Rating Methodology
April 2005
Contact
Phone
New York
Jerome S. Fons
Vincent J. Truglia
Christopher T. Mahoney
1.212.553.1653
The Application of Joint Default Analysis to
Government Related Issuers
Introduction
This Rating Methodology describes the extension of joint-default analysis (JDA), as announced in a February 2005
Special Comment, to government-related issuers (GRIs).1 That Special Comment followed a Request for Comment,
issued in December 2004, in which Moody’s proposed incorporation of joint-default analysis to rated, non-structured
finance entities.
We define a GRI as an entity with full or partial government ownership or control, a special charter, or a publicpolicy mandate from the national or local government. An issuer fully or partially owned (or controlled) by a GRI may
also be included in the new approach.
The joint-default methodology represents an elaboration and systematization of Moody’s prior approach to rating
issuers and obligations with full or partial support. It explicitly accounts for a) the GRI’s baseline, or stand-alone, risk
assessment (detailed below); b) the supporting government’s rating; c) an estimate of default correlation between the
two entities; and d) the degree of government support.
The first section of this special comment reviews the methodology underlying joint-default analysis. The second
section characterizes GRIs and provides guidance on implementing JDA for GRIs. The final section offers examples
in which JDA is applied to hypothetical situations.
Overview of Joint-Default Analysis
JDA formally incorporates the following principle: The risk that two obligors will both default should be less than or
equal to the default risk of the stronger obligor. It allows us to approach the problem of rating obligations subject to
the credit strength of two obligors. The most straightforward situation is where one party (the supporter) unconditionally guarantees the obligations of another. But it is not difficult to extend the analysis to situations in which the
strength of the guarantee, or “support,” is less than full.2
1.
2.
Please see Moody’s Special Comment “The Incorporation of Joint-Default Analysis into Moody's Corporate, Financial and Government Rating Methodologies,” February 2005.
The Appendix to this document provides a technical summary of the joint-default methodology.
The methodology, as described in the appendix, relies on conditional default analysis to characterize the credit
dependence between two obligors. We propose a weighting parameter W to represent the degree of dependence
between two obligors’ baseline, or stand-alone, default risk. The more highly dependent – or correlated – the two
obligors’ baseline default risk, the lower the benefits achieved from joint support.
As applied to GRIs, the dependence parameter W is a function of any intrinsic economic relationship between
the GRI and its sponsoring government. It captures shared sources of business or credit risk, but it is independent of
the two parties’ vulnerability to default. That is, while the baseline credit profiles of the GRI and the sponsoring government may change over time, their default dependence need not change.
Extending the analysis to incorporate partial support is accomplished by considering two extremes: no support
and full support. Where support is non-existent, the default risk faced by an investor is simply the baseline default risk
of the GRI. On the other hand, full support reduces default risk to that of the joint-default (i.e., guarantee) situation,
in turn a function of credit dependence, as described above. We therefore model support as a second weighting
parameter S which places the final credit risk somewhere between these two outcomes.
In order to successfully apply JDA, one must have estimates of baseline default risk for each party to a transaction.
In most situations, this is not difficult. But an estimate of the baseline risk of, say, a state-owned railway, becomes
somewhat more complicated, as described below.
At least initially, Moody’s intends to publicly release only ranges for the model inputs (aside from the published
rating of the supporting government). The dependence ratio and support probability will be expressed as being low,
medium or high. Likewise, the baseline risk assessment for the GRI will be expressed as falling within a risk scale
ranging between 1 and 6, with 1 representing the lowest risk. The next section provides an overview of each of these
inputs.
Implementation Guidelines for GRIs
Definition of GRI
In order to apply joint default analysis to a government related issuer, one must first estimate the baseline credit risk of
the underlying obligor. To be considered a GRI, an issuer should meet the following criteria:3
• The issuer should have full or partial (national or local) government ownership or have a charter from the
(national or local) government.4 An issuer fully or partially owned or controlled by a GRI may be considered a GRI.
• The issuer does not have taxing authority.5
Examples of GRIs are state-owned electric utilities, railroads, government-sponsored enterprises (GSEs), development banks and highway authorities.
