Document Sample

                   Thomas Link, Prof. Dr. Svetlozar Rachev, Stefan Trück
                             University of Karlsruhe, Germany

Abstract. In contrast to the situation in the United States, rather weak tendencies towards an
infrastructure for external ratings by independent agencies have been observed in Europe in recent
decades. Despite a change in this structure due to a new framework for bank's capital adequacy, our
paper shows that, especially for middle-class enterprises, thus far there has been no possibility of
obtaining an objective rating. The reason is, that information is very asymmetric between investors and
capital demands on small and medium-sized companies. A rating which would remove this asymmetry,
is not offered by the traditional rating agencies. In our study, not only will we examine the need for the
rating of medium-sized companies, but we will also give an overview of characteristics that have to be
incorporated in models pricing the debt of such companies. We will also show why so far the traditional
models for evaluating default risk have to be adapted to medium sized companies and further describe
models, products and companies in this sector and suggest ideas for model refinements.

Key-words: Rating, Credit Risk, Medium-Sized Companies, Risk Weights, Rating Agencies, Spread,
Reduced Form Models, Structural Models, Rating Criteria, Discrete-Time Markovian Model, Brownian
Motion, Black-Scholes

1) Introduction
1.1) Rating Culture in Germany
Compared to the situation in the United States rather weak tendencies towards an
infrastructure for external ratings by independent agencies have been observed in Germany in
the last decades. So, according to a study, published by the IWK, Munich, in January 2000
only 170 German enterprises and banks exhibited a rating by an international agency, in the
USA about 8000 companies had a rating.2 Certainly, German banks do internal ratings to
decide about whether a company fulfills the requirements to get a credit, but their outcome is
not published and often influenced by subjective goals of the respective bank. “Even the firm
being rated is typically not informed about its current internal rating”3. In contrast to those,
external ratings usually are created explicitly in order to publish them and to give a neutral
and objective evaluation of a company’s credit status and additionally should give an outlook
on its future potential and chances on the market.4
What are the reasons for this exceptional difference in the amount of rated companies? One
cause can be identified in the typical German house bank system: The bank plays the role of
an intermediary between investors and credit demand and gives bank credits to its customers

    See Becker 2000, p. 56
    Krahnen/Weber 2000, p. 4

to conditions, which are determined by their internal rating. In the Anglo-Saxon bank system,
investment banks are not allowed to issue bonds themselves and, therefore, deal with loans
thrown on the market by companies which have a demand for debt financing.5 This private
issuing makes it necessary for those companies to prove their general ability to pay to the
public, which can be achieved by a rating created by an acknowledged agency. Another reason
is that the demand for industrial loans is much smaller in Germany, because of the broad
supply of government bonds and the investors’ rather risk averse attitude. These facts lead to a
totally different debt structure in Germany compared to the USA. In Germany nearly half of
the debt market is covered by bank credits, in the USA about 80 percent of the market volume
for debt are corporate loans.6
    Recent developments on the international capital markets in the EU obviously are leading to
a change of this structure. First, the improvements in the information technology sector and
the introduction of the Euro lead to a globalization of the capital markets and bear new
possibilities for the provision of capital by issuing industrial loans. Second, the “new
framework for banks’ capital adequacy (proposed by) ... the Basle committee on banking
supervision (at July,3 1999), intends to reward their portfolios on their aggregate credit
rating”7. So, banks are induced to evaluate their credit risk by a standardized rating or use
external ratings by External Credit Assessment Institutions (ECAI), to categorize it and to
back it with equity corresponding to its risk classification.
According to the latest Consultative Document (January 2001) of the Basle Committee the
capital requirements for the banks will be higher for loans credits given to unrated companies.
This is due to the higher risk weight assigned to unrated corporations entering the capital
requirement calculation.
This makes it inevitable, for companies with a strong demand for capital to acceptable
conditions, to get rated, as otherwise banks don’t have a clue about the company’s standing
and have to conduct a time consuming internal rating process or just give them bad credit
conditions. These facts indicate that industry loans as well as ratings are gaining importance
on the European capital markets.
This trend, however, causes problems especially for middle-class enterprises, as information is
very asymmetric between investors and capital demanding small and medium-sized

