Increasingly demanding standards in terms of relevance and scientific rigor

The rating of mutual funds has been growing considerably not only in the United States but also in Europe. It provides an answer to a legitimate expectation on the part of investors and their consultants, who are concerned about comparing the performances and value-added of management offerings that are both increasing in number and becoming increasingly easy to switch in and out of, as a result of the arrival of “open” distribution architecture.

A corollary of the success of ratings and their growing influence on the collection of capital is that users are increasingly demanding with regard to the relevance and methodological rigor of the performance evaluation system implemented.

In point of fact, the classifications of the main rating services on the market no longer meet the principles of reliability demanded by both investors and management companies.

These demands are the following:

• Genuine integration of risk in performance measurement
• Performance persistence
• Robustness and reliability of the results obtained
• Transparency and legibility of the ratings

The EUROPERFORMANCE-EDHEC Style Rating: methodology

The EUROPERFORMANCE-EDHEC Style Rating is put together from three criteria:

• Risk-adjusted performance (or alpha)
• Potential extreme loss (or VAR)
• Performance persistence

For each of these dimensions of the rating, the EUROPERFORMANCE-EDHEC Style Rating relies on state-of-the-art conceptual and technical tools.

1. Measuring alphas

The risk-adjusted performance is measured in a two-step process.

The first step aims to select the style indices that represent the strategic allocation and therefore the risks taken over the rating period. This analysis is carried out in the form of a constrained multi-linear regression such as developed by William Sharpe.

Using this fund-specific style index selection and the related weights (customised benchmark), it is possible to calculate the fund’s excess performance while taking the risks to which it was really exposed into account. In order to do this, we use a multifactor index model which constitutes a robust and practical application of modern portfolio theory (Arbitrage Pricing Theory).

In order to limit the risks of collinearity between the indices, the fund universe was divided into 58 distinct categories using the EuroPerformance classification groupings.

2 Determining the potential extreme loss (VAR)

The concept of Value-at-Risk (VaR) allows all the portfolio risks, which are spread between several asset classes, to be summed up in a single value. While a measure such as the variance characterises the average risk of the portfolio (mean dispersion in the return distribution), the Value-at-Risk refers to a value for the possible loss; in this sense, it is a measure of extreme risk.

The principle behind the VaR concept is therefore simple, but its practical implementation is more complex. The asset returns are assumed to be subject to common risk factors, which enables the portfolio’s return to be decomposed. By modelling the future evolution of these factors, it is possible to estimate the value of the portfolio at a fixed horizon and to deduce its VaR from there.

By estimating a Cornish Fisher-type semi-parametric VAR, the EUROPERFORMANCE-EDHEC Style Rating allow for the integration of the potential (to a 99% threshold) extreme loss for funds that present non-Gaussian return profiles, i.e. funds that do not respect a “normal distribution”, either because they invest in markets where the extreme losses are considerable, or because they use derivative instruments.

3 Performance persistence analysis

The performance persistence measure implemented by the EUROPERFORMANCE-EDHEC Style Rating relies on two indicators:

• The calculation of the gain frequency, which measures the frequency of positive alpha over the whole period on a weekly basis, and which aims to identify the managers who “repeat” their performance
• The Hurst exponent, which is a measure of the regularity of the outperformance, and which aims to evaluate the probability that, at the subscription stage, the outperformance will not be too different from that observed at the decision-making stage

4 Rating eligibility rules

The funds that are eligible for the Style Rating have the following elements:

• at least 3 years of historical weekly return data
• not more than 2 returns missing for the previous 156 weeks
• an adjusted determination coefficient (R2) greater than 70% for the alpha calculation model exept for diversified funds and funds belonging to the international categories whose level is fixed to 60%.
• a Value-At-Risk at the time the rating is calculated that is lower than the average VaR of its analysis category increased by 2 standard deviations

The funds belonging to the following categories are excluded:

• Treasury Funds
• Guaranteed Funds
• Gold and Raw Material Sector Funds

• ETFs and all funds practicing index or alterntive management

Finally, the Convertible Bonds category is not analysed because there is no style index for this category.

5. The rating presentation categories

The scores attributed by the EUROPERFORMANCE-EDHEC Style Rating are independent from any category and, as such, are perfectly comparable, whatever the type of fund rated.

Nonetheless, in order to simplify the presentation of the ratings, six categories have been defined to make them easier to read and to understand.

6. Rules for attributing stars

Rating Attribution Table

Rating	Diminishing Alpha	Average Alpha	Gain Frequency	Hurst Exponent
		>= 0	>= .5	>= .5
		>= 0	>= .5
		>= 0	< .5
		>= - Average Mgt fees
	50% +	< - Average Mgt fees
	50% -	< - Average Mgt fees
VaR too high (V)	Funds whose Value at Risk is greater than the average of the VaR in the analysis category + 2 standard deviations are identified and do not receive any star.
R2 too low (R2 )	In order to be graded, the coefficent of determination resulting from the style analysis must be higher than 70% exept for diversified funds and funds belonging to international categories whose level is fixed to 60%.

The EUROPERFORMANCE-EDHEC Style Rating is updated every month. The behaviours identified by the EUROPERFORMANCE-EDHEC Style Rating are stable and do not require the implementation of rating management rules that limit its variability.

7. Communication

The Style Rating is a measure of the quality of active management. The Style Rating criteria are demanding and fewer than 10% of the funds rated obtain the maximal rating of .

The and categories contain funds that do not outperform their objective on average. These funds were able to deliver positive performances over the period analysed while not exceeding the average performance produced by the indices that are representative of their strategy.

The category contains funds whose performance is very similar to that of the markets in which they are invested. Their outperformance nonetheless allows them to cover the average management fees in their category.

The and categories contain funds that outperform over the analysis period. This excess performance is the fruit of the manager’s decisions: active stock picking and/or market or style timing.

Among these funds that excel, some provide a significant gain frequency that characterises persistence of outperformance. These funds are distinguished by the maximal rating of .

Finally, the (Hurst exponent) category distinguishes funds that regularly outperform their benchmark. The ‘H’ is a decision support tool because it indicates that the investor has a better than average chance, at the subscription stage, of benefiting from good performance of the fund chosen.

The right to use the EUROPERFORMANCE-EDHEC Style Rating is provided free of charge on condition that the source of the rating reproduced is cited. Details on the conditions and limitations of the use of the Style Rating are given on our website.

The Style Rating results are published on the 4th Friday of the month on this website: http://stylerating.com. The site provides details on the evolution of the rated funds and lists the funds excluded from the style rating. The technical documentation is also available on the website.

The cooperation between Europerformance and EDHEC

The EUROPERFORMANCE-EDHEC Style Rating is an association between Europerformance and the Edhec Risk and Asset Management Research Centre.

The Edhec Risk and Asset Management Research Centre was set up in 2001. Its team of 28 research engineers, specialised associate researchers and professors work together on applied research programmes in the areas of asset allocation and performance measurement in the traditional and alternative universes.

Edhec is the scientific adviser to the EUROPERFORMANCE-EDHEC Style Rating.

