization of Covid-19. Based on

TABLE 4 Downgrade vulnerability of BBB-rated bonds based on Z-score as of 2019.
Bond rating equivalents of BBB-rated bonds

‚ …„ ƒ
00
Z -score determined Z -score determined
From BBB to BB 57/298 (19%) 78/371 (21%)

From BBB to B 45/298 (15%) 56/371 (15%)
Total 102/298 (34%) 134/371 (36%)
Source: author’s computations from Capital IQ data.
a sample of 298 BBBC, BBB and BBB firms for which stock price data, balance-
sheet and profit and loss statements were available, and 372 firms with either market
or book equity data available, we examined the bond rating equivalents (BREs) of
their scores.3 Our sample includes essentially 100% of the approximately 384 non-
financial issuers of the almost US$3 trillion BBB market in 2019. The results of our
analysis, presented in Table 4, show that 34% of our sample of BBB firms with BREs
based on Z-scores and 36% based on Z 00 -scores were classified as noninvestment-
grade, BB- or B-rated companies as of December 2019. Our analysis also shows that
about 15% of the sample had BREs above BBB, so the percentage of BREs below
investment grade was more than double that above. For the balance, about half of
the sample, we agreed with the rating agency’s BBB rating. So, the BREs are not
symmetrical below and above the BBB rating.
If, as our analysis implies, the percentage that will be downgraded to junk levels
in 2020 and 2021 is 20–25% (35% is not likely over a two-year downturn), then this
would result in about US$500–625 billion of new fallen angels or about a 33–42%
increase in the US$1.5 trillion HY “junk” market: not a trivial amount. Indeed, Light
(2020) estimated that as much as US$1 trillion of BBB bonds would be downgraded
to noninvestment grade. We have some doubts that CRAs will actually downgrade
this large an amount (20–25%) since their estimate of a maximum of 10% down-
grades could be something of a self-fulfilling prophecy. Still, based on our analysis
of data at the end of 2019, we project a sizable amount (more than 10%) would be
downgraded to noninvestment grade in the next downturn, whatever the catalyst. To
be fair to the CRAs, in the early stage of the Covid-19 crisis, at least one rating
agency recognized the already large amount of fallen angel downgrades and the even
3For a discussion of Z and Z 00 models and our experience with these models over the last fifty
years, see Altman (2018) and Altman et al (2019). BREs are determined by calibrating Z- and
Z 00 -scores to median values of each of the Standard & Poor’s (S&P) rating categories for various
years over the last fifty or more years. For example, a Z-score below 1.81 in 1968 was classified
as a likely bankrupt (Altman 1968), whereas the cutoff level drops to zero (0.0) in recent years.

18 E. I. Altman
larger numbers of additional vulnerable BBBs (see Kesh et al 2020). Although just
about all corporate bonds suffered significant price declines in March, the subsequent
support shown by the Fed for corporate bonds, including, notably, fallen angels, con-
tributed to price rebounds for most bonds and the vast amounts of new issue liquidity
since.
In the first two months after the Covid-19 crisis realization, estimates of reported
US downgrades of BBB bonds to “junk” ranged from US$135 billion (Bank of
America 2020) to US$158 billion (Economist 2020; CreditSights 2020), and others
estimate that another US$280 billion (Bank of America 2020) will be downgraded
by May 2021, bringing the fifteen-month total to US$430 billion. Recall that we
believe that a two-year downgrade total in the current downturn credit cycle could
reach US$500–625 billion. We also observed that every one of the downgraded fallen
angel bonds from March to mid-May 2020 had a BRE below investment grade, based
on Z-scores, and all but three of the eighteen companies based on the Z 00 -score
model (Table 5); the latter three recorded a Z 00 -score BRE of BBB, the same as their
actual CRA rating in 2019. Therefore, our analysis implies that the expected default
amounts, and possibly the short-term default rate, will be increased by the recent
explosion of BBB-rated debt and subsequent downgrades in a major and sustained
downturn.
4.1 Ratings inflation or persistent rating overvaluation?

