CreditMetrics and Constant Level of Risk Sep 2010

CreditMetrics® and Constant Level of Risk
Analyzing and benchmarking the effect of rebalancing under the Incremental Risk
Charge
RiskMetrics Group is now part of MSCI Inc.
September 2010
Christopher C. Finger
MSCI Applied Research

© 2010 MSCI. All rights reserved. 1 of 12
Please refer to the disclaimer at the end of this document.
| September 2010
Introduction
Among the new contributions to banking capital standards, the Incremental Risk Charge (IRC)
1
stands out for its reliance on the internal risk models of banks. The high level intent of this
charge is evident, and the Basel Committee provides some level of guidance on the details, but
largely this charge is left to the banks to define and defend. In the past, model-based capital
standards were based on short forecast horizons, and strict empirical backtesting was possible.
With the IRC, banks are faced with the problematic combination of a complex modeling task and
forecast horizons that are too long to admit rigorous empirical validation. This combination can
lead to intuitive and subjective modeling arguments, and subsequently to suspicions on the part
of banks and supervisors alike. In this piece, we will attempt to provide a framework for
interpreting the IRC requirements, and to evaluate the impact of the various potential modeling
assumptions on the ultimate capital charges produced.
At first glance, the IRC is a standard portfolio risk measure: the Value-at-Risk at 99.9%
confidence and a one-year horizon, accounting for changes due to defaults and credit migrations.
However, because the IRC is to cover trading portfolios, the definition acknowledges that it is
unrealistic to assume that a bank will hold these positions without trading or mitigating risk for the
full year. The IRC definition stipulates that according to the portfolio’s liquidity, a bank will
rebalance over the course of the year to maintain a “Constant Level of Risk” (CLoR). The
frequency of the rebalancing is governed by the liquidity horizon of the positions – that is, the
horizon over which a position can be established, closed or hedged away completely without
suffering a material market impact. Presumably, other things being equal, more liquidity is a good
thing, and the rules should assign less capital to more liquid portfolios. This is not explicit in the
definition, however, but is only implicit in the description of the rebalancing. The dependence of
the IRC on liquidity, and indeed on most relevant factors, falls ultimately to an exercise left to the
banks themselves: the interpretation of the Constant Level of Risk.
In the first half of this piece, we discuss of the Constant Level of Risk rebalancing, identifying
three mechanisms by which rebalancing changes portfolio risk over the one-year horizon.
Following this discussion, we analyze three representative trading portfolios, under a number of
different model specifications, in order to attribute the impact of rebalancing to the three
mechanisms we have defined. In addition, we compare our results to the Basel banking book
(that is, IRB) capital rules, and to another portfolio model from the literature.
Rebalancing the Shape of Credit to cover the required sources of risk, the IRC model should treat
as the “states of the world” the future rating states (including default) of the credits underlying the
portfolio. All other potential sources of risk are static. The rating states serve two purposes in the
model: first, they determine the likelihood of default or rating transitions over the relevant risk and
2
liquidity horizons; second, they determine the market value of the positions in the portfolio.
Consider a trader holding a single A-rated bond position with a six-month liquidity horizon. To
model the risk of the bond over six months, we utilize the set of six-month transition probabilities
associated with the bond’s current rating. After six months, the bond may default or downgrade
or stay in the same rating. Under an assumption of constant positions, the trader holds the bond
for the second six months, and the likelihoods of default and transition over this new period would
depend on the bond’s rating at the beginning of this period. Thus if the trader maintains the same
position, the risk for the second period would differ from the risk for the first. In order to maintain
a CLoR, we assume the trader rebalances into a new bond whose risk over the second period is
the same as the risk of the original bond over the first period. In our example, after six months,
assuming no default, the trader rebalances into another (perhaps hypothetical) newly A-rated
1
See Basel Committee (2009) for a full description of the IRC.
2
It is not necessary that we build the model on rating states, but it is necessary that we have a model to capture these two aspects. For
purposes of exposition, we will utilize the ratings framework through the remainder of this article.