  See Becker 2000, p. 73
  See Becker 2000, p. 34
  See Becker 2000, p. 36
  Shirreff 1999 (expression in brackets added by the author)

companies8 and a rating, which would remove this asymmetry, is not offered by the traditional
rating agencies as Standard & Poors, Moody’s or Fitch IBCA for this category of firms9. For
smaller companies both non-rated and not listed on the stock market the so-far most
commonly used models couldn't be applied anyway. Neither the structural approach by
Merton and its refinements - e.g. by KMV – nor the reduced form approaches using the rating
of a company as an input variable can be used for medium-sized companies since they are not
listed on the stock market and do not have a rating. We will show that, especially for middle-
class enterprises, thus far there has been no possibility of obtaining an objective rating.
Furthermore, it even might not meet their requirements, because these agencies are specialized
mainly on rating bonds and loans, while middle-class companies often are in need of equity
capital. Bigger, rated companies, e.g. such listed at the Frankfurt Stock Exchange, saturate
their demand for equity on the stock market and naturally spread more information about the
firm’s projects and its credit status, which leads to lower capital costs for them, if their
prospects are good. On the other hand, medium-sized companies which are not as well-known
get worse conditions even if their ability to pay and projects promise prosper future
developments, just because of a lack of information on the capital markets. Recognizing this
“niche for an agency rating medium-sized companies”10 several rating agencies for this
market sector were established in Germany in 1998. These were the first activities towards a
supply of rating products for medium-sized companies in the world.

1.2)    Objective of the Paper
This work shall explain why there is a need for a rating for medium-sized companies and
describe the problems one might face applying the traditional approached to evaluating Credit
Risk. We will further give a brief overview of already existing companies and products for
this sector and, furthermore, come up with necessary future steps which have to be taken.
The next chapter shows the consequences of the new Basel Capital Accord (Basel II), the
functions a rating for middle-class enterprises has to fulfill, Furthermore, advantages of the
existence of middle-class ratings and an exemplary rating process are shown.
In the third chapter we will then have a look at the traditional approaches to estimate default
and downgrade risk of companies and make suggestions how they could be applied to
Medium Sized Companied.

  See Gerke et al. 1995, p. 16
  See Becker 2000, p. 91
   Shirreff 1999

In the fourth chapter we will then introduce some of the new rating agencies, give information
about their target groups and methods. We will also point out the difference to traditional
ratings offered by the older agencies. Finally, some further theoretical approaches which
might be realized in the future are explained.
The paper ends with a short conclusion of its contents and results.

2) The Need for Rating Products and Refined Credit Risk Models for
Medium-sized Enterprises

2.1) The Effect of The New Basel Capital Accord

Due to the New Basel Capital Accord (Basel II)11 until 2004 every European Bank is obliged
to provide approaches in their Internal Rating Systems "which are more comprehensive and
sensitive to risks" (Basel II). The new framework – a revision of the 1988 Basel Accord - is
due to the problems arising from evaluating new complex instruments in the credit sector and
a change in the optimistic view on Credit Risk Modeling as a result of e.g. the Asian Crisis.
The capital adequacy can be measured by different methods including external and internal
The new framework requests banks to evaluate their credit risk by a standardized rating or use
external ratings by External Credit Assessment Institutions (ECAI). Further the banks are
obliged to categorize the risk and to back it with equity corresponding to its risk classification.
Table 1 shows that according to the latest Consultative Document (January 2001) of the Basle
Committee the capital requirements for the banks will be higher for loans credits given to
unrated companies than for companies with a rating better than BB-. This is due to the higher
risk weight assigned to unrated corporations entering the capital requirement calculation.

                (Corporate)     AAA      to A+ to   A- BBB+   to Below
                Credit          AA-                    BB-       BB-       Unrated
                Risk Weights    20%        50%          50%     150%       100%*

 Figure 1: Risk Weights for Corporates According to the Consultative Document of the Basel
                         Committee of Banking Supervision (January 2001)

     Basel Committee on Banking Supervision "The new Basel Capital Accord" (issue January 2001)

Therefore, especially smaller banks dealing with medium sized companies will have to revise
their rating system and may be forced to provide higher capital requirements to meet the
obligations. The problem is not only that many smaller banks so far didn't have an adequate
internal rating system but that at the current state of discussion the revised accord may even
not allow for the most commonly used models like KMV or S&P's CreditMetrics. For smaller
companies both non-rated and not listed on the stock market the so-far most commonly used
models cannot be applied. Neither the structural approach by Merton and its refinements - e.g.
by KMV – nor the reduced form approaches using the rating of a company as an input
variable can be used for medium-sized companies since they are not listed on the stock market
and do not have a rating. We will explain this more thoroughly in the following sections.
The consequence is that the banks will either have to implement a new internal rating system
to obtain ratings also for medium sized companies or get these ratings from ECAI (External
Credit Assessment Institutions).
Also for companies with demand for capital to acceptable conditions, it is important to get
rated, as otherwise banks due to higher capital requirements give them bad credit conditions.
Since not all banks will have an appropriate rating systems or the companies may not be
satisfied with their rating given by the housebank there will me more and more need for
independent external ratings from ECAIs.