Since the GFC, credit rating agencies have been criticized by regulators, lawmakers
and market practitioners for being too lenient in their assessment of credit instru-
ments. In particular, mortgage-backed securities received sudden and massive down-
grades from high investment grades to CCC and sometimes to default during 2008
and 2009. We have heard some of the same criticisms of late with respect to corpo-
rate debt ratings (see, for example, Economist 2020). In order to assess this critique,
we extended our analysis of the aforementioned BBB investment grade segment of
the corporate bond market to various times over the last dozen years. This category
is particularly relevant due to its enormous growth and its position on the borderline
between investment and below investment grade. Thus, this rating category repre-
sents a particularly good opportunity to assess possible rating inflation in the bond
market in recent periods.
Herpfer and Maturama (2020) investigate the rating inflation issue by analyz-
ing US$900 billion of “performance sensitive loans,” whereby the interest paid is
a dynamic function of the CRA rating. They conclude that, despite the lawsuit set-
tlements with the government over conflicts of interest during the GFC in the securi-
tized mortgage-backed security market, rating inflation of those interest rate sensitive
loans had been prevalent and remained unchanged after the lawsuit settlements. They

TABLE 5 Fallen angel Z- and Z 00 -scores and their bond rating equivalents, May 2020.
General information Financial information

‚ …„ ƒ ‚ …„ ƒ
Face
downgrade Date of Z -score Z 00 -score
Issuer name FI ticker Industry (US$m) data Z -score BRE Z 00 -score BRE
Ford Motor F Autos 34 572 12/31/2019 0.91 CC+ 4.13 B
Occidental Petroleum OXY Energy 29 059 12/31/2019 0.80 CC 4.71 B+
Western Midstream WES Energy 7 820 12/31/2019 0.77 CC 3.95 B
Partners
Continental CLR Energy 5 300 12/31/2019 1.54 B 5.73 BBB
Resources
Cenovus Energy CVECN Energy 4 781 12/31/2019 1.39 CCC+ 5.18 BB
Delta Air Lines DAL Transportation 4 100 12/31/2019 1.30 CCC+ 3.04 CCC+
Macy’s M Retail 2 456 11/02/2019 2.05 B+ 5.63 BB+
ZF NA Capital ZFFNGR Autos 1 699 12/31/2019 — — 5.15 BB
Methanex MXCN Chemicals 1 550 12/31/2019 1.28 CCC+ 5.30 BB
Adani Abbot Point ADAABB Transportation 500 03/31/2019 — — 3.87 B
Terminal
Marks & Spencer MARSPE Retail 300 09/28/2019 2.36 BB 5.76 BBB
Pemex PEMEX Energy 58 621 12/31/2019 — — 2.93 D
Rockies Express ROCKIE Energy 2 050 12/31/2019 — — 5.37 BB+
Pipeline
Royal Caribbean RCL Leisure 1 450 12/31/2019 1.81 B+ 4.25 B
Cruises
Trinidad Generation TRNGEN Utility 600 12/31/2019 — — 5.68 BBB
Growthpoint Properties GRTSJ Real estate 425 12/31/2019 0.81 CCC 5.02 BB
Hillenbrand HI Capital goods 375 12/31/2019 1.37 B 4.94 BB
Source: author calculations from NYU Salomon Center and Capital IQ data.