Please refer to the disclaimer at the end of this document. RV0710
| September 2010
bond, rather than remaining invested in a bond that is either an aged A-rated bond, or a bond that
has drifted to a different rating category. In the end, rebalancing to a CLoR means the trader is
exposed to the same six-month risk twice, rather than a full twelve months of risk on the same
bond. The trader repeats the short term rather than experiencing the long term.
In this context, we can view rebalancing as a mechanism to counter the drift in the risk profile of
individual assets. Of course, this is not necessarily beneficial, but rather depends on what we
assume about this drift. Typically, the data at our disposal are default probabilities over an
annual horizon. To obtain probabilities for shorter horizons, we work with conditional default
probabilities; for example, the probability of a default in month seven, conditional on survival until
month six. The simplest method is to assume these probabilities are constant in time. Under this
assumption, rebalancing has no effect; there is no difference between repeating the short term
and experiencing the long term.
One criticism of this approach is that it does not match the common intuition about the evolution
of credit. Particularly with high quality names, we expect that the longer a credit stays alive, the
more likely it is to migrate into poorer rating categories; said differently, the highest quality credits
are less likely to jump suddenly to default than they are to deteriorate through the entire credit
spectrum. The conditional default probabilities should thus increase with time. One method to
produce non-trivial term structures of conditional default probabilities is to infer them from the
annual default and transition probabilities together. Assuming that transitions behave according
to a Markov process – essentially, that the likelihood of transition or default depends only on the
current rating, not on the history of ratings, and that future transitions are independent of past
3
ones – we can infer shorter horizon default and transition probabilities from the annual ones.
We display a number of conditional default probability curves obtained from the Markov
assumption in Figure 1. As expected, for the highest rating categories, the curves slope upward.
Rebalancing with these credits counters this drift, leaving the trader exposed only to the short,
sudden default events. For the lowest rating category, the curve slopes downward, indicating that
the longer the credit survives, the more likely it is to have improved. For these credits, the
rebalancing actually increases risk relative to holding a constant position.
Rebalancing and Portfolio Effects

In a portfolio context, there is another mechanism for rebalancing to influence the risk profile.
Consider two bonds with highly correlated issuers. Suppose that in the first half of the year, one
of the bonds defaults, but the other does not. Given the high correlation, we expect that the
second bond would have at least deteriorated, and that its chance of defaulting in the second half
of the year would have increased. This effect is distinct from the drifting we discussed earlier:
even if there is no drift in the standalone default probabilities, the presence of correlation should
make the conditional default probability in the second period greater if there have been defaults in
the first period by related issuers. If the trader were to hold the second bond for the second
period, he would experience this higher conditional default probability, and thus experience a
correlation across the two periods. Instead, we assume the trader rebalances, reinvesting into a
bond whose default probability is “reset” back to what was experienced in the first period. This
does not always reduce risk, however: in cases where the two correlated issuers both improve in
the first period, rebalancing sets them back to their initial (higher) default probabilities, increasing
the risk for the second period.
The rebalancing that we have described takes a strict view on the interpretation of CLoR.
Specifically, we assume that the bank manages its portfolio to have constant risk conditionally,
3
There are in fact a number of numerical subtleties to this. See Kreinen and Sidelnikova (2001) for further details.

| September 2010
that is, in any economic scenario. If in the first part of the year, the general economy
deteriorates, causing the trader to expect more defaults in the latter part, he would shift into
higher-rated assets in order to preserve the overall default probability. This rebalancing activity
essentially breaks any links (that is, any correlations) from one period to the next. Holding a
constant set of positions, we might argue that there is some autocorrelation across periods, and
indeed, this is the modeling framework proposed by Dunn (2008). While we acknowledge that
such autocorrelation may exist in the economy, we assert that the implication of a CLoR strategy
is always to rebalance back to the same absolute risk profile. Thus, even if the economy
experiences autocorrelation, the returns on the rebalanced portfolio should not.
Paying for Reduced Risk