Presentation of Style Rating

Technical documentation

©2004-2006 Europerformance-Edhec

The Europerformance-Edhec Style Rating
Methodology
April 6, 2006
THE EUROPERFORMANCE – EDHEC RATINGS
WHY FUND RATINGS?
The measurement of the performance of investment funds is an important issue
for professionals, because it is through performance measurement that their
know-how and their products are evaluated.
Ratings have a determining role in the sale of products, as can be seen in the
American market, where more than 70% of subscriptions involve 4 and 5-star
funds. In Europe, ratings have not yet reached that level of importance, but their
influence on the collection of capital is growing. A corollary of this is that
investors are increasingly demanding with regard to the relevance and the
scientific and methodological rigor of the performance evaluation system
implemented.
Ratings respond to a legitimate expectation from investors and their consultants,
and their concern to compare the performance and value-added of management
offerings that are increasing in number and are increasingly easy to switch
between, due to the arrival of “open” distribution architectures.
However, today, the rankings of the main existing rating agencies no longer
satisfy the principles of reliability demanded by both investors and the
management companies.
These demands are the following:
1. Genuine integration of risk in performance measurement
2. Measurement of performance persistence
3. Robustness and reliability of the results obtained
4. Transparency and legibility of the ratings
As an agency for measuring and analysing the performance of investment funds,
it was incumbent upon EuroPerformance to come up with a response to these
requirements. That is why EuroPerformance entered into a partnership with
EDHEC in 2002 to engage in research into performance measurement and style
analysis. This research work led to ratings that take into account both the riskadjusted
performance and the persistence of the results.
1 THE THREE DIMENSIONS OF THE EUROPERFORMANCE-EDHEC
RATINGS
The EUROPERFORMANCE-EDHEC ratings are constructed using three criteria:
o Risk-adjusted performance (or alpha)
o Extreme loss potential (or VaR)
o Performance persistence
Alpha measures the outperformance (“abnormal return” or alpha)
compared to the risks taken by the manager. This magnitude is not directly
observable; it is measured by the difference between the fund’s returns and the
“normal” returns that reward all of the portfolio’s risks. This approach is based on
a multifactor model drawn from the style analysis model developed by the Nobel
Prize winner William Sharpe in 1992.
The decomposition of the performance between the alpha and the reward
for risks is specific to each fund. The alpha is an arithmetic magnitude that is
easy to compare and is independent from the prevailing situation for an asset
class or the definition of a ranking category.
Extreme risks: most existing ratings handle average risks. Whether it
involves volatility or the information ratio, the risk-adjusted measures consider
that the investor is averse to the average risk of the investment but has no
particular opinion in relation to extreme losses. In order to face up to this
insufficiency the Style Ratings implement a Value-at-Risk calculation that
provides a good estimate of the fund’s maximal loss.
Persistence: the ratings are used by investors to make investment
decisions, not to reward past performance. From this point of view, a fund’s
capacity to reproduce its performance is an important element for the person
who uses the ratings. The EUROPERFORMANCE-EDHEC ratings take two
indicators into account to measure a manager’s (or a fund’s) capacity to
reproduce its past performance:
o The capacity to deliver alphas frequently over the calculation period
(gain frequency)
o The regularity of the outperformance (Hurst exponent)
These indicators are absolute measures that are specific to each fund. The
detailed summary represented in the final grade does not depend on a
performance ranking or the fact that is belongs to a category (country in which it
is marketed, awards received, etc.). The fund has a grade that evolves according
to its own performance and not according to its relative positioning in a
predefined universe. Since the grade is not related to a given typology, all
comparisons between funds are justified.
2 MEASURING ALPHAS
To do this, we propose to use the conceptual framework of the performance and
risk style analysis proposed by the Nobel Prize winner William Sharpe (1992).
2.1 Style analysis
William Sharpe (1988, 1992) introduced the following model to provide an
objective breakdown of the manager’s real style, as opposed to the style
breakdown announced by the manager. This method is known as return-based
style analysis.
Model:
where:
• Ri = excess return (net of fees) of a given portfolio or fund
• Fk = excess return compared to index j for period t
• wik = weight of the style (the sum of the weights is equal to one)
• eit = error term
Style analysis is differentiated however from the standard linear regression by
the fact that specific constraints are imposed on the coefficients so that they can
be interpreted directly as weightings:
• Portfolio constraint: the sum of the coefficients must be equal to one
• Positivity constraint: the wik coefficients must be positive
In essence, the model divides the fund’s return into two components:
• The “Style”: wi1F1t + wi2F2t + … wikFkt (the part attributable to market
movements)
• The “Skill”: eit (the part specific to the manager)
The term of “skill” can itself be attributed to:
• The manager’s exposure to asset classes that are not included in the analysis
• The manager’s active anticipations: active selection within the asset classes
and/or timing in relation to these same classes.
it i t i t iK Kt it R = w F + w F + …+ w F + e 1 1 2 2
It is tempting to interpret the term of “skill” or total excess return eit in a style
analysis as a measure of abnormal return. There are however two major
opposing elements:
1. Distortion of the results: Introducing the portfolio and positivity
constraints into the style analysis distorts the results of the standard
regression. As a result, the standard desirable properties for linear
regression models are not respected.
2. The absence of appropriate risk adjustment: As recalled above, an
analysis of that type does not provide an explanation for the abnormal
return, on a risk-adjusted basis.
2.2 The multi-index model
These two imperfections of style analysis are handled by using a factor model,
which constitutes a satisfactory theoretical response to the problem of riskadjusted
performance measurement.
In practice, one important question remains: the choice of factors.
There are four types of factor models:
1. Implicit factor model. In this approach, one carries out a factor analysis
(for example, a principal component analysis) to extract the return series
factors statistically. It is without doubt the best approach, because it
avoids the problems of including the wrong factors or omitting the right
factors. However, the factors can be difficult to interpret.
2. Explicit macroeconomic factor model. In this approach, economic
variables are used as factors. For example, Chen, Roll, Ross (1986), use
the inflation rate, the growth of industrial output, the difference between
long-term and short-term interest rates and the difference in ratings
between bonds.
3. Explicit microeconomic factor model. In this approach, microeconomic
attributes are used as factors. The BARRA model is a popular example of
this type of model.
4. Factor model with explicit factors made up of indices. In this approach,
market indices are used as factors. This is consistent with the idea of
using portfolio returns as factors. The most popular example of this
approach is the CAPM (Sharpe (1964)), where the return on the market
portfolio, approximated by a broad index, is used as a factor. The idea of
using “replicating portfolios” as an approximation of the true factor,
which is not known, is also found in Fama and French (1992).
In this context, in order to provide the methodology with factor transparency,
traceability and easy-to-interpret results, we have chosen to use a multi-index
factor model, which is written as follows:
where:
• α = represents the abnormal performance of the fund
• Ri = return (net of fees) of a given portfolio or fund
• rf is the risk-free rate
• βik = sensitivity of the fund to factor k
• Fk = return of factor k for period t
This factor model is similar in its form to that used by Elton et al. (1993) to
estimate the performance of the managers’ fund.
This equation can be seen as a weak form of style analysis, which consists of
relaxing the positivity constraint and the portfolio constraint, and including a
constant term in the regression.
From a practical point of view, this approach also allows us to consider the
question of benchmarking and performance measurement in a unified manner:
once the appropriate indices have been selected, they can be used both for
return-based style analysis (strong form of the style analysis with the positivity
constraint and portfolio constraint) and for abnormal return measurement (weak
form applied to the excess returns).
2.3 Implementation in the ratings
2.3.1. The analysis categories
In William Sharpe’s model, all asset classes are represented. In certain market
conditions this can lead to statistical adjustments that no longer describe the
fund’s real allocation. Experience shows that return-based style analysis and
portfolio decomposition-based style analysis work best when they are associated.
To be more precise, we use a decomposition-based method for the observable
attributes, for example the geographical zone. We further use a return-based
method for the non-observable characteristics, in particular the investment
styles, as opposed to the investment styles that the managers themselves claim
to use, and for which information may be missing or uncertain.
In order to limit the risks of collinearity between the indices, the fund universe
has been divided into 21 distinct categories using the groupings in the
EuroPerformance classification. There are 13 of these categories for equities, 5
for bonds and 2 for diversified funds. A table in appendix 1 describes the relation
between the EuroPerformance categories and the analysis categories.
it f i i ( t f ) i ( t f ) iK ( Kt f ) it R − r =α + β F − r + β F − r +…+ β F − r + e 1 1 2 2
2.3.2. Style indices
For each category, a set of style indices averages the returns of the stocks in
which the fund is liable to invest for a given management objective. For the
indices, to comply with the issue of differences in valuation days between funds,
EuroPerformance calculates weekly performance for each day of the week. In
that way, each fund is compared to a set of indices that value at the same date
as the fund. In the case of international management, the international indices,
which are often expressed in USD, have been converted into euros on the basis
of exchange rates provided by the AFG (the French Investment Management
Association), before calculating their returns.
For the equity class, the MSCI indices are still the obligatory reference point for
nearly all funds on the international level. They are thus, de facto, the best
proxies for the markets that they evaluate and represent. For the indices that
cover the bond class, there is no consensus. Merrill Lynch was selected from
among the providers, because it offers the broadest panel of indices, and it is
one of the major references in the fixed-income world.
Appendix 4 describes the indices chosen for each analysis category.
2.3.3 Sector funds
Growth-Value and Large Cap–Small Cap style indexes are unable to fully relate
the changes in returns for certain sectors to specific characteristics. The style on
its own is insufficient in explaining the performance of certain sectors, whereas
the objective of Style Rating consists in extracting the alpha of the fund’s
performance, by taking out the performance accounted for by factors common to
the investment universe.
For sector funds we use a group of indexes covering all sectors to analyse these
funds.
This approach is consistent with the objective of Style Rating: we impose the
minimum of preconditions on the content of the fund. This choice is also justified
by the fact that sector funds are very often invested in several sectors.
Furthermore whilst there is no ambiguity as to the notion of country; this is not
the case for sector segmentation.
2.3.4 Selecting regressors
The constrained regression of the style analysis allows for the selection of the
“right” regressors which are used in the second regression to determine the
fund’s alpha.
To limit the effects of collinearity between the indices, regressors with a low or
null weight in the style analysis regression are removed from the alpha
calculation model. In more precise terms, the regressors are classified according
to their weight. The regression for the alpha calculation uses the first three at
least. The following regressors are only integrated into the model if their weight
is above the threshold of 10% in the fixed-income class and 3% for the
diversified funds.
2.3.5 The calculation period
Style Analysis and alpha are calculated on a weekly basis over the last 156 returns (3
years). Alphas are annualized by multipliying the weekly alpha by 52.
2.3.6 – Management of fund valuation
It is important in the assessment of different ratios and style valuations to use
indexes having the same valuation method as the fund being studied (indexes
valued at opening prices for funds valued on opening and indexes valued at
closing prices for funds valued on closing). As the style indexes for funds based
on opening rates are not available, funds valued at the start of the day are
analysed using closing prices from the previous day.
One opts for an analysis based on the previous day:
- if the coefficient of determination (R²) for style analysis in the regression
model calculated on a given day and for the previous day are higher than the
threshold for the category (70% or 60%) and there is a more than 15% variance
between the two coefficients in favour of the previous day.
- if the calculated R² on the given day is lower than the threshold and the
calculated R² on the previous day is higher than the threshold for the category
plus 5% (75% or 65%).
In the Gaussian framework the VaR may be calculated explicitly by using the
following formula:
( )
(μ σ )
α
c VAR W z
P dW VAR
= −
≤ =
where Zc = number of the standard deviation at (1-α)
σ = annual standard deviation
W = current value of the portfolio
α = VaR threshold (typically 1%)
The analytical side of this standard VaR formula was then adjusted by using the
Cornish-Fisher (1937) expansion as follows:
2 3 (2 3 5 ) 2
36
( 3 ) 1
24
( 1) 1
6
z Z 1 Z S Z Z K Z Z S c c c c c c = − − + − − −
where Zc = the critical value for the probability (1-α)
S = the skewness
K = the excess kurtosis (i.e., kurtosis minus 3)
The adjusted VaR is therefore equal to: VAR =W(μ − zσ )
Most often the VaR is a negative number. By convention it is displayed as an
opposite value representing the loss: a VaR displayed as 10% represents a
maximum weekly loss of 10%.
3 MEASURING PERSISTENCE
The ratings are used by investors to make investment decisions. From this point
of view, a fund’s capacity to reproduce its performance is very important for the
person who uses the ratings.
We are very often tempted to believe that a manager who has performed well
one year has a better chance of performing well the following year than one
whose result was mediocre. The fund rankings published by the financial press
are based on this idea. However, the results of the studies that tend to verify this
assumption are contradictory and do not allow one to affirm that past
performance is a good indicator of future performance.
The EUROPERFORMANCE-EDHEC ratings do not focus on the repetition of past
performance but on the manager’s capacity to deliver alphas frequently over the
analysis period (gain frequency), while ensuring that the outperformance is
regular, without too much volatility (Hurst exponent).
3.1 The gain frequency
The gain frequency measures, as a percentage, the number of times that the
fund has delivered performance that was better than that of its benchmark.
3.2 The Hurst exponent
The basic idea is relatively simple: if the funds’ excess return compared to the
customised benchmark follows a random walk, which means that there is no
correlation between the excess returns observed at date t and date t+1, this
leads one to believe that the portfolio manager has no persistence in obtaining
returns. A high return is followed by a low or high return with a probability equal
to ½. If, on the contrary, there is persistence in the series of returns, one can
expect to find positive correlation between the returns at date t and the returns
at date t+1. In that case, the process followed is no longer a random walk.
The H exponent is in fact constructed precisely as a measure of deviation
compared to the random walk, which corresponds to the basic case of an
exponent equal to ½. If the H exponent is greater than ½, we speak in that case
of persistence and positive correlation in the return series. If the H exponent is
lower than ½, we speak in that case of anti-persistence and negative correlation
in the return series.
3.3 Calculating the Hurst exponent
We first calculate the normalised series of outperformances (ERt) where we
subtract the average at each observation.
We thus define:
ERt =Rt −Bt
Z(t)=ERt −m,
where Rt : fund’s returns
Bt : benchmark’s returns
Σ=
=
T
t
t R
T
m
1
1
is the average of the returns in the sample
We then define:
( ) ( ) Σ=
=
t
s
Y t Z s
0
Which gives Y1 = Z1, Y2 = Z1 + Z2, Y3 = Z1 + Z2 + Z3, …, YT = Z1 + Z2 + …+ ZT
with:
Y Y(t)
<t<T
=
1 0max
Y Y(t)
<t<T
=
2 0min
Finally, the Hurst exponent is given by the following formula:
( )