Covid-19 and the credit cycle
19
20 E. I. Altman
also concluded that CRAs are reluctant to downgrade the issuers when the costs of
the downgrade are high (in other words, there is continued “stickiness” in downgrad-
ing clients), and that CRAs are slow to downgrade, or avoid downgrades entirely.
Our own results from prior periods (Altman and Rijken 2004) demonstrated that
CRA downgrades lag point-in-time models, such as Z-score-type models, by more
than one year, indicating this same stickiness. Further, Posch (2011) investigated the
timeliness of rating changes in the wake of the GFC and found that several factors
result in stickiness of rating actions and that default probability estimates have to
change by at least two notches before a change, up or down, takes place.
Our past experience and observations before Covid-19 were that the aforemen-
tioned stickiness would continue, especially with respect to fallen angel downgrades.
The current cycle’s early results, however, may prove us wrong, as fallen angels
have been numerous and, as shown above, at least one rating agency (S&P Global)
recognized that lots more are vulnerable. Curiously, Herpfer and Maturama (2020)
conclude there was no evidence from their sample that this stickiness at the bor-
der between investment grade and HY took place. Bruno et al (2016) assessed
results comparing the symmetry between upgrades and downgrades crossing the
investment-grade threshold in a traditional issuer-paid CRA model firm and an
investor-paid CRA model firm. They showed that the certification of the investor-
paid CRA firm affects the issuer-paid CRAs, which become more symmetrical in
their subsequent up- and downgrade rating changes.
The Z-score models’ results, presented in Table 4, suggest overevaluation of about
one-third of BBB-rated companies. However, these results are not necessarily indica-
tive of rating inflation, since this represents a single reference-date data point. To pro-
vide some evidence of a time series inflationary trend, we compared the BBB class
of 2019 with those of 2007, 2013 and 2016. We ran Z- and Z 00 -score calculations
on 108 BBB firms in 2007, 332 in 2013 and 416 in 2016 and compared them with
our 2019 sample, discussed earlier, to observe if this overvaluation was a very recent
occurrence or one that has persisted for some time. The results of these earlier test
years were essentially the same, or even greater than in 2019; ie, between 35% and
45% of the firms received a lower-than-BBB BRE during these periods, depending
upon the Z-score test and the period analyzed, and less than 15% received higher
than BBB BREs. Our BREs are calibrated based on median Z-scores for each CRA
bond rating class over the last thirty years, thereby recognizing changes in median
scores over time. We conclude, therefore, that CRAs have been overvaluing a sig-
nificant number of BBB corporates over the last dozen years, as well as in 2019. Of
course, rating agency ratings can differ from point-in-time Z-score BREs, since dif-
ferent criteria are used in the two methods. However, these different methods should
not manifest continuous bias of higher versus lower ratings.

On the other hand, there is no evidence of overall rating inflation. What there
appears to be is persistent rating overvaluation of investment-grade BBB-rated debt.
Clearly, the deterioration in creditworthiness that will cause an investment-grade rat-
ing to be downgraded to “junk” is a more dramatic one than just about any other rat-
ing migration rating change. The arbitrary and, in some cases, regulatory distinction
between investment grade and HY for fixed-income securities is steeped in tradi-
tion. For example, the insurance industry has required specific reserve requirements
for noninvestment-grade debt in investment-grade portfolios for at least twenty-five
years, and the US Securities and Exchange Commission regulates how much spec-
ulative grade debt an investment-grade corporate bond mutual fund can own. For
some investors (such as exchange-traded funds and some mutual funds), sale of the
security upon observing a fallen angel migration is required, potentially causing a
free fall in their bond and loan prices. During the stress conditions of the Covid-
19 environment, investors who have the flexibility to hold these newly downgraded
HY securities may choose to do so if they feel the firms (eg, airlines and energy
firms) will likely return to BBB in more normal times. Of course, the downgrade
may be a mere first move toward eventual insolvency and default. Altman and Kao
(1992) suggest that a downgrade is more likely to be followed by another downgrade
(ie, positive autocorrelation) rather than returning to the higher rating. This propen-
sity to observe a downgrade, rather than an upgrade, in subsequent rating changes
varied from two to five times greater, depending upon the industrial sector of the
firm. Perhaps it is time to update these propensities?
4.2 Collateralized loan obligations

Rating downgrades that always accompany stressed credit cycles could be partic-
ularly critical in the Covid-19 economic and credit market crisis period. As noted
earlier, CLOs have grown enormously in the United States and now amount to over
US$650 billion (see Figure 6) compared with just US$300 billion in 2007 and about
US$400 billion in 2014. Increasingly, the leveraged loans that make up the bulk of
CLOs are without protective covenants for investors, and are known as “cov-lite”
loans (see Griffin et al (2019) and Berlin et al (2020) for analyses on this increasing
trend). Griffin et al (2019) shows that loan covenant violations have dropped by 70%
in their sample period, due mostly to less restrictive covenants in loan contracts. This
should not only reduce the incidence of loan defaults in the short run, but also reduce
recovery rates on future defaults of these same companies. Exceptions to these cov-
lite loans are when the same firm draws down on existing revolving credit agree-
ments, which almost always retain traditional financial constraints. These derivative
CLO instruments are held widely by institutional and mutual fund investment com-
panies, exchange-traded funds and others. CLOs are very sensitive to downgrades,