Baseline Default Risk
In many instances, the continuing operation of a GRI depends on some form of subsidy, tariff or capital support
scheme. Its charter may require the GRI to provide a public service which otherwise would not be met through private enterprise, or if offered privately, might entail unacceptable private costs or pose national security concerns. In
many countries, natural monopolies are prime candidates for GRI status.
In most other applications, an assessment of baseline default risk simply means that the analysis excludes parent,
state or third party support. When applied to a government-related institution that requires a subsidy to survive, the
concept of baseline default risk becomes more complex. Because such institutions would fail absent financial ties to a
supporting government, we have chosen to refine our criteria for baseline risk. In particular:
The baseline risk assessment for a government-related institution measures the likelihood that the issuer will
require an extraordinary bailout. It takes into account all aspects of the entity's existing (or anticipated) business
model, including benefits (such as regular subsidies or credit extension) and/or drags associated with the government relationship.
In other words, the baseline risk assessment for a GRI can incorporate normal operating subsidies and therefore
contemplates the risk that it would need an extraordinary bailout from the government.6 By including maintenance, a
GRI’s financial attributes may be compared to global peers (fully private firms as well as other GRIs) in determining
its baseline default risk.7
3.
4.
5.
2
We exclude from this analysis US Public Finance state and local governments.
Here, partial government ownership is generally considered be 20% or greater.
Sub-national government ratings will be evaluated under joint-default analysis at a later date.
Moody’s Rating Methodology
Separating bailout risk from on-going
financial assistance is one of the challenges
in applying this methodology to GRIs. The
question arises: At what point do subsidies
become a de facto bailout? The guiding principle is that any normal maintenance factored into the baseline risk assessment must
not be viewed as extraordinary “support”
when determining the degree of government support (as discussed below) so as to
avoid double-counting government support
benefits.
Baseline Risk Assessments
A baseline risk assessment is an opinion of the likelihood that an issuer will require an
extraordinary bailout.
Moody's Baseline Risk Assessment Definitions:
1
2
3
4
5
6
Entities with baseline risk assessments of 1 are judged to exhibit minimal credit risk.
Entities with baseline risk assessments of 2 are judged to be of very low credit risk.
Entities with baseline risk assessments of 3 are judged to be of low credit risk.
Entities with baseline risk assessments of 4 are judged to be of moderate credit risk.
Entities with baseline risk assessments of 5 are judged to exhibit substantial credit risk.
Entities with baseline risk assessments of 6 are judged to be of high credit risk.
Default Dependence/Correlation
To calculate the joint-default risk between a GRI and its sponsoring government, one needs an estimate of their default
dependence.8 Maximum possible dependence holds if, given a default by the supporting government, the GRI will
default with certainty. In other words, the baseline credit profiles of the government and the GRI are inextricably
linked. In such a situation, the joint-default risk will equal the sovereign’s default risk. Any ratings on such fully supported obligations would therefore be capped at the sovereign’s rating.
Minimum possible dependence holds if, given a default by the supporting government, the GRI’s default risk
(absent extraordinary support) remains consistent with its baseline default risk assessment. In other words, their
default risks are independent of one another and the joint-default risk is therefore equal to the product of their respective default probabilities.9
One can imagine situations in which the credit profile of a GRI could be independent of the supporting government. A commercially run commodity exporter located in a low-rated developing country would be one example. A
rating committee might assign a dependence factor W of just 20% in such a case.
On the other hand, an electric utility operating in the same country would likely experience high default dependence with the sovereign. Here, a committee might assign a dependence factor W as high as 70% or higher. For situations where there is no compelling guidance in either direction, a dependence factor W of 50% – meaning that the
joint-default risk between the GRI and the sovereign lies halfway between the product of their default risks and the
default risk of the sovereign – is an acceptable choice.
Degree of Support
The final input to the rating process is an assessment of the degree of government support for a GRI. This is the likelihood that the government will step in and bail out a GRI if it were to experience a catastrophic loss. An explicit guarantee would be an example of full support (S=100%). In this case, the default risk faced by a GRI’s bondholders is
simply the joint-default risk of the GRI and the supporting government – in turn, a function of their respective baseline ratings and the dependence factor.