 −
=
σ
ln 1 2
ln
1 Y Y
T
H
where
( ) Σ=
−
−
=
T
t
t R m
T 1
2
1
σ 1
is the unbiased estimator of the dispersion of
returns in the sample.
If the H exponent is greater than ½, we speak in that case of persistence and
positive correlation in the return series. If the H exponent is lower than ½, we
speak of anti-persistence and negative correlation in the return series.
3.4 Implementation within the framework of the ratings
At the date of the ratings, the gain frequency and the Hurst exponent are
calculated for the previous 156 weeks.
Two types of behaviour are identified in terms of persistence:
1. The funds whose H value is > 0.5, which tend to be persistent in their
performance
2. The funds whose H value is ≤ 0.5, which tend to be anti-persistent in their
performance
In the same way, two types of behaviour are identified for the gain frequency:
1. The funds whose gain frequency is > 0.5, which tend to be persistent in
their alphas
2. The funds whose gain frequency is ≤ 0.5, which tend to be anti-persistent
in their alphas
4 MEASURING EXTREME RISKS
More often than not, the notion of risk is reduced to the variability of the returns.
The most common indicator is the volatility. While this measure characterises the
average risk of the portfolio (average uncertainty in the distribution of returns), a
complementary approach consists of focusing on the fund’s potential for extreme
loss. The Value-At-Risk (VaR) responds to this need. The VaR allows all the risks
of a portfolio split between several asset classes to be reduced to a single value.
4.1 Value-At-Risk
There are essentially three methods for calculating a portfolio’s VaR: the
parametric method, the historical method and the Monte Carlo method.
The parametric method is often based on normal return assumptions. However,
the historical return distributions of market variables are often fairly far removed
from a normal distribution. In particular, one observes distributions that are
skewed and present fat tails. It is obviously possible to postulate a distribution
law that accounts for the existence of distribution tails that are fat compared to
those of the normal distribution (Student’s law, Pareto’s law, etc.), but the
requirement to impose a prior choice leads to a strong specification risk called
“model risk”.
The historical methods, on the other hand, do not set a prior assumption on the
shape of the return distribution and therefore only take into account the data in
the sample to calculate the VaR. In this case, the quality of the historical series is
important. There is a significant sample risk and it would require only a small
amount of missing or aberrant data to render the result obtained irrelevant. Like
for the parametric method, this method can be implemented rapidly.
The Monte Carlo method uses the same principle as the historical VaR but the
data used is obtained through stochastic simulation. This method is cumbersome
to implement because it requires many simulations to be carried out to obtain a
good level of accuracy in the results, which leads to numerous calculations. It is
used primarily in estimating VaR for complex instruments (options) which
provide flows that are non-linear functions of the base assets.
4.2 Cornish-Fisher type VaR
It is possible to implement a VaR calculation method based on an “intermediary”
approach that achieves a balance between the advantages of the historical VaR
approach (no model risk) and the parametric approach (limitation of the sample
risk). The method in question is the semi-parametric VAR, in an environment of
fat distribution tails. This method for determining the maximal potential loss
initially involves calculating a VaR by using a normal distribution formula and
then correcting it through a development referred to as Cornish-Fisher to take
into account the skewness and the kurtosis (third and fourth order moments of
the return distribution).
Within the Gaussian framework, the VaR can be calculated explicitly by using the
following formula:
( )
0.5
1
VAR n Wdt
P dW VAR
σ
α
=
≤ − = −
where n = number of standard deviations at (1-α)
σ = annual standard deviation
W = the portfolio’s current value
dt = fraction of year
The analytical side of this normal VaR formula is then adjusted by using the
Cornish-Fisher (1937) extension as follows:
2 3 (2 3 5 ) 2
36
( 3 ) 1
24
( 1) 1
6
z Z 1 Z S Z Z K Z Z S c c c c c c = + − + − − −
where Zc = the critical value of the probability (1-α)
S = the skewness
K = the excess kurtosis (i.e., kurtosis minus 3)
The adjusted VaR is therefore equal to: VAR =W(μ − zσ )
4.3 Implementation within the framework of the ratings
For each fund, a VaR is calculated using a Cornish-Fisher type semi-parametric
approach, with a threshold of 1%.
5 CRITERIA AND RULES FOR ATTRIBUTING GRADES
5.1 Criteria
The ratings are calculated from the following weekly data:
o The average alpha of the last 4 weeks
o The gain frequency for the last 156 weeks
o The Hurst exponent calculated from the last 156 weekly returns
o The positioning of the Value at Risk of the week in the VaR
distribution within each analysis category
o The average of the last 4 determination coefficients of the
unconstrained regression.
Information from outside the analysis, namely the average management fees
practiced for each of the rating categories, is also necessary to calculate the
grade. This information is obtained through a half-yearly survey of management
companies carried out by EuroPerformance.
Grade Attribution Table
Grade Population Diminishing Alpha
Gain
Frequency
Hurst
Exponent
***** H >=0 >= .5 >= .5
***** >=0 >= .5
**** >= 0 < .5
*** >=-avge(mgt fees)
** 50% + < avge(mgt fees)
* 50% - < avge(mgt fees)
VaR too high (V)
VaR too high : Funds whose Value at Risk is greater than the
average of the VaR in the analysis category + 2 standard
deviations are identified and do not receive any star.