2.2) The Advantages of Ratings for Medium-sized Enterprises

One positive effect of a rating for medium-sized companies certainly is an improvement of the
chances of middle-class firms to economically succeed, which recently has been an issue in
Germany. This category of companies has been underrepresented and the bankruptcy rate
among them too high. A rating in a standardized rating scale (e.g. Moody’s or S&P’s scale)
enables an investor without any knowledge of the company and its products to estimate the
firms standing and the risk associated with an investment. So, by disclosing the international
capital markets for them and, thereby, providing them with better credit conditions and better
chances on the stock markets, such a rating system can lead to an improvement of the
economical environment for small innovative firms and, in consequence, new industry
branches can arise. This is necessary, in order to stay a highly developed country and not to
miss the worldwide ongoing economical and industrial change.

Besides this financial aspect, the entrepreneur can extract even more advantageous effects out
of a rating. So, he can use it as a means of advertising: The media get information about his
company’s success and, thereby, obtain a more positive picture of its products, which is traded
on to the customers as well as to suppliers, who are interested in having a successful company
as a customer and, therefore, offer better conditions, as for instance discounts. This
automatism also works in the field of recruitment; a firm with a rating, that promises a prosper
future outlook, will easily attract well-qualified personnel. Another aspect is sales financing.
If marketers have a good opinion of the company’s products and their chances on the market,
they probably will think about supporting the firm in the marketing sector (see figure 2, p.8).
Another advantage obviously is that investors can choose from a bigger pool of loans and
equity titles. The current state in Germany just gives them the opportunity to carry their
money to the bank or buy assets of the already listed corporations. If owners of middle-class
companies recognize, that by using rating means, they can sell their enterprise to a fair price
by going public, more and more of them will take this way of capital funding into account and
the number of joint-stock companies will rise.
A side effect, the existence of a rating culture in the medium-sized sector can have, is an
enhanced motivation and greater discipline of the business management of companies, which
need a rating. Usually a company is given a period of time after it has ordered the rating, in
which it will try to improve its current status to achieve a better rating result. The enterprise
has – induced by the rating – an incentive to use their potentials more effectively.12 Moreover,
provided that the criterions to be rated are known to the companies, SMEs have an orientation
for their own risk management.13 A better resource allocation is the consequence, which – out
of the sight of national economics – leads to an increase of welfare.
Banks can take advantage of the existence of a medium-sized company rating as well. They
can “calibrate (their) ... internal rating systems by comparison with public ratings”14 and
hence, when having ratings of middle-class firms, can use their slightly varied internal rating
process for a large new group of enterprises.
The last point to be mentioned here is the effect of an improved protection of investors. A
rating of a company is available for everybody in newspapers or online and free of charge.15 If
there are formally acknowledged rating agencies, or – even better – if there are agencies,

   See Becker 2000, p. 74
   See RS Rating Services AG brochure 01/ 2000, p.5
   Shirreff 1999 (expression in brackets added by the author)
   See Becker 2000, p. 73

which can refer to a successful “back-testing”16 of their ratings, which means that their
predications about their customers proved to be true in most of the cases, they have additional
information about the opportunities on the capital markets and can rely on this information.
This is currently only accomplished in the case of large-scale enterprises.

3) Application of the Major Approaches in Credit Risk to Medium Sized

In this chapter we will describe the two major approaches to evaluate credit risk – the
structural and reduced form models and show why without a rating for medium sized
companies it will be very difficult to apply any of them.

3.1) Structural Models – The Approach of Merton and the KMV company
Technically, Merton’s work is a so-called structural model, as it also includes the value of the
firm’s assets V(t) in valuing the default-risky bond. The probability of default is determined
by the volatility of the assets. Default occurs if the value of the firm hits a certain lower

     Krahnen/ Weber 2000, p.12

boundary of the obligations at maturity. The stochastic process driving the dynamics of the
assets is assumed to be
                                  d Vt    Vt dt   Vt d Wt 

where  and  are the firm’s asset value drift rate and volatility, and dW(t) is a Wiener
process17. The figure of such a path of the value of a companies assets is shown in the figure
The fundamental statement is that a firm defaults, if the values of its assets fall below its
outstanding debt. This means that at maturity of the debt18, the bondholder is paid the face
value of the bond, if the market value of the firm exceeds the face value of the bond. If the
assets fall below this amount, the payment to the bondholder will be the residual value of the
firm. Therefore the bondholder gets back the smaller of two quantities: either the face value of
the bond or the firm’s assets. This payoff BT  at maturity T specifically amounts to the face

value of the bond B minus the price of a put option PVt , t  on the firm’s value with a
strike price equal to the face value of the bond.19:

                                     BT   B  PVt , t 

                                     BT   B  maxB  Vt ,0

            Figure 3: P. Crosbie (1999). Modeling Default Risk, KMV Corporation.