22 E. I. Altman
FIGURE 6 US CLOs outstanding, 2006 Q1–2019 Q2.
700 US$640bn
600
US dollars (billions)
500
400
300
200
100
0
2006 Q1
2006 Q3
2007 Q1
2007 Q3
2008 Q1
2008 Q3
2009 Q1
2009 Q3
2010 Q1
2010 Q3
2011 Q1
2011 Q3
2012 Q1
2012 Q3
2013 Q1
2013 Q3
2014 Q1
2014 Q3
2015 Q1
2015 Q3
2016 Q1
2016 Q3
2017 Q1
2017 Q3
2018 Q1
2018 Q3
2019 Q1
Sources: 2006–18 Securities Industry and Financial Markets Association (SIFMA) data from Liu and Schmidt-
Eisenlohr (2019, Figure 1); 2019 Q1 from SIFMA data; and 2019 Q2 from Bank of America.
especially to CCC and to defaults of the portfolio-collateral companies. Indeed, it is

common for overcollateralization triggers to be invoked if CCC leveraged loans in
CLOs increase to a certain percentage, eg, 7.5%. Perhaps equally as important, CLOs
are constrained to purchase new issues once the 7.5% threshold has been violated. In
addition, the CLO manager is required to take a “haircut” on the CCC debt above the
7.5% threshold, which itself causes problems in the compliance of the overcollater-
alization test. Indeed, any cash and new interest coming into the CLO must be used
first to pay off the most senior tranche of the CLO until the overcollateralization ratio
comes back into compliance. The triggering of these constraints therefore restricts
the CLO’s ability to invest in new loans. The increased risk profile of the CLO will be
more severe than originally proposed to the market, resulting in likely price declines
and potential redemptions of the senior tranches and losses to the junior and equity
tranches, if not all tranches.
As of May 7, 2020, roughly 15% of US CLOs (182 in all) were failing the over-
collateralization test, but only 1% were failing the senior tranche test (Preston et al
2020). This compares favorably to the peak of the GFC, when 56% failed the over-
collateralization test and 11% the senior test. Almost all of these 2020 “failures”
occurred after the realization of the Covid-19 crisis. Over half of the new overcol-
lateralization test failures were already past their reinvestment period of new loans

in their pools and about 20% had reinvestment periods ending after 2022. The latter,
therefore, are the most constrained, going forward.
4.3 “Zombie” firms

The concept of “zombie” firms is well established (see Acharya et al 2019a,b; Cohen
et al 2017; Banerjee and Hofmann 2018). Although the definition of these “walking
dead” firms is itself a controversial issue, all attempts connote firms that are kept
alive somewhat artificially or dependent on a specific credit environment. For exam-
ple, firms may be granted loans by banks, which themselves may be having capital
problems, for the sole purpose of not wanting to write off marginal customers. Or,
firms who would have serious liquidity and interest rate constraints to raise capital
in a normal credit environment may be able do so in a low interest rate, cov-lite,
“risk-on” environment, such as we found ourselves in at year-end 2019.
The BIS studies by Cohen et al (2017) and Banerjee and Hofmann (2018) define
“zombies” as firms whose interest coverage ratios are less than 1.0. To evaluate this
definition, we may observe that the median CCC-rated firm in the United States
and Europe has an interest coverage ratio less than 1.0.4 As of 2020, this would
include almost US$100 billion of CCC bonds outstanding today and perhaps a simi-
lar amount of leveraged loan companies, especially since many CLO portfolio firms
have been downgraded in the Covid-19 crisis environment. Banerjee and Hofmann
(2018) estimated that as much as 16% of all US listed corporations had this financial
profile in 2017 compared with only 2% in the 1990s. This percentage will certainly
increase in 2020. Acharya et al (2019a) follow Caballero et al (2008) and Gian-
netti and Simonov (2013) in including firms receiving loans at below-market interest
rates, ie, if the firm’s interest rate expense is below that paid by the most creditwor-
thy firms in the economy. Acharya et al (2019a,b) analyze zombie European firms in
non-GIIPS countries.5 Many of these companies had received loans from banks that
had a stake in the company from prior loans. Acharya et al estimated that roughly
8% of loans in their sample were zombie loans.
We propose a different method for identifying zombies in the United States by
selecting those firms with existing low credit ratings that we estimate have very
high probabilities of default within two years. Starting out with a relatively large
sample of firms whose senior unsecured bond rating from S&P Global was B
or below at year-end 2019, we ran Z- and Z 00 -score tests to assess which had a
BRE of D D default classification, ie, scores below zero (0.0).6 We found data on
ninety-nine firms rated B or below as of year-end 2019 (sixty-seven were B and
4 See, for example, https://www.spglobal.com.