Thus far, we have described two mechanisms where the constant level of risk rebalancing has
the potential to decrease risk: 1) rebalancing counters the drift of individual issuers to higher
default probabilities across the term structure, and 2) it breaks the correlation across periods
between the deterioration of different issuers. However, if this rebalancing strategy reduces risk,
then presumably it should carry some cost; the risk reduction should not be a free lunch. A
realistic model should account for this cost, and this is where rating migrations come into play.
Consider again the example of a single A-rated bond, priced currently at 100. Suppose a default
occurs in the first period, and the bond falls to some recovery value (for example, 40). To
rebalance to a CLoR, the trader adds 60 to the portfolio, and holds a new A-rated bond at 100 for
the second period. Here, there has been no risk reduction from the rebalancing; the trader simply
realizes the loss due to the default event. What if the bond does not default? The trader still
rebalances to a new A-rated bond priced at 100, but absent a model for rating transitions, we
have no information about the bond’s price at the end of the first period. The natural assumption
is that the bond is still at 100, meaning the rebalancing – that is our lunch – is free.
4
If instead we model rating migrations, then we would know, for instance, that the bond had
downgraded to a B rating in the first period. In this case, the trader rebalances by selling the B-
rated bond (say for 80), adding 20 to the portfolio and purchasing a new A-rated bond at 100. In
this case, the rebalancing reduces risk by shifting from a B-rated to an A-rated bond for the
second period, but this risk reduction comes at the cost of realizing the loss due to the
downgrade. Remember that for highly rated issuers, as time passes and no default occurs, the
expectation is that the default probability for subsequent periods increases. Thus even if no
default occurs, we expect the credit to deteriorate and the rebalancing to counter this
deterioration, at the cost of realizing a loss on the position.
One point to clarify is the notion of “simply” adding to the portfolio to compensate for losses.
First, adding to the portfolio to compensate for losses is a necessary component of CLoR. The
motivation for CLoR, as we understand it, is to model the bank’s trading activities, assuming that
it will continue to trade, make markets, and otherwise service its clients, even in the face of
significant losses. The rebalancing must therefore achieve the objective of putting the same
amount at risk in each period. This can lead to the apparent paradox of losses being greater than
the initial portfolio (for instance, the trader could experience two defaults on the A-rated bond, for
a loss of 120), meaning in special cases, the capital requirement could be greater than the
portfolio value. We resolve the paradox by observing that capital here is not simply to support
losses on the trading portfolio, but rather to support losses on the trading business, including
positions that may be put on in the future.
4
Or any other notion of credit quality changes short of default.

| September 2010
Second, we have assumed that the additions to the portfolio are free. This is clearly a
simplification. A slight modification could be to account for the bank’s funding cost in these
additions. Assuming a fixed level for this funding cost would produce the same effect as
assuming slightly greater losses accompanying ratings downgrades. A richer treatment would be
to assume the funding cost is tied to aggregate losses, capturing the presumed liquidity pressure
the bank would experience in the most adverse scenarios. We will not consider this effect further.
Measuring the Effects