R² too low (R² )
R² to low : In order to be graded, the coefficent of
determination resulting from the style analysis must be higher
than 70% exept for diversified funds and funds belonging to the
international categories whose level is fixed to 60%.
Funds do not receive a grade if:
o the Var is greater than the average + twice the standard deviation
in its analysis category
o the determination coefficient (R²) resulting from the style analysis is
lower than 70%
2 stars and 1 star:
groups together the funds that do not achieve the performance of
their reconstituted benchmark (alpha < 0). This population is
ranked according to the value of the alpha. The stronger 50% have
2 stars, the others 1 star.
3 stars: the funds are very close to their management objective
(reconstituted benchmark). By taking into account the average
management fees of the analysis category, they conserve a slight
advantage.
4 stars: the funds in this class have an alpha that is strictly positive but they
have not reached the requisite levels on the persistence criteria
(gain frequency and Hurst coefficient).
5 stars: this category contains funds with a strictly positive alpha and a gain
frequency that is greater than or equal to 50%.
5 star H: a sub-category within the 5-star category distinguishes the funds
with a Hurst exponent greater than 0.5.
5.2 The management rules
The ratings are calculated monthly on the basis of weekly data. The calculations
are closed off on the 2nd Friday of each month and the results are published on
the 4th Friday of the month. This time period is required for verifications and for
correcting errors in the calculation of the returns in the database.
In the event of an error, the corrections are made in the database but the
regressions and indicators are not recalculated. The funds concerned conserve
their rating from the previous month for one month. The corrections will
potentially modify the fund’s grade for the following months.
The rate of change remains low from one month to the next, confirming the
relative stability in management behaviour. It is not therefore necessary to
implement management rules on conserving grades in order to reduce the
variability of the ratings.
5.3 The ratings website
The results of the ratings are published on the 4th Friday of the month on the
following website: www.stylerating.com
Which funds are moving up? Which funds are moving down? What are the new
entries? Why is such and such a fund no longer rated? Those are the kinds of
questions that the website will reply to.
It is easy to launch queries on the:
• New funds rated
• Upgraded funds
• Downgraded funds
• Funds with a too high VaR
• Pending rating funds
The list of funds excluded from the rating is available each month with the
reason for the exclusion.
• Funds whose R² is too low
• Funds whose VaR is too high
The persistence criteria are monitored within the 5-star category.
All of the documentation is accessible on the website.
APPENDICES
1. COUNTRIES COVERED
The rating is calculated on a pan-European basis; the grade does not depend
on either the country in which the fund is marketed or the classifications.
Progressively, all the funds monitored in the EuroPerformance database will be
integrated.
The marketing zones covered are as follows: France, Luxembourg, Switzerland,
Spain, Ireland, Italy, Belgium and the Netherlands.
The goal is to cover the whole of the Euro zone.
2. REGRESSORS BY ANALYSIS CATEGORY
The style analysis and the rating use the market style indices that are the most
widely followed by European managers.
Equities South Africa
MSCI SOUTH AFRICA VALUE
MSCI SOUTH AFRICA GROWTH
Equities Germany
MSCI GERMANY VALUE
MSCI GERMANY GROWTH
MSCI GERMANY SMALL CAP
Equities America
MSCI ALL COUNTRIES AMERICAS FREE VALUE
MSCI ALL COUNTRIES AMERICAS FREE GROWTH
MSCI NORTH AMERICA SMALL CAP
Equities North America
MSCI USA VALUE
MSCI USA GROWTH
MSCI USA SMALL CAP
Equities Asia
MSCI ALL COUNTRIES ASIA PACIFIC FREE VALUE
MSCI ALL COUNTRIES ASIA PACIFIC FREE GROWTH
MSCI PACIFIC SMALL CAP
Equities Asia Ex Japan
MSCI ALL COUNTRIES ASIA PACIFIC FREE EX JAPAN VALUE
MSCI ALL COUNTRIES ASIA PACIFIC FREE EX JAPAN GROWTH
MSCI PACIFIC EX JAPAN SMALL CAP
Equities Australia
MSCI AUSTRALIA VALUE
MSCI AUSTRALIA GROWTH
MSCI AUSTRALIA SMALL CAP
Equities Austria
MSCI AUSTRIA VALUE
MSCI AUSTRIA GROWTH
MSCI AUSTRIA SMALL CAP
Equities Belgium
MSCI BELGIUM VALUE
MSCI BELGIUM GROWTH
MSCI BELGIUM SMALL CAP
Equities Canada
MSCI CANADA VALUE
MSCI CANADA GROWTH
MSCI CANADA SMALL CAP
Equities China
MSCI CHINA FREE VALUE
MSCI CHINA FREE GROWTH
Equities Korea
MSCI KOREA VALUE
MSCI KOREA GROWTH
Equities Denmark
MSCI DENMARK VALUE
MSCI DENMARK GROWTH
MSCI DENMARK SMALL CAP
Equities Emerging Countries
MSCI EMERGING MARKETS FREE VALUE
MSCI EMERGING MARKETS FREE GROWTH
Equities Spain
MSCI SPAIN VALUE
MSCI SPAIN GROWTH
MSCI SPAIN SMALL CAP
Equities Europe
MSCI EUROPE VALUE
MSCI EUROPE GROWTH
MSCI EUROPE SMALL CAP
Equities Europe Ex UK
MSCI EUROPE EX UK VALUE
MSCI EUROPE EX UK GROWTH
MSCI EUROPE EX UK SMALL CAP
Equities Finland
MSCI FINLAND VALUE
MSCI FINLAND GROWTH
MSCI FINLAND SMALL CAP
Equities France
MSCI FRANCE VALUE
MSCI FRANCE GROWTH
MSCI FRANCE SMALL CAP
Equities Hong Kong
MSCI HONG KONG VALUE
MSCI HONG KONG GROWTH
MSCI HONG KONG SMALL CAP
Equities India
MSCI INDIA VALUE
MSCI INDIA GROWTH
Equities Ireland
MSCI IRELAND VALUE
MSCI IRELAND GROWTH
MSCI IRELAND SMALL CAP
Equities Israel
MSCI ISRAEL VALUE
MSCI ISRAEL GROWTH
Equities Italy
MSCI ITALY VALUE
MSCI ITALY GROWTH
MSCI ITALY SMALL CAP
Equities Japan
MSCI JAPAN VALUE
MSCI JAPAN GROWTH
MSCI JAPAN SMALL CAP
Equities Malaysia
MSCI MALAYSIA FREE VALUE
MSCI MALAYSIA FREE GROWTH
Equities Morocco
MSCI MOROCCO VALUE
MSCI MOROCCO GROWTH
Equities Norway
MSCI NORWAY VALUE
MSCI NORWAY GROWTH
MSCI NORWAY SMALL CAP
Equities Netherlands
MSCI NETHERLANDS VALUE
MSCI NETHERLANDS GROWTH