5 GIIPS denotes Greece, Italy, Ireland, Portugal and Spain.
6 See Altman et al (2019) for more details on the mapping of bond rating equivalents.

24
TABLE 6 Financial profile of low-rated bonds and their Z-score and Z 00 -score default predictions, December 2019.
E. I. Altman
(a) Z-score default prediction
Rating EBIT/Interest BRE of D EBITDA/Interest BRE of D

S&P Sample with BRE ‚ …„ ƒ ‚ …„ ƒ
rating size Z -score of D (%) <1.0 1.0–1.5 1.5–2.0 >2.0 <1.0 1.0–1.5 1.5–2.0 >2.0
B 59 12 20.3 10 — 1 1 4 2 1 5
CCCC 21 7 33.3 6 — 1 — 3 1 1 2
CCC 9 3 33.3 2 1 — — 2 1 — —
CCC 8 4 50.0 3 — 1 — 2 1 — 1
Total 97 26 26.8 21 1 3 1 11 5 2 8
(b) Z 00 -score default prediction
Rating EBIT/Interest BRE of D EBITDA/Interest BRE of D

S&P Sample with BRE ‚ …„ ƒ ‚ …„ ƒ
rating size Z -score of D (%) <1.0 1.0–1.5 1.5–2.0 >2.0 <1.0 1.0–1.5 1.5–2.0 >2.0
B 67 8 11.9 6 — — 2 3 2 1 2
CCCC 23 3 13.0 3 — — — 2 — 1 —
CCC 9 1 11.1 1 — — — 1 — — —
CCC 9 4 44.4 3 — 1 — 3 — — 1
Total 108 16 14.8 13 — 1 2 9 2 2 3
EBIT, earnings before interest and taxes. EBITDA, earnings before interest, tax, depreciation and amortization. Sources: author data, NYU Salomon Center.
forty-two CCCs) to which we could apply the Z-score model and 108 firms that had
sufficient data to apply the Z 00 -score model. Of these, 26.8% had an implied BRE
of D using the Z-score model and 15.8% had an implied BRE of D using the Z 00 -
score approach (see Table 6). All of the Z-score D firms also had D-scores with the
Z 00 test. Incidentally, twenty-one of the twenty-six B , CCC and CC firms that we
predicted to default had interest coverage ratios below 1.0 (the BIS test) and eleven
had earnings before interest, tax, depreciation and amortization (a cashflow proxy)
coverage ratios below 1.0, a more stringent test. Our sample of about 100 firms is
around 40% of the entire population of HY low-quality firms in the Bank of Amer-
ica ICE High Yield index, since many of the remaining firms are owned by private
equity firms and do not provide financial data publicly.
Another interpretation of our tests is that we are measuring default risk, not nec-
essarily zombie firms. Since the market was not expecting default in most of these
companies, ie, their bond prices were not below seventy in December 2019, we feel
it is legitimate to consider them as zombies. We posit that many of these firms, espe-
cially those who will not be able to meet their interest or maturity payments during
the Covid-19 pandemic, will now default, due partially to the crowding-out effect
from new-entry fallen angels, discussed above, as well as rising interest rates on new
financing. Of course, the exact number of these firms is difficult to estimate with
precision, but we feel it will be nontrivial. As estimated earlier, about US$15 billion,
or 1% of the HY bond market, and perhaps a similar amount of leveraged loans, will
fall into this crowded-out default category.
5 CONCLUSION
No doubt the situation we find ourselves in today is unprecedented in terms of the
speed of recent asset price declines, subsequent rapid rebound after government
credit and job market supports, and the expected impact and forecast of corporate
defaults. These rapid fire dynamics make the future extremely difficult to predict. As
such, we may not have reliable models to capture these dynamics based on histori-
cal, modern credit market experience. Despite these uncertainties, our best estimate
at this point in time is a default rate on the HY bond and leveraged loan markets
for the next twelve months of between 8.6% and 9.0%, or about US$130–135 bil-
lion in each of these leveraged finance markets. With respect to Chapter 11 corpo-
rate bankruptcy reorganization filings, in 2020 we expect a record annual number of
mega-bankruptcies over US$1 billion in liabilities, which could, along with smaller
firm filings, challenge the bankruptcy court system. Our forecast for filings with more
than US$100 million in liabilities is second only to 2009. These large bankruptcies
do not capture the impact of Covid-19 on small and medium-sized enterprises, a
topic which is beyond the scope of this paper.