At the other extreme, where support is non-existent, the default risk faced by an investor is simply the baseline
default risk of the GRI. In most cases, however, support for a GRI can not be characterized as a guarantee, in which
case judgment is required to place support along a continuum. We rarely assume that government support for a GRI
is non-existent (S=0%), but is instead a positive value that is itself a function of several factors. Among these are the
percentage of state ownership, national importance of the GRI, privatization status and political tolerance towards
government intervention. The table below provides guidance as to how these and other factors might map to a support assessment.
6.
7.
8.
9.
The sovereign’s baseline default risk is typically represented by its government bond rating. However, a sovereign’s ability to bail out an entity may, under certain circumstances, be stronger than that suggested by the published government bond rating – which focuses on the ability and willingness of the sovereign to make debt
payments as promised. The willingness to support an entity is captured in the analysis by the parameter S, as discussed below.
This holds for GRIs with a well-defined, commercial function. Please refer to Moody’s Industry Rating Methodology for the sector in question.
The term “dependence” as used here is synonymous with, but not exactly the same as, “default correlation.”
In accordance with prevailing practice, we exclude the possibility of negative default correlation. If it were allowed, the minimum possible joint-default probability
would instead be zero.
Moody’s Rating Methodology
3
TYPICAL ATTRIBUTES OF STATE SUPPORT FOR GOVERNMENT-RELATED ISSUERS
LOW SUPPORT
0-30%
MEDIUM SUPPORT
31-70%
HIGH SUPPORT
71-100%
- Less than 51%
- Between 51 and 100%
100%
2 - Privatisation status
- In the process of being privatised
- Legal minimum stake to be
maintained is below 50%
- Will not be privatised within 5
years, or full guarantee required if
stake is reduced
3 - Governance and business model
- Government is an arm's length
shareholder
- Company is managed and funded
on a fully standalone basis
- No differentiating legal status
- No immediate prospects, but
possible over medium-term
(5 years)
- Legal minimum stake is 50%
- History of fund flows between
entity and state (e.g., dividends and
capital injections), but not mandatory
or budgeted
- Management is mostly stateappointed, beyond proportional
representation
- No differentiating legal status
4 - Political tolerance for government
intervention and support
- No government economic
intervention allowed
- No evidence of direct support, or
statements from government that
direct support is likely
- Legislative/regulatory body (e.g. EU)
is very likely to object and prevail
5 - National importance of issuer
- Low
- Financial health of entity is not
strategic to government
- If willing to support, there may be
very significant internal impediments
to timely support (e.g., bureaucratic
processes, unclear or very politicised
mechanisms, etc.)
1 - State ownership
6 - Possible sources of delays in providing
support
- Moderately interventionist
government
- Indirect State support likely (e.g.,
willing to act as "deep-pocket"
shareholder)
- Legislative/regulatory body (e.g.,
EU) likely to raise objections to
additional future support
- Entity is likely "flagship" national
company
- Internal processes are well defined,
but support process can be exposed
to complexity of mechanism (e.g.,
multiple municipalities) or external
interference (e.g., legislation)
- Legal status at or close to EPIC/Ente
Publico, or Authority
- Stable legal status, or assets/
liabilities may be transferred to State
in future
- If no specific legal status, entity
budget relies substantially on
government funding
- Business model is not viable
- Highly interventionist government
- Documented support intentions
(e.g., letters of comfort, consolidation
on government books)
- Entity is key to country's economic
health
- Legislative/regulatory body (e.g.,
EU) not very likely to object
- Avoidance of default deemed
critical to financial reputation of
State
- Support deemed to be timely in all
cases
Application and Examples
We now illustrate how the joint-default methodology would be applied to GRIs. The case studies below use hypothetical issuers and do not necessarily reflect real-world firms or countries.
State-Owned Electric Utility
Consider a 50% state-owned electric utility which is located in a developed country rated Aaa. A rating committee has
determined that, based on its intrinsic financial profile, the baseline default risk for the utility is a Baa2 risk. This rating incorporates a statutory tariff enjoyed by the firm, but it excludes the likelihood of an extraordinary bailout.