MSCI NETHERLANDS SMALL CAP
Equities Portugal
MSCI PORTUGAL VALUE
MSCI PORTUGAL GROWTH
MSCI PORTUGAL SMALL CAP
Equities United Kingdom
MSCI UNITED KINGDOM VALUE
MSCI UNITED KINGDOM GROWTH
MSCI UNITED KINGDOM SMALL CAP
Equities Russia
MSCI RUSSIA VALUE
MSCI RUSSIA GROWTH
Equities Scandinavia
MSCI NORDIC COUNTRIES VALUE
MSCI NORDIC COUNTRIES GROWTH
MSCI NORDIC COUNTRIES SMALL CAP
Equities Singapore
MSCI SINGAPORE VALUE
MSCI SINGAPORE GROWTH
MSCI SINGAPOUR SMALL CAP
Equities Switzerland
MSCI SWITZERLAND VALUE
MSCI SWITZERLAND GROWTH
MSCI SWITZERLAND SMALL CAP
Equities Sweden
MSCI SWEDEN VALUE
MSCI SWEDEN GROWTH
MSCI SWEDEN SMALL CAP
Equities Taiwan
MSCI TAIWAN VALUE
MSCI TAIWAN GROWTH
Equities Thailand
MSCI THAILAND FREE VALUE
MSCI THAILAND FR GROWTH
Equities Turkey
MSCI TURKEY VALUE
MSCI TURKEY GROWTH
Equities International
MSCI WORLD INDEX VALUE
MSCI WORLD INDEX GROWTH
MSCI WORLD INDEX SMALL CAP
Equities Euro
MSCI EMU VALUE
MSCI EMU GROWTH
MSCI EMU SMALL CAP
Equities Latin America
MSCI EMERGING MARKETS FR LATIN AMERICA VALUE
MSCI EMERGING MARKETS FR LATIN AMERICA GROWTH
Equities Emerging Europe
MSCI TURKEY
MSCI RUSSIA
MSCI CZECH REPUBLIC
MSCI POLAND
MSCI HUNGARY
Balanced Euro Zone
MSCI EMU VALUE
MSCI EMU GROWTH
MSCI EMU SMALL CAP
EUROMTS GLOBAL
MERRILL LYNCH EURO HIGH YIELD
EURIBOR 3 MOIS
Balanced Europe
MSCI EUROPE VALUE
MSCI EUROPE GROWTH
MSCI EUROPE SMALL CAP
MERRILL LYNCH PAN-EUROPEAN BROAD MARKET INDEX
MERRILL LYNCH EUROPEAN CURRENCY HIGH YIELD
REFERENCE EURIBOR 3 MOIS
Balanced international
MSCI WORLD INDEX VALUE
MSCI WORLD INDEX GROWTH
MSCI WORLD INDEX SMALL CAP
JP MORGAN GLOBAL GOVERNMENT BONDS TOTAL RETURN USD-TRADED
MERRILL LYNCH GLOBAL HIGH YIELD
MERRILL LYNCH US TREASURY BILLS 0-3M
Bonds GBP
MERRILL LYNCH UK GILTS 1-3Y
MERRILL LYNCH UK GILTS 10+Y
MERRILL LYNCH STERLING HIGH YIELD
MERRILL LYNCH GBP LIBOR 3M CONSTANT MATURITY
Bonds Swiss Franc
MERRILL LYNCH CH GOVERNMENT 1-3
MERRILL LYNCH CH GOVERNMENT 10+
MERRILL LYNCH EUROPEAN CURRENCY HIGH YIELD
MERRILL LYNCH CHF LIBOR 3M CONSTANT MATURITY
Bonds Europe
MERRILL LYNCH PANEUR GOVERNMT 1-3 Y
MERRILL LYNCH PANEUR GOVERNMT 10+Y
MERRILL LYNCH EUROPEAN CURRENCY HIGH YIELD
EURIBOR 3 MOIS
Bonds High Yield Euro
MERRILL LYNCH EMU CORPORATE
MERRILL LYNCH EURO HIGH YIELD
EURIBOR 3 MOIS
Bonds Europe High Yield
MERRILL LYNCH PANEUR LARGE CAP CORP
MERRILL LYNCH EUROPEAN CURRENCY HIGH YIELD
EURIBOR 3 MOIS
Bonds International High Yield
MERRILL LYNCH GLOB BRD MKT CORP INDEX
MERRILL LYNCH GLOB HIGH YIELD
MERRILL LYNCH US TREASURY BILLS 0-3M
Bonds International Hedged
MERRILL LYNCH GLOBAL GOVERNMENT BOND II 1-3 Y HEDGED (vs EURO)
MERRILL LYNCH GLOBAL GOVERNMENT BOND II 10+Y HEDGED (vs EURO)
MERRILL LYNCH GLOBAL HIGH YIELD HEDGED (vs EURO)
EURIBOR 3 MOIS
Bonds International
MERRILL LYNCH GLOBAL GOVERNMENT BOND II 1-3 Y
MERRILL LYNCH GLOBAL GOVERNMENT BOND II 10+Y
MERRILL LYNCH GLOBAL HIGH YIELD
MERRILL LYNCH US TREASURY BILLS 0-3M
Bonds Euro zone
EUROMTS 1-3 YEARS
EUROMTS 10-15 YEARS
MERRILL LYNCH EURO HIGH YIELD
EURIBOR 3 MOIS
Sectors Europe
MSCI EUROPE ENERGY
MSCI EUROPE MATERIALS
MSCI EUROPE INDUSTRIALS
MSCI EUROPE CONSUMER DISCRETIONARY
MSCI EUROPE CONSUMER STAPLES
MSCI EUROPE HEALTH CARE
MSCI EUROPE FINANCIALS
MSCI EUROPE INFORMATION TECHNOLOGY
MSCI EUROPE TELECOM SERVICES
MSCI EUROPE UTILITIES
Sectors World
MSCI WORLD ENERGY
MSCI WORLD MATERIALS
MSCI WORLD INDUSTRIALS
MSCI WRLD CONSUMER DISCRETIONARY
MSCI WORLD CONSUMER STAPLES
MSCI WORLD HEALTH CARE
MSCI WORLD FINANCIALS
MSCI WORLD INFORMATION TECHNOLOGY
MSCI WORLD TELECOM SERVICES
MSCI WORLD UTILITIES
Sectors Euro Zone
MSCI EMU ENERGY
MSCI EMU MATERIALS
MSCI EMU INDUSTRIALS
MSCI EMU CONSUMER DISCRETIONARY
MSCI EMU CONSUMER STAPLES
MSCI EMU HEALTH CARE
MSCI EMU FINANCIALS
MSCI EMU INFORMATION TECHNOLOGY
MSCI EMU TELECOM SERVICES
MSCI EMU UTILITIES
3. RULES FOR INCLUSION IN AND EXCLUSION FROM THE RATINGS
The fund must:
• belong to the categories analysed (§2)
• have three years of weekly historical returns
• not have more than two returns missing for the previous 156 weeks
• the adjusted R2 of the style analysis must be greater than 70% exept for
diversified funds and funds belonging to the international categories whose
level is fixed to 60%.
• the VaR at the calculation date must be lower than the average of the VaR
in the analysis category + 2 standard deviations
4. CATEGORIES EXCLUDED FROM THE RATINGS
Funds belonging to the following categories have been excluded:
Treasury Funds
Guaranteed Funds
Gold and Raw Material Funds
Real Estate Funds
ETFs and all funds practicing index management or alternative
management.
The “Convertible Bond” category is not analysed because there is no style
index for this category.
REFERENCES
-Amenc N., Le Sourd V., “Portfolio Theory and Performance Analysis”, John Wiley, 2003.
-Chen N., Roll R., Ross S., 1986, “Economic Forces and the Stock Market”, Journal of
Business, 59, 383-403.
-Cornish and Fisher, 1937, “Moments and Cumulants in the Specification of
Distributions”, Review of the International Statistical Institute, 307-320.
-Elton E. Gruber M., Das S., Hlavka M., “Efficiency with Costly Information: a
Reinterpretation of Evidence from Managed Portfolios”, The Review of Financial Studies,
6, 1-20.
-Fama E., French K., 1992, “The Cross-Section of Expected Stock Returns”, Journal of
Finance, 442-465.
-Hurst, H.E., 1951, “Long-Term Storage of Reservoirs“, Transactions of the American
Society of Civil Engineers, 116; formalised by Mandelbrot, B., “The Fractal geometry of
Nature”, W.H. Freeman and Co., New York, NY, 1977.
-Sharpe W. F., December 1988, “Determining a Fund’s Effective Asset Mix”, Investment
Management Review, 2, 6, 59-69.
-Sharpe W. F., 1992, “Asset Allocation: Management Style and Performance
Measurement”, Journal of Portfolio Management, 18, 7-19.
-Sharpe W.F., July-August 1998, “Morningstar’s Risk-Adjusted Ratings”, Financial
Analysts Journal