To this point, we have discussed a number of mechanisms by which the constant level of risk
rebalancing can impact portfolio risk or capital, with some of these appearing to reduce risk and
others appearing to increase it. To quantify the impact of the different mechanisms, we analyze
three sample portfolios under a variety of modeling assumptions. The basic model we work with
in all cases is the CreditMetrics® framework, with the default and migration probabilities specified
exogenously, and the dependencies between defaults and migrations driven by the correlated
underlying asset values of the issuers. We assume that movements of asset values, and thus
migrations, are independent across time periods.
In order to partially isolate the effect of the different rebalancing mechanisms, we consider three
distinct model specifications. In each case, we utilize the same annual default probabilities and
the same asset correlations. First, we model defaults only (without ratings migration), and scale
the default probabilities to shorter liquidity horizons, while assuming constant conditional default
probabilities. Under this specification, as we increase the rebalancing frequency, there is no drift
effect (since returning to the initial default probability is the same as continuing along the curve),
and there is no cost due to ratings moves. Thus, under this specification, as we increase the
rebalancing frequency, the only mechanism at work is the breaking of the correlation in
deterioration of different issuers across time.
Under the second specification, we continue with the default-only model, but scale the default
probabilities based on the assumption that ratings migrations follow a Markov process, as
discussed earlier. Under this specification, we see the effect of the correlation breaking, as
before, and also the effect of countering the drift. Our third specification adds the impact of
migrations, and thus covers all three of the rebalancing mechanisms. For each specification, we
analyze the portfolio under the assumption of quarterly, semiannual and annual rebalancing.
For comparison, we also examine the capital required under the Basel II Internal Ratings Based
(IRB) formula. This is the capital treatment that would apply if the portfolio were to be reclassified
as part of the banking book. The IRB formula is based on a simple portfolio default model, with
the assumption that the portfolio is well diversified (in the sense that there are no position size
concentrations). The IRB “model” is highly stylized, but it is based on the same underlying
CreditMetrics-style structure as our model, meaning that it also works with issuer asset
correlations. The asset correlations in the IRB model are set between 12% and 24% as a
5
function of default probability. The output of the formula is then adjusted to account for maturity.
We examine three versions of the IRB formula: the standard output with no maturity adjustment,
which only accounts for default risk; the standard output, including the maturity adjustment; and
the maturity-adjusted formula, but with the correlation values replaced by our model assumption.
The IRB formula only accounts for long positions, and does not penalize for concentrations in
position size.
We analyze the same set of benchmark portfolios as Dunn (2008). The first portfolio – “Long
Only” – contains 87 positions, with 70% of the exposure to investment grade issuers. Within each
5
See Basel Committee (2005) for further details.

| September 2010
rating category, the exposures to the individual issuers are equal. The second portfolio – “Long
Bias” – contains the same long positions as “Long Only,” but also contains short positions in each
rating category, producing an overall exposure mix of 60% long and 40% short. The third
portfolio – “Long Bias with Lumps” – contains the same overall exposure mix as “Long Bias,” but
rather than a uniform distribution of exposures, contains three concentrated long positions whose
size is five to seven times as large as the smaller positions. Note that since the three portfolios
each contain the same total long exposures by rating category, they are charged equal levels of
capital under the IRB approach.
Dunn analyzes only the impact of defaults, and so does not need to specify the maturity of the
positions. Because we will analyze the impact of rating migrations, and their accompanying
changes in credit spread, we need to specify the sensitivity of the portfolios to spread moves. We
assume all positions have a four-year spread duration and five-year final maturity. Note as well
that the positions each reference a unique issuer, and as such should be understood as net
positions across an issuer. The expected reduction in risk arising from the short positions derives
from the positive correlation between the short and long positions, rather than deriving from an
explicit risk offset.
We fix the overall correlation level at 30% – higher than the IRB levels, but lower than typical
levels for tranched corporate credit indices – and compute the 99.9% VaR under all model
specifications using Monte Carlo with 100,000 scenarios. We report the VaR as a proportion of
the face value of the long portion of the portfolio. The results are displayed in Table 1.
Beginning with the Long Only portfolio, we track the impact of the various rebalancing
mechanisms as we increase the liquidity of the portfolio (or equivalently, the frequency of the
rebalancing). Under the first model specification, where the only mechanism at work is the
correlation breaking, we see a reduction of capital from 9.8% to 8.2% (a relative reduction of
16%) as we move from annual to quarterly liquidity. From the 8.2% capital level, we obtain a
further reduction of 5% (to 7.8%) by moving to the second model specification, and thus layering
on the effect of the drift cancelation. As we expect, this leads to a capital reduction, since the
conditional default probability curves are upward sloping for the majority of the portfolio. Finally,
adding migration raises the capital to 10.1% (a relative change of 30%). Overall, beginning from
the simple default model with no rebalancing, the benefits of liquidity are more than offset by the
additional risk due to non-default events.
For the Long Only portfolio, regardless of the assumed liquidity, the IRC capital is greater than
what would be charged by the IRB formula; the portfolio, even with high liquidity, would be
charged more capital as part of the trading book than as part of the banking book. With no
rebalancing, the IRC charge is 13.1%, while the IRB charge is only 9.3%. However, notice that if
we replace the IRB correlation levels with the 30% we have used for the IRC model, the IRB
charge becomes 14.6%. The difference with respect to our IRC model at this point is primarily
due to the differing treatments of maturity and migration. From this, we can conclude that for the
Long Only portfolio, the IRC produces greater capital than the IRB, driven primarily by the higher
correlation level we selected.
Moving to the Long Bias portfolio, the first thing we observe is that the IRC capital charge is in all
cases less than the IRB charge. The impact of the IRC’s higher correlation, which we observed
with the Long Only portfolio, is more than offset by the impact of the short positions, which are not
accounted for under the IRC. The effect of rebalancing on this portfolio is more muted, however.
We see a reduction only from 5.1% to 4.9% under the first model specification as we move from
annual to quarterly liquidity, and a further reduction only to 4.7% with the incorporation of the
default probability curves. The difference between the default and migration approaches is less
than it was for the Long Only portfolio, indicating in some sense that the migration (that is, mark-
to-market) changes are better offset by the short positions than are the default events.