26 E. I. Altman
A related issue is the enormous growth and downgrade potential of the BBB-rated
investment-grade corporate bond category. This group could produce as much as
US$500–625 billion of new HY bonds, ie, “fallen angels,” over the next two years.
Thus, the likelihood that many of these low-quality credits will be able to fund them-
selves during the credit downturn is low, thereby increasing defaults. We call this the
“crowding-out effect”. Additional evidence is provided to assess the criticism that
we are observing ratings inflation, especially in the massive BBB-rating sector. Our
analysis suggests no evidence of ratings inflation, but we do find evidence of persis-
tent rating overvaluation in the BBB-rating sector over the last dozen years. Finally,
we present some evidence of the existence of “zombie companies,” and we posit that
many of these firms will be among those that will default during the Covid-19 crisis
period. We hope that our analysis has provided some guidance on these important
issues.
DECLARATION OF INTEREST
The author reports no conflicts of interest. The author alone is responsible for the
content and writing of the paper.
ACKNOWLEDGEMENTS
The author thanks Ryan Fitzgerald and Siddhartha Kaparthy of the NYU Salomon
Center for the data assistance; Viral Acharya, NYU Stern faculty, for his insightful
comments; the HY research staff at Bank of America, especially Oleg Melentyev
and Eric Yu, together with Eric Rosenthal (Fitch), Martin Fridson (Lehmann Livian
Fridson Advisors LLC), Robert Benhenni (Classis Capital, SpA), Edith Hotchkiss
(Boston College), David Smith (Virgina), Katherine Waldock (Georgetown), Wei
Wang (Queens University) and Cristiano Zazzara (NYU Stern VRI), for their com-
ments and data assistance; and Linda Allen and Michael Gordy, editors-in-chief of
the Journal of Credit Risk, for their considerable thought-provoking and motivating
comments.
REFERENCES
Acharya, V., Eisert, T., Eufinger, C., and Hirsch, C. (2019a). Whatever it takes: the real
effects of unconventional monetary policy. Review of Financial Studies 32(9), 3366–
3411 (https://doi.org/10.1093/rfs/hhz005).
Acharya, V., Crosignani, M., Eisert, T., and Eufinger, C. (2019b). Zombie credit and (dis-)
inflation: evidence from Europe. Working Paper, October NYU Stern Business School
(https://doi.org/10.3386/w27158).

Altman, E. (1968). Financial ratios, discriminant analysis, and the prediction of corporate
bankruptcy. Journal of Finance 23(4), 589–609 (https://doi.org/10.1111/j.1540-6261
.1968.tb00843.x).
Altman, E. (1989). Measuring corporate bond mortality and performance. Journal of
Finance 39(4), 909–922 (https://doi.org/10.1111/j.1540-6261.1989.tb02630.x).
Altman, E. (1990). Investing in distress securities. Report, Foothill Group, Los Angeles,
CA.
Altman, E. (2018). A fifty-year retrospective on credit risk models, the Altman Z-score
family of models and their applications in financial markets and managerial strategies.
Journal of Credit Risk 14(4), 1–34 (https://doi.org/10.21314/JCR.2018.243).
Altman, E., and Kao, D. L. (1992). The implications of corporate bond rating drift. Financial
Analysts Journal 48(3), 64–75 (https://doi.org/10.2469/faj.v48.n3.64).
Altman, E., and Kuehne, B. (2020). Defaults and returns in the high-yield bond and dis-
tress debt market. NYU Salomon Center Special Report, February 25, The Salomon
Center.
Altman, E., and Rijken, H. (2004). How rating agencies achieve rating stability. Journal of
Banking and Finance 28, 2679–2714 (https://doi.org/10.1016/j.jbankfin.2004.06.006).
Altman, E., Resti, A., and Sironi, A. (2004). Default recovery rates in credit risk modeling:
a review of the literature and recent evidence. Economic Notes 33, 183–208 (https://
doi.org/10.1111/j.0391-5026.2004.00129.x).
Altman, E., Brady, B., Resti, A., and Sironi, A. (2005). The link between default and
recovery rates: theory, empirical results and implications. Journal of Business 78(6),
2203–2228 (https://doi.org/10.1086/497044).
Altman, E., Gande, A., and Saunders, A. (2010). Bank debt versus bond debt: evidence
from secondary market prices. Journal of Money, Credit and Banking 42(4), 755–767
(https://doi.org/10.1111/j.1538-4616.2010.00306.x).
Altman, E., Hotchkiss, E., and Wang, W. (2019). Corporate Financial Distress, Restruc-
turing and Bankruptcy, 4th edn. Wiley (https://doi.org/10.1002/9781119541929).
Banerjee, R., and Hofmann, B. (2018). The rise of zombie firms: causes and con-
sequences. BIS Quarterly Review, September, 67–78. URL: https://www.bis.org/publ/
qtrpdf/r qt1809g.pdf.
Bank of America (2020). Do fallen angels make HY higher quality? BofA Global Research,
May 8.
Bank of America Merrill Lynch (2019). High yield strategy: where we are in this credit
cycle. Report, September 13, Bank of America Merrill Lynch.
BankruptcyData.com (2020). Business bankruptcy filings year-end report 2019. New
Generation Research, Inc., Boston, MA.
Berlin, M., Nini, G., and Yu, E. G. (2020). Concentration of control rights in leverage loan
syndicates. Journal of Financial Economics 137(1), 249–271 (https://doi.org/10.1016/
j.jfineco.2020.02.002).
Board of Governors of the Federal Reserve System (2019). December 11, 2019: FOMC
projections materials, accessible version. Report, December 11, Federal Open Market
Committee. URL: https://bit.ly/3hhaE24.
Bruno, V., Cornaggia, J., and Cornaggia, K. (2016). Does regulatory certification affect the
information content of credit ratings? Management Science 62(6), 1578–1597 (https://
doi.org/10.1287/mnsc.2015.2188).