The Alpha League Table score: alpha intensity

The objective of the Alpha League Table is to allow asset management firms to be ranked according to their capacity to generate alpha frequently: alpha intensity.

Alpha intensity is the product of two indicators calculated using information from the EuroPerformance-EDHEC Style Rating:

EDHEC Alternative Indexes: May 2007 (Estimates)

Asset Allocation

The authors show that the Two-moments CAPM can be extended to a Four-Moment CAPM where not only mean and variance are considered, but skewness and kurtosis. They derive a unified theory in a Four-Moment Capital Asset Pricing Model framework where variance, skewness and kurtosis receive a risk premium.

The Two-moments CAPM is linked to return normality and to a quadratic utility function. This model is not empirically verified by finance studies. During the last 20 years, models going beyond mean and variance have been developed such as the Lower CAPM , Gini CAPM and VaR CAPM . To take account of the asymmetry effect of the return distribution, Rubinstein (1973) and Krauss, Litzenberger (1976) propose an extension of the classical Two-moments CAPM. The characteristic of their model is based on a systematic skewness premium. Jurcenzko and Maillet extend the work of previous researches with kurtosis and the first four co-moments.

Like beta is related to variance, systematic skewness is related to skewness and sytematic kurtosis is related to kurtosis. The linked between variance, systematic skewness, systematic kurtosis and an asset expected return is the following:

        (1)

where

and Rm the market expected return, Sm the market skewness, Km the market kurtosis, α1 and α2 are respectively the premium required for skewness and kurtosis risk aversion. An asset with a negative contribution to portfolio skewness has a negative sytematic skewness, thus the second term in equation (1) will be positive and the E(R) will be positive. An asset with a positive contribution to portfolio kurtosis has a positive systematic kurtosis, thus the third term in equation (1) will be positive. Adding these two terms with the beta market premium gives the total asset expected return in a Four-Moment CAPM space. An asset with extreme risks (i.e. during equity/bond/currency market shocks , the asset negative returns is extemely lower with respect to thenormal case) should receive an additonal premium coming from equation (1) second and third terms.

The Four-Moment CAPM has not been empirically tested except by Hwang and Satchell (1999). Further research needs to be done to test this model on assets with high devations from normality such emerging market and hedge fund indices.

Footnotes

{1} Nantel, , Price, Price (1982)

{2} Okunev (1990)

{3} Alexander, Baptista (2000)

{4} For example a short put option or a merger arbitrage hedge fund index.

{5} For example, an emerging market index or a low liquid asset.

References

Nantel T., Price B., Price K., Variance and lower partial moments measures of sytematic risk: some analytical and empirical results, Journal of Finance, 37, p.843-855, (1982)

Okunev J., An alternative measure of mutual fund performance, Journal of business finance and accounting, 17, p.247-264, (1990)

Alexander C., Baptista A., Economic implications of using a mean-VaR model for portfolio selection: a comparison with mean-variance selection, working paper, (2000)

Hwang S., Satchell S., Modelling emerging markets risk premia using higher moments, International journal of finance and economics, 4, p.271-296, (1999)

http://www.edhec-risk.com/site_edhecrisk/public/research_news/

choice/RISKReview1063697659662713107

	ABCP Suite ABS Suite Accounting Advisory Services CDO Suite Collateralized Debt Obligations Global Transaction Services Operations and Technology Strategies Tax Advisory Services Transaction Services < Back to previous page		Securitization   Understanding that securitization is a combination of process, regulation and technology, we have assembled the wealth and depth of experience – from the beginning of the securitization process to the end – to offer you a comprehensive business solution to your securitization challenges. Securitization Accounting: The Ins and Outs of FASB 140, FIN 46R, IAS 39 and More… The seventh edition of this booklet deals with securitizations, mainly those employing term structures and traditional asset types. Learn more about this practice Contact us for more information

Deloitte

Securitization Accounting

The Ins and Outs (And Some Do’s and Don’ts) of FASB 140, FIN 46R, IAS 39 and More …

The seventh edition of this booklet deals with securitizations, mainly those employing term structures and traditional asset types. We provide reference material for all the relevant, but separate, guidance issued by the Financial Accounting Standards Board (FASB), the Emerging Issues Task Force (EITF), the Securities and Exchange Commission (SEC), the American Institute of Certified Public Accountants (AICPA) and the International Accounting Standards Board (IASB) assembled in one place. Read our “Securitization Accounting” report in the attached PDF file below.

Attachments

Securitization Accounting (3493 KB)

The Ins and Outs (And Some Do’s and Don’ts) of FASB 140, FIN 46R, IAS 39 and More… Speaking of Securitization Accounting, tax, regulatory and other developments affecting transfers and servicing of financial assets Regulation AB requires, to the extent material, disclosure of original pool summary information and static pool information. The static pool information is to be disclosed on a periodic basis, either monthly or quarterly, and to include disclosures regarding delinquencies, cumulative losses and prepayments for the respective asset type. The rules call for information, to the extent material, for a minimum of five years or for such shorter period the sponsor has been either securitizing assets of the same asset type (in the case of seasoned sponsors) or making originations or purchases of assets of the same type (in the case of unseasoned sponsors). Given the many questions regarding what sponsors will disclose, we are providing the attached examples of: 1. Original pool characteristics and static pool or performance data 2. Disclaimers used on static pool Web sites 3. Diagrams for transaction structures and flow of funds The attached survey covers disclosures from sponsors of Residential Mortgage, Commercial Mortgage, Auto Loan, Credit Card and Student Loan securitization transactions. The data were gathered from sponsors’ static pool Web sites or from preliminary prospectus supplements through January 22, 2006. Attachments SOS. Speaking of Securitization (610 KB) Reg AB: Static Pool Survey and Structural Diagrams of Flow of Funds

Subprime: Point to Where It Hurts

Steps to Modify Loans
And Avert Foreclosures
Has Investors Clashing

By LINGLING WEI, RUTH SIMON and JAMES R. HAGERTY
June 29, 2007; Page C1

As defaults on home loans mount, mortgage companies are scrambling to work out deals to help as many borrowers as possible stay in their houses.

On the surface, it seems an obvious tactic. Lenders usually end up losing money on foreclosed homes because of legal and other costs and the need to sell those properties fast, often at a knockdown price. Also, politicians are pressing mortgage companies to minimize the damages foreclosures cause to families and neighborhoods.

Still, the effort to hold down foreclosures threatens to create clashes between mortgage companies and investors in securities backed by bundles of home loans, a $6 trillion market that has been shaken recently by losses on some of the riskier types of mortgage bonds. And because of the way these securities are sold, these efforts can pit groups of holders against each other.

“It’s going to create a class warfare” among different types of investors, predicts Kyle Bass, managing partner of Hayman Capital Partners LP, a Dallas hedge fund.

Foreclosures are rising fast partly because lenders in recent years rushed to make no-money-down loans to people with weak credit records and didn’t always verify their income. At the same time, the decline in housing prices in much of the country makes it hard for borrowers who fall behind to sell their homes for enough money to pay off the loan.

When borrowers can’t keep up, lenders typically consider whether it makes sense to offer a loan modification. Such workout deals, known as “mods,” often involve lowering the interest rate or stretching out the term. Lenders have used mods for years, but the practice is expected to proliferate as defaults rise.