| September 2010
The size concentrations in the third portfolio produce expected effects. Capital is in general
higher – by 35% to 45% depending on the model specification – as a result of the concentrated
positions. These are still lower than the IRB capital, however; the lower correlation and lack of a
concentration penalty in the IRC are not enough to compensate for its neglect of the short
positions. Liquidity has a lesser impact, with capital reductions of less than 10% for moving from
annual to quarterly rebalancing. We also note that the contribution of the migration effects to total
IRC capital is lower again than with the Long Bias portfolio; the impact of the potential large
defaults of the long positions is driving much of the capital.
Comparing to an Alternative Model

Finally, since we have used the same portfolios as Dunn (2008), it is worth comparing our results
to his. Dunn’s model covers only default losses, but does account for the shape of the default
probability curves. For comparison, we work with our second model specification, incorporating
the default probability curves that Dunn specifies on page 15. Dunn’s model also builds the
dependence between issuers through a CreditMetrics approach, with underlying asset variables
driving default.
There is one important difference between our model and Dunn’s. In our case, the correlation is
purely contemporaneous. Moves in the underlying assets of distinct firms over the same time
period are correlated, but there is no correlation across different time periods. In Dunn’s
formulation, there is both contemporaneous and temporal correlation. The temporal correlation
links movements in a general market factor from one period to the next. Dunn specifies a
calibration in order to produce a desired annual correlation (such as the 30% we have used) by a
combination of a non-zero temporal correlation in the market factor and a lower level of short-
term contemporaneous correlation across issuers. We have argued already that a true CLoR
rebalancing would counteract any temporal correlation, but nonetheless proceed with our
comparison.
In the tables on pages 17-18, Dunn reports capital for the three portfolios of 5.8%, 3.6% and 5.7%
respectively under quarterly rebalancing and a 30% annual correlation. Under our model, we
obtain 6.2%, 3.8% and 6.0%. In these examples, the contemporaneous correlation seems to play
a dominant role, and our model, with only this effect, produces somewhat higher capital than
Dunn’s with a blend of the two correlation sources.
Conclusion
We have presented a framework for interpreting the Constant Level of Risk rebalancing scheme
in the Incremental Risk Charge, identifying three distinct mechanisms – default probability drift,
correlation breaking and migration costs – that govern the effect of rebalancing on long-term risk
and capital. We have then investigated the impact of these mechanisms on a number of stylized
portfolios, and compared these to the capital that would be assessed to these portfolios under the
banking book treatment. Overall, we see a mild capital reduction coming from rebalancing, and a
significant impact from incorporating non-default events into the model. For long-short portfolios,
the IRC approach does appear to provide a capital benefit relative to the banking book model,
though this is largely due to the IRC’s coverage of short positions, and only to a lesser extent due
to the liquidity-related concessions.