The default dependence between the utility and the state is estimated to be medium at W=50%, reflecting the fact
that the linkage between electricity demand and the country’s overall economic performance is thought to be moderate, as well as the moderate risk that the government would implement price controls coincident with a sovereign
default on local currency instruments.
Finally, it is estimated that the probability that the government would bailout bondholders in the event of a failure
by the utility is moderate, and the committee therefore votes on a value of S=60%. For this combination of inputs, the
resulting, supported rating is A3.
State-owned Oil Refinery
We now consider an oil refiner, 100% owned by the government. Assume that the government itself is rated A3. A
rating committee determines that the baseline default risk of the refiner is equivalent to Baa3. Dependence is considered low at W=25% and support is considered relatively high at S=75%. Here the GRI’s supported rating would equal
Baa1, one notch below the supporting government.
State Railway
Many railway systems throughout the world operate at a loss, with government subsidies required to maintain operations and debt service. Consider a state-owned railway system located within a country rated A2. A rating committee
has estimated that the railway’s baseline default risk is equivalent to a Ba1 risk. This risk assessment incorporates normal subsidies, but excludes any support likely to be extended in the event of a catastrophe.
The default dependence between the railway and the sovereign is considered to be moderate and a rating committee has agreed on a value for W of 50%. State support is thought to be relatively strong, resulting in an estimate for S
of 85%. The resulting, supported rating for the GRI would therefore be Baa1.
4
Moody’s Rating Methodology
Appendix – A Review of Joint-Default Analysis
CONDITIONAL DEFAULT PROBABILITIES
The probability that two parties will jointly default depends on a) the probability that one of them defaults, and b) the
probability that the second will default, given that the first has already defaulted. Expressed algebraically, one can write
this for events A and B as:10
P(A and B) = P(A | B) x P(B)
(1)
Or equivalently,
P(A and B) = P(B | A) x P(A)
(2)
We define A as the event “obligor A defaults on its obligations” and B as the event “obligor B defaults on its obligations.” Likewise, “A and B” is the joint-default event “obligors A and B both default on their obligations.”11 The operator P(•) represents the probability that event “•” will occur and P(• | *) is defined as the conditional probability of
event “•” occurring, given that event “*” has occurred.
Moody’s ratings can be used to infer directly the probability that a particular issuer will default (P(A) and P(B)).12
But in order to estimate the conditional default probabilities P(A | B) and P(B | A), one must take into account the
relationship between the drivers of default for both obligors. Each of these four probabilities – P(A), P(B), P(A | B)
and P(B | A) – are intended to represent unsupported risk measures. That is, they represent the likelihood of an obligor default in the absence of any joint support or interference. We present in the Applications and Examples section
below a framework for modeling both support and interference.
Although in theory, one can tackle this problem directly by estimating either one of the conditional default probabilities described in equations (1) and (2), it may be more intuitive to focus on the product of the conditional probability of default for the lower-rated, or supported, firm and the unconditional probability of default for the higher-rated,
or supporting, firm. Using L to denote the event “lower-rated obligor L defaults on its obligations” and H to denote
“higher-rated obligor H defaults on its obligations,” we can rewrite equation (1) as:
P(L and H) = P(L | H) x P(H)
(3)
It is not difficult to imagine situations where the conditional probability P(L | H) might be at its theoretical maximum (i.e., 1) or at its minimum (i.e., P(L)).13 Let us consider these extreme outcomes in turn by way of example.
• P(L | H) = 1. Suppose that the financial health of an issuer is crucially linked to the operations of another,
higher-rated entity. For example, the default risk of a distributor in a competitive distribution market dominated by a single supplier may be highly dependent on the financial health of that supplier. In other words,
the conditional probability of the distributor’s default given a default by the higher-rated supplier, P(L | H),
is equal to one. In this case, events L and H are maximally correlated.14 Under such a scenario, the jointdefault probability P(L and H) in equation (3) above is simply P(H). That is, the rating applied to such
jointly supported obligations would equal the supplier’s rating, without any ratings lift, regardless of issuer
L’s standalone rating.
10. Statisticians will recognize these equations as axioms of probability theory that underlie Bayes’ Theorem.
11. The implication here is that the default events occur simultaneously, but we require only that the timing be such that a holder of the supported obligation suffers credit
loss within a specified horizon.