28 E. I. Altman
Caballero, R., Hoshi, T., and Kashyar, A. K. (2008). Zombie lending and depressed
restructuring in Japan. American Economic Review 98, 1943–1977 (https://doi.org/10
.1257/aer.98.5.1943).
Celik, S., Demirtas, G., and Isaksson, M. (2020). Corporate Bond Market Trends, Emerg-
ing Risks and Monetary Policy. OECD Capital Markets Series, Paris.
Cohen, B., et al (2017). International and financial market developments. BIS Quarterly
Review, September. URL: http://www.bis.org/publ/qtrpdf/r qt1709.htm.
CreditSights (2020). US IG: new index weights after supply, downgrades. Report, May 6,
CreditSights.
Economist (2020). Credit-rating agencies are back under the spotlight. The Economist,
Finance and Economics blog, May 9.
Ellias, J., et al (2020). Proposal to the US Congress for support for the bankruptcy court
system. Report, May 6, Large Corporations Committee of the Bankruptcy and Covid-19
Working Group.
Fitch Ratings (2020). Fitch US leveraged loan default insight. Special Report, May 22.
URL: https://bit.ly/3cTB9r4.
Giannetti, M., and Simonov, A. (2013). On the real effect of bank bailouts: micro evidence
from Japan. Macroeconomics 5(1), 35–67 (https://doi.org/10.1257/mac.5.1.135).
Griffin, T., Nini, G., and Smith, D. C. (2019). Losing control? The 20-year decline in loan
covenant restrictions. Working Paper, Social Science Research Network (https://doi.org/
10.2139/ssrn.3277570).
Herpfer, C., and Maturama, G. (2020). When credit rating agencies avoid downgrading:
the effects of performance sensitive debt. Working Paper, Emory University (https://
doi.org/10.2139/ssrn.3579030).
JP Morgan (2020). Global fixed income views 2Q 2020. JP Morgan Asset Management,
March 18, JPMorgan Chase.
Kesh, S. K., Conti, V. R., and Limbach, S. (2020). Credit trends: potential fallen angels hit
a record-high 111. S&P Global Market Intelligence, May 14, S&P Global Ratings. URL:
https://bit.ly/3hg68kE.
LCD News (2020). Deep dive: US leveraged loan market flashes yellow on recovery. S&P
Global Market Intelligence, May 12, S&P Global Ratings. URL: https://bit.ly/3cOJX1v.
Light, L. (2020). Despite FED backstop, junk-bonds still face big trouble. CIO Magazine,
April 28.
Liu, E., and Schmidt-Eisenlohr, T. (2019). Who owns US CLO securities? FEDS Notes,
July 19, Board of Governors of the Federal Reserve System, Washington, DC (https://
doi.org/10.17016/2380-7172.2423).
Posch, P. (2011). Time to change: rating changes and policy implications. Journal of Eco-
nomic Behavior and Organization 80, 641–656 (https://doi.org/10.1016/j.jebo.2011.06
.026).
Preston, D., Bilskie, R., and Eddins, A. (2020). CLO: this week down in the OC. Structured
Products Research, May 7, Wells Fargo Bank.