| September 2010
Figure 1: Monthly conditional default probabilities.
Source: Author’s calculations based on S&P annual transition matrix.

| September 2010
Table 1: VaR at 99.9% with a 30% correlation.
Long Bias
Long Only Long Bias w/ Lumps
Annual step
Default, const haz 9.8% 5.1% 7.2%
Default, Markov 9.8% 5.1% 7.2%
Migration 13.1% 6.2% 7.9%
Semiannual step
Migration 11.6% 5.9% 7.9%
Quarterly step
Migration 10.1% 5.4% 7.4%
IRB Capital
Default only 5.7% 5.7% 5.7%
Standard 9.3% 9.3% 9.3%
Corr adjusted 14.6% 14.6% 14.6%
Source: Author’s calculations.

| September 2010
References
Basel Committee on Banking Supervision (2005). An Explanatory Note on the Basel II IRB Risk
Weight Functions. July.
Basel Committee on Banking Supervision (2009). Guidelines for Computing Capital for
Incremental Risk in the Trading Book. July.
Dunn (2008). A Multiple Period Gaussian Jump to Default Risk Model. FSA Occasional Papers
Series. August.
Kreinen and Sidelnikova (2001). Regularization Algorithms for Transition Matrices. Algo
Research Quarterly. March.

| September 2010
Contact Information
clientservice@msci.com
Americas
Americas 1.888.588.4567 (toll free)

Atlanta + 1.404.551.3212
Boston + 1.617.532.0920
Chicago + 1.312.675.0545
Montreal + 1.514.847.7506
Monterrey + 52.81.1253.4020
New York + 1.212.804.3901
San Francisco + 1.415.836.8800
Sao Paulo + 55.11.3706.1360
Stamford +1.203.325.5630
Toronto + 1.416.628.1007
Europe, Middle East & Africa
Amsterdam + 31.20.462.1382
Cape Town + 27.21.673.0100
Frankfurt + 49.69.133.859.00
Geneva + 41.22.817.9777
London + 44.20.7618.2222
Madrid + 34.91.700.7275
Milan + 39.02.5849.0415
Paris 0800.91.59.17 (toll free)
Zurich + 41.44.220.9300
Asia Pacific
China North 10800.852.1032 (toll free)

China South 10800.152.1032 (toll free)
Hong Kong + 852.2844.9333
Seoul + 827.0768.88984
Singapore 800.852.3749 (toll free)
Sydney + 61.2.9033.9333
Tokyo + 81.3.5226.8222
www.msci.com | www.riskmetrics.com