12. Moody’s ratings are defined as ordinal (or relative) measures of default risk and not in terms of cardinal (or absolute) default rates. However, as long as ratings can
provide a constant measure of relative default risk, with actual default probabilities rising and falling proportionately by rating category over a credit cycle, the methods
proposed here will produce logically consistent measures of jointly supported ratings.
13. Technically, the conditional default probability P(L | H) could be as low as zero, a situation which would occur if the default correlation between the two obligors was at
its theoretically maximum negative value. However, throughout this discussion, we follow the standard practice of ignoring the highly unlikely possibility that the default
experience of the two obligors will be negatively correlated.
14. This use of the term “correlation” applies to default events that follow a binomial distribution and should not be confused with potential correlation in rating transitions
(or default intensities). When the default profiles of two obligors are maximally correlated, P(L | H) = 1 and P(H | L) = P(H)/P(L). That is, the weaker entity always
defaults when the stronger entity defaults, and the stronger entity will only default if the weaker entity also defaults. This leads to the result P(H | L) = P(H)/P(L). Note
that maximum correlation will be less than 1 in cases where obligors have different ratings.
Moody’s Rating Methodology
5
•
P(L | H) = P(L). Suppose a highly rated European bank provides a letter of credit to a lower-rated agribusiness in the US. While there may be circumstances in which the agribusiness might face financial difficulties
on its own, its intrinsic operational health is generally unrelated to the circumstances that might lead the
European bank to default on its obligations. Under this scenario, the conditional probability of a default by
the agribusiness, given a default by the bank – i.e., P(L | H) – is simply the standalone default risk P(L) of
the agribusiness. That is, events L and H are uncorrelated and independent of one another. In this case,
their joint-default probability is the product of their standalone default probabilities, P(L)*P(H). The
jointly supported obligation rating implied by such a relationship is generally higher than the rating of the
supporting entity H.
In practice, the conditional default risk of the lower-rated entity, given a default by the stronger entity, will vary
somewhere between these two extremes, maximum correlation (i.e., where P(L | H) = 1) and independence, (i.e.,
where P(L | H) = P(L))
INTERMEDIATE LEVELS OF CORRELATION
We propose here a simple tool for modeling intermediate cases of default risk linkage. Let us denote the variable W as
a correlation weighting factor, where W = 1 corresponds to a maximum theoretical correlation between the default of
the lower-rated entity and that of the higher-rated entity; and W = 0 corresponds to a complete independence (i.e.,
zero correlation) between default events. Fractional values of W indicate intermediate levels of correlation between
the two default events.
Using the correlation weighting concept, we can express the joint-default probability between obligors L and H as:
P (L and H) =W* P(L and H | W=1) + (1-W)* P(L and H | W=0)
(4)
Or more compactly,
P(L and H) = W*P(H) + (1 - W)*P(L)* P(H)
(5)
In other words, once we have determined standalone ratings for the two obligors, the task of assigning a rating to
a jointly supported obligation may be reduced to the assignment of a correlation weight.15
PARTIAL SUPPORT
In many cases, an obligation benefits from external support, but that support falls short of an iron-clad guarantee.
Examples include bonds issued by a weak subsidiary of a relatively strong parent firm, or bonds issued by an issuer with
partial government ownership. In the latter case, the government's incentive to bail the issuer out, should it run into
difficulties, may be a function of the share of government ownership or of the importance of that issuer to the national
economy.
It is helpful to think of the two extreme situations in which an investor faces losses. The first is where the issuer of
the obligation defaults and there is no external support. The probability of this event occurring is simply P(L), the
probability that issuer L will default on its own. The second is where there is full support, but both the issuer and the
support provider default on their obligations. As above, this is given by P(L and H). The degree of support can also be
thought of as a probability and can therefore vary between 0 and 1. We model the risk to the investor as a shifting
probability between the two risk outcomes P(L) and P(L and H):
P(L and H | S) = (1-S)*P(L)+S*P(L and H)
(6)
Here, the weighting parameter S represents the likelihood of support. Full support (i.e., S = 1) leads to the jointdefault outcome and no support (i.e., S = 0) yields the standalone default risk of the obligor, P(L).