Altman (2020)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Altman (2020)

Uploaded by

Copyright:

Available Formats

Journal of Credit Risk 16(2), 1–28

Covid-19 and the credit cycle

Print ISSN 1744-6619 j Online ISSN 1755-9723

(1) low and below-average default rates and forecasted rates;

(2) high and above-average recovery rates on actual defaults;

Journal of Credit Risk www.risk.net/journals

www.risk.net/journals Journal of Credit Risk

Total corporate debt*/total equity market cap Total corporate debt*/GDP

2 CREDIT AND ECONOMIC CONDITIONS DURING THE GLOBAL

Journal of Credit Risk www.risk.net/journals

FIGURE 2 US nonfinancial corporate debt (credit market instruments) to GDP: com-

%NFCD to GDP Four-quarter moving average default rate 16

average default rate (%)

www.risk.net/journals Journal of Credit Risk

Journal of Credit Risk www.risk.net/journals

3 FORECASTING COVID-19 DEFAULT AND RECOVERY RATES

www.risk.net/journals Journal of Credit Risk

Journal of Credit Risk

default rate.t C 1/ D 3:15 C 1:28.yield spread.t //; adjusted R2 D 59:6%:

www.risk.net/journals Journal of Credit Risk

default rate.t C 1/ D 0:923 C 0:240.distress ratio.t //; adjusted R2 D 75%:

R2 was not increased by using nonlinear estimates.

mortality rate D 5:75%;

Journal of Credit Risk www.risk.net/journals

3.1 Forecasting Covid-19 recovery rates

www.risk.net/journals Journal of Credit Risk

FIGURE 4 Recovery rate/default rate association: US dollar weighted average recovery

1Color figures are available in the online version of this paper.

Journal of Credit Risk www.risk.net/journals

3.2 Summary of default rate forecasts at different points in time in

 March 26, 2020: 9.0%.

 May 19, 2020: 8.6%.

3.3 Bankruptcies in the Covid-19 crisis

(3) the most dramatic indication of insolvency: bankruptcy filings.

www.risk.net/journals Journal of Credit Risk

TABLE 2 Default rate estimates on HY bonds (%).

4 THE BBB DEBT MARKET AND FUTURE CREDIT CONDITIONS

Journal of Credit Risk www.risk.net/journals

(a) Chapter 11 filings of more than US$100 million

Number of Chapter 11 filings 66

(b) Chapter 11 filings of more than US$1 billion

Number of Chapter 11 filings 23

www.risk.net/journals Journal of Credit Risk

FIGURE 5 US BBB-rated corporate bonds outstanding, 2005–18.

BBB-rated bonds outstanding as a percentage

of US investment-grade corporate bonds

Source: Bloomberg Barclays US Corporate Investment Grade Index.

Journal of Credit Risk www.risk.net/journals

TABLE 4 Downgrade vulnerability of BBB-rated bonds based on Z-score as of 2019.

Bond rating equivalents of BBB-rated bonds

From BBB to BB 57/298 (19%) 78/371 (21%)

Total 102/298 (34%) 134/371 (36%)

Source: author’s computations from Capital IQ data.

www.risk.net/journals Journal of Credit Risk

4.1 Ratings inflation or persistent rating overvaluation?

Journal of Credit Risk www.risk.net/journals

General information Financial information

Journal of Credit Risk

Journal of Credit Risk www.risk.net/journals

4.2 Collateralized loan obligations

www.risk.net/journals Journal of Credit Risk

FIGURE 6 US CLOs outstanding, 2006 Q1–2019 Q2.

especially to CCC and to defaults of the portfolio-collateral companies. Indeed, it is

Journal of Credit Risk www.risk.net/journals

Total corporate debt/total equity market cap Total corporate debt/GDP

March 26, 2020: 9.0%.

May 19, 2020: 8.6%.