| September 2010
Notice and Disclaimer
 This document and all of the information contained in it, including without limitation all text, data, graphs, charts
(collectively, the “Information”) is the property of MSCl Inc., its subsidiaries (including without limitation Barra, Inc.
and the RiskMetrics Group, Inc.) and/or their subsidiaries (including without limitation the FEA, ISS, and CFRA
companies) (alone or with one or more of them, “MSCI”), or their direct or indirect suppliers or any third party
involved in the making or compiling of the Information (collectively (including MSCI), the “MSCI Parties” or
individually, an “MSCI Party”), as applicable, and is provided for informational purposes only. The Information may
not be reproduced or redisseminated in whole or in part without prior written permission from the applicable MSCI
Party.
 The Information may not be used to verify or correct other data, to create indices, risk models or analytics, or in
connection with issuing, offering, sponsoring, managing or marketing any securities, portfolios, financial products or
other investment vehicles based on, linked to, tracking or otherwise derived from any MSCI products or data.
 Historical data and analysis should not be taken as an indication or guarantee of any future performance, analysis,
forecast or prediction.
 None of the Information constitutes an offer to sell (or a solicitation of an offer to buy), or a promotion or
recommendation of, any security, financial product or other investment vehicle or any trading strategy, and none of
the MSCI Parties endorses, approves or otherwise expresses any opinion regarding any issuer, securities, financial
products or instruments or trading strategies. None of the Information, MSCI indices, models or other products or
services is intended to constitute investment advice or a recommendation to make (or refrain from making) any kind
of investment decision and may not be relied on as such.
 The user of the Information assumes the entire risk of any use it may make or permit to be made of the Information.
 NONE OF THE MSCI PARTIES MAKES ANY EXPRESS OR IMPLIED WARRANTIES OR REPRESENTATIONS
WITH RESPECT TO THE INFORMATION (OR THE RESULTS TO BE OBTAINED BY THE USE THEREOF), AND
TO THE MAXIMUM EXTENT PERMITTED BY LAW, MSCI, ON ITS BEHALF AND ON THE BEHALF OF EACH
MSCI PARTY, HEREBY EXPRESSLY DISCLAIMS ALL IMPLIED WARRANTIES (INCLUDING, WITHOUT
LIMITATION, ANY IMPLIED WARRANTIES OF ORIGINALITY, ACCURACY, TIMELINESS, NON-INFRINGEMENT,
COMPLETENESS, MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE) WITH RESPECT TO
ANY OF THE INFORMATION.
 Without limiting any of the foregoing and to the maximum extent permitted by law, in no event shall any of the MSCI
Parties have any liability regarding any of the Information for any direct, indirect, special, punitive, consequential
(including lost profits) or any other damages even if notified of the possibility of such damages. The foregoing shall
not exclude or limit any liability that may not by applicable law be excluded or limited, including without limitation (as
applicable), any liability for death or personal injury to the extent that such injury results from the negligence or wilful
default of itself, its servants, agents or sub-contractors.
 Any use of or access to products, services or information of MSCI requires a license from MSCI. MSCI, Barra,
RiskMetrics, ISS, CFRA, FEA, EAFE, Aegis, Cosmos, BarraOne, and all other MSCI product names are the
trademarks, registered trademarks, or service marks of MSCI in the United States and other jurisdictions. The
Global Industry Classification Standard (GICS) was developed by and is the exclusive property of MSCI and
Standard & Poor’s. “Global Industry Classification Standard (GICS)” is a service mark of MSCI and Standard &
Poor’s.
© 2010 MSCI. All rights reserved.
About MSCI
MSCI Inc. is a leading provider of investment decision support tools to investors globally, including asset managers,
banks, hedge funds and pension funds. MSCI products and services include indices, portfolio risk and performance
analytics, and governance tools.
The company’s flagship product offerings are: the MSCI indices which include over 120,000 daily indices covering more
than 70 countries; Barra portfolio risk and performance analytics covering global equity and fixed income markets;
RiskMetrics market and credit risk analytics; ISS governance research and outsourced voting and reporting services; FEA
valuation models and risk management software for the energy and commodities markets; and CFRA forensic accounting
risk research, legal/regulatory risk assessment, and due-diligence. MSCI is headquartered in New York, with research and
commercial offices around the world.


CreditMetrics and Constant Level of Risk Sep 2010

Caricato da

Informazioni sul documento

Descrizione originale:

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

CreditMetrics and Constant Level of Risk Sep 2010

Caricato da

Copyright:

Formati disponibili

CreditMetrics® and Constant Level of Risk

RiskMetrics Group is now part of MSCI Inc.

MSCI Applied Research

MSCI Applied Research

Rebalancing and Portfolio Effects

MSCI Applied Research

Paying for Reduced Risk

MSCI Applied Research

Measuring the Effects

MSCI Applied Research

MSCI Applied Research

Comparing to an Alternative Model

MSCI Applied Research

Figure 1: Monthly conditional default probabilities.

Source: Author’s calculations based on S&P annual transition matrix.

MSCI Applied Research

Table 1: VaR at 99.9% with a 30% correlation.

Source: Author’s calculations.

MSCI Applied Research

MSCI Applied Research

Americas 1.888.588.4567 (toll free)

Europe, Middle East & Africa

China North 10800.852.1032 (toll free)

MSCI Applied Research

Notice and Disclaimer

© 2010 MSCI. All rights reserved.

MSCI Applied Research

Potrebbero piacerti anche