15. While this derivation focused on P(L | H), it could also be approached through a focus on P(H | L). (See footnote 15.) An alternative methodology is described in a
paper published by Douglas Lucas, “Default Correlation and Credit Analysis,” The Journal of Fixed Income, Vol. 4, No. 4, March 1995.
6
Moody’s Rating Methodology
Related Research
Special Comment:
The Incorporation of Joint-Default Analysis into Moody's Corporate, Financial and Government Rating
Methodologies, February 2005 (91617)
To access any of these reports, click on the entry above. Note that these references are current as of the date of publication of this
report and that more recent reports may be available. All research may not be available to all clients.
Moody’s Rating Methodology
7
To order reprints of this report (100 copies minimum), please call 1.212.553.1658.
Report Number: 92432
Author
Production Manager
Jerome Fons
William L. Thompson
© Copyright 2005, Moody’s Investors Service, Inc. and/or its licensors including Moody’s Assurance Company, Inc. (together, “MOODY’S”). All rights reserved. ALL INFORMATION
CONTAINED HEREIN IS PROTECTED BY COPYRIGHT LAW AND NONE OF SUCH INFORMATION MAY BE COPIED OR OTHERWISE REPRODUCED, REPACKAGED, FURTHER
TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY SUCH PURPOSE, IN WHOLE OR IN PART, IN ANY
FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODY’S PRIOR WRITTEN CONSENT. All information contained herein is obtained by
MOODY’S from sources believed by it to be accurate and reliable. Because of the possibility of human or mechanical error as well as other factors, however, such information is provided “as
is” without warranty of any kind and MOODY’S, in particular, makes no representation or warranty, express or implied, as to the accuracy, timeliness, completeness, merchantability or fitness
for any particular purpose of any such information. Under no circumstances shall MOODY’S have any liability to any person or entity for (a) any loss or damage in whole or in part caused by,
resulting from, or relating to, any error (negligent or otherwise) or other circumstance or contingency within or outside the control of MOODY’S or any of its directors, officers, employees or
agents in connection with the procurement, collection, compilation, analysis, interpretation, communication, publication or delivery of any such information, or (b) any direct, indirect,
special, consequential, compensatory or incidental damages whatsoever (including without limitation, lost profits), even if MOODY’S is advised in advance of the possibility of such
damages, resulting from the use of or inability to use, any such information. The credit ratings and financial reporting analysis observations, if any, constituting part of the information
contained herein are, and must be construed solely as, statements of opinion and not statements of fact or recommendations to purchase, sell or hold any securities. NO WARRANTY,
EXPRESS OR IMPLIED, AS TO THE ACCURACY, TIMELINESS, COMPLETENESS, MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OF ANY SUCH RATING OR OTHER
OPINION OR INFORMATION IS GIVEN OR MADE BY MOODY’S IN ANY FORM OR MANNER WHATSOEVER. Each rating or other opinion must be weighed solely as one factor in any
investment decision made by or on behalf of any user of the information contained herein, and each such user must accordingly make its own study and evaluation of each security and of
each issuer and guarantor of, and each provider of credit support for, each security that it may consider purchasing, holding or selling.
MOODY’S hereby discloses that most issuers of debt securities (including corporate and municipal bonds, debentures, notes and commercial paper) and preferred stock rated by
MOODY’S have, prior to assignment of any rating, agreed to pay to MOODY’S for appraisal and rating services rendered by it fees ranging from $1,500 to $2,400,000. Moody’s Corporation
(MCO) and its wholly-owned credit rating agency subsidiary, Moody’s Investors Service (MIS), also maintain policies and procedures to address the independence of MIS’s ratings and rating
processes. Information regarding certain affiliations that may exist between directors of MCO and rated entities, and between entities who hold ratings from MIS and have also publicly
reported to the SEC an ownership interest in MCO of more than 5%, is posted annually on Moody’s website at www.moodys.com under the heading “Shareholder Relations — Corporate
Governance — Director and Shareholder Affiliation Policy.”
8
Moody’s Rating Methodology