Sharon Greenberg, an analyst for Credit Suisse Group in New York, estimates that before this year modified loans typically accounted for less than 2% of all those outstanding. Within the next couple of years, she says, they may peak at several times that level.

Investors holding mortgage-backed bonds are watching nervously because mods may not always be in their best interest. Some investors fear that loan servicers — the firms, often owned by lenders, that collect payments and deal with defaults — will make too many mods. Generally, investors favor mods that ease a normally reliable borrower through a rough patch, but not those that merely buy time for deadbeats.

Investors doubt some homeowners merit a rescue plan. In some cases, says Kishore Yalamanchili, a fund manager with BlackRock Inc., New York, “by making these people current, you are pushing losses to another year or so.”

Credit Suisse analysts recently examined loans that had been modified over the past few years by one nationwide lender and found that borrowers missed at least one monthly payment after a mod in nearly 40% of the cases. (That failure rate may have been skewed upward by victims of Hurricane Katrina who never returned to their homes.)

If the borrower is unlikely to keep up with payments even after a mod, many investors would prefer that servicers pursue a foreclosure quickly, especially in regions where house prices are falling, reducing the value of the collateral.

Servicers are required by their contracts to act in the interests of the investors and modify loans only when that can be expected to reduce losses. That puts servicers in the tricky position of trying to figure out which borrowers are basically sound and when it makes more sense to foreclose quickly.

One complication is that different classes of investors have different interests, reflecting the complicated mechanics of mortgage securities. Issues of mortgage-backed securities are divided into slices with various ratings, depending on the level of risk. Holders of the highest-rated slices (those with the lowest risk) are first in line to collect payments of interest and principal flowing from the loans. Many such bond issues are structured so that there is initially more than enough cash flow available to cover obligations to all the investors, leaving a cushion to cover potential losses from loan defaults. If after three years or so the loans have performed well enough to meet certain performance measures, the cushion may be reduced.

In that case, some of the excess cash available goes to holders of lower-rated securities and “residuals,” the highest-risk parts of the securities that are last in line for payments.

If loan mods delay the onset of foreclosures, holders of the lower-rated securities and residuals are more likely to get those payments. But, holders of AAA and other high-rated securities may argue that the loan mods have artificially boosted the performance of the loans and that the holders of lower-rated securities and residuals are getting payments that should be preserved to protect owners of higher-rated paper against the risk of a resurgence of defaults later.

Even where there are no clashes among investors, servicers face restrictions on how they modify loans.

Moody’s Investors Service, a ratings provider, recently reviewed roughly 400 subprime mortgage-security transactions issued last year: 5% of those deals prohibit any kind of loan mod; among those that allow mods, about a third stipulate that no more than 5% of the loans backing the securities can be modified. Subprime loans are those to people with spotty credit histories.

“Those restrictions may prevent servicers from doing the things they need to do,” says Larry Litton Jr., chief executive of Litton Loan Servicing, a unit of C-BASS LLC, New York. Mr. Litton says his firm hasn’t bumped up against any ceilings. Still, he favors eliminating restrictions in future issues of mortgage securities to give servicers more flexibility.

Write to Lingling Wei at lingling.wei@dowjones.com, Ruth Simon at ruth.simon@wsj.com and James R. Hagerty at bob.hagerty@wsj.com

 http://www.reuters.com/article/bondsNews/idUSN2834578820070628

CDOs: ABS and other sundry collateral

Thu Jun 28, 2007 5:52PM EDT

June 28 (Reuters) - Collateralized debt obligations, or CDOs, are a rapidly growing class of securities created by bundling portions of other bonds into a new, more diversified structure that spreads the risk among investors around the world.

Demand for CDOs in the past few years has fueled demand for other debt such as subprime mortgage asset-backed securities.

While there are many kinds of debt used as collateral, the surge in subprime bonds as a percentage of total ABS has led to their domination in many CDOs. Investors watching the prolonged U.S. housing slump cause rising delinquencies and losses on the ABS expect CDO values and ratings will soon be affected.

Losses at two Bear Stearns Cos. (BSC.N: Quote, Profile, Research) hedge funds after bad bets on subprime came to a head in June as bank creditors tried to extricate their investments in the fund. Most of the funds’ CDO assets put up for sale never found buyers at prices acceptable to the banks, fueling speculation that ABS CDOs would suffer widespread markdowns, sources said.

High-grade CDOs buy the top, “AAA” rated portions, while so-called mezzanine CDOs are comprised of the riskier parts of underlying bonds structured to take the first losses, if any.

U.S. CDO issuance surged in 2006 to $386 billion from about $201 billion in 2005, according to JPMorgan Chase & Co. Managers have packaged $182 billion so far in 2007. (CDO volume in Europe last year was $141.2 billion. In Asia, it was $21.9 billion.) PRIMARY TYPES OF CDOs:

STRUCTURED FINANCE CDO: Includes CDOs backed by asset-backed securities supported by residential mortgages (subprime, “Alt-A” and prime), student loans, manufactured housing loans, auto and credit card receivables.

High-grade structured finance CDO issuance increased to $128.5 billion last year from $59 billion in 2005. Volume of mezzanine structured finance CDOs was $55.6 billion in 2006 and $30.7 billion in 2005. (Home-equity loans, consisting mostly of subprime mortgages, are the fastest-growing segment of ABS, rising to 26 percent this year from 14 percent in 2000.)

Issuance of high-grade and mezzanine structured finance CDOs this year is on pace to exceed 2006’s record.

C0RPORATE BOND CDO: A CDO backed by debt issued by a corporation, both junk and investment grade Investment-grade debt CDO issuance rose to $34.9 billion in 2006 from nearly $18 billion in 2005. High-yield bond CDO issuance that pioneered the CDO market in the late 1980s was $1.2 billion last year and $1.6 billion in 2005. Volume is just above $1 billion for both categories this year.

COLLATERALIZED LOAN OBLIGATIONS, or CLOs: A type of CDO packaged with slices of high-yield bank loans. Issuance rose to $106 billion in 2006 from $59.7 billion in 2005. More than $53 billion has been sold since January this year.

OTHER TYPES OF CDOs:

COMMERCIAL MORTGAGE BOND CDO: A CDO backed by commercial mortgage-backed securities, or CMBS. Volume of $28.3 billion last year was more than double that of 2005. Issuance this year has topped $12.6 billion

EMERGING MARKET BOND CDO: A CDO backed by emerging market debt. Sales of $881 million in 2006 compare with $1.6 billion in 2005.

CDO OF CDO: A “CDO squared” is collateralized with pieces of existing cash flow or synthetic CDOs. Volume of $9.6 billion in 2007 has nearly matched full-year issuance of $9.9 billion in 2006 and more than double that of 2005.

STRUCTURES:

CASH FLOW CDO: These CDOs pay investors directly from the interest and principal payments to the underlying debt. These make up the bulk of CDOs.

SYNTHETIC CDO: A structured or corporate bond CDO created with contracts referencing the value of a bond, instead of the bond itself. Synthetic CDOs sell insurance against default of the referenced collateral with credit default swaps.

MARKET VALUE CDO: A security designed to rise or fall with the value of the underlying collateral.