Assuming the term structure of the riskless interest rate r to be constant, the value of the risky
zero bond at time t maturing at T (t<T) is equal to

   also called ‘Brownian Motion’. A Wiener Process is a continuous-time random walk pn(t) of
statistically independent, normally distributed increments.
   a zero-coupon bond
   In comparison to that, the payoff to the shareholders at maturity, amounts to either the difference
between the assets and the debt or zero in case of default. This means they hold a call option
C Vt , t  on the assets with strike price equal to the face value of the firm’s debt, i.e. the bond.
                                     Bt , T   Be  r T t   PV t , t 

The price of the call option C Vt , t  can be derived using the Black-Scholes formula:

                      C Vt , t   Vt  d1   B e  r T t   d 2 

                                Vt         1 2
                           log           r   T  t 
                      d1       B            2 
                                          T t

                                Vt         1 2
                           log           r   T  t 
                      d2       B            2 
                                                              d1   T  t
                                          T t

Using the put-call parity

                               CVt , t   B e  r T t   PVt , t   Vt 

we calculate the value of a risky zero bond for the time t maturing at T

                             Bt , T   V (t )(1  (d1 ))  Be  r (T t ) (d 2 )

Despite its simplicity and intuitive appeal, Merton’s model has many limitations. Especially
considering the often quite sparse information about the value of a medium sized company we
may run into difficulties:
To determine the value of a bond of a company we need the value of the firm's assets and also
the volatility of the value. However, we can hardly observe the volatility of the market value
of a company. But if the company is listed on the stock market we can use the following
relationship that can be derived by Ito's Lemma

                                        stocks            N (d 1) market

to estimate the volatility of the value of a companies assets. KMV uses exactly this
relationship in their famous Credit Monitor to estimate the expected default frequencies
But for companies not listed on the stock market it will be very difficult to estimate the
volatility of the assets of the company. Since this paper focuses on evaluating default risk of
Medium Sized companies we face the following problems:

           Medium Sized Enterprises are not listed on the stock market and even information
            about the value of the companies assets is rather sparse. Thus, it is obvious that the
            method described above can hardly be used to rate the companies credit quality.

           Even if it possible to determine the value of a company at a certain point in time t
            the question rises whether the value of a medium sized company really follows a
            Brownian Motion as it is assumed in the models. The value of the assets of smaller
            companies may be much more volatile and have a greater volatility than those of
            large companies. Maybe it will be necessary to apply another stochastic process
            (e.g. stable Brownian Motion21 that also allows for fat tails and asymmetry in the
            return distribution of a companies assets) to describe the process of value changes
            of such enterprises.

Due to the difficulties the Structural Approach may give us we will now examine whether the
other major approach to evaluating default risk – the so-called reduced form approach is more
appropriate to estimate the default risk of medium sized companies.

3.2) The Reduced Form Approach

3.2.1) Default Models

  see P. Crosbie (1999). Modelling Default Risk, KMV Corporation
  see forthcoming working paper Menn, Rachev, Trueck (2001), Refinement of a Structural Model for
Estimating Default Risk of Medium Sized Companies

In 1994, Jerome S. Fons22 developed a reduced form model to derive credit spreads using
historical default rates and a recovery rate estimate. Fons’ approach is based on the results of
Moody’s corporate bond default studies, which at that time covered 473 defaults of issuers
that ever held a Moody’s corporate bond rating between January 1, 1970 and December 31,
1993. He found out that the term structure of credit risk, i.e. the behavior of credit spreads as
maturity varies, seems to depend on the issuer’s credit quality, i.e. its rating. For bonds rated
investment grade, the term structures of credit risk have an upward sloping structure. The
spread between the promised yield-to-maturity of a defaultable bond and a default-free bond
of the same maturity widens as the maturity increases. On the other hand, speculative grade
rated bonds behave in the opposite way: the term structures of the credit risk have a downward
sloping structure.
Fons’ findings are equivalent to the crisis at maturity hypothesis23, which assumes that highly
leveraged firms with near term debt face a great uncertainty with respect to their ability to
meet their obligations. Speculative grade rated firms, once past these obstacles and having
survived without a default, face a lower risk of default for time horizons of five years or more.
Well-established, large and solid investment grade firms, on the other hand, face a low default
risk on the near term, while their credit outlook over longer time horizons, such as 10 or more
years is less certain.
In every rating category, Fons compares term structures of credit spreads with weighted-
average marginal default rates, using data from Moody’s investigations.
In his model, Fons assumes that investors are risk neutral. The risky bond price B(0, T ) with
face value B maturing at time T supplied by Fons can be used to infer the credit spread on
that bond by means of a formula which links the price of the bond to its yield to maturity.
The price of a risky bond in t  0 can be expressed in terms of its yield, with r being the
riskless yield and s being the credit spread:
                                           B(0, T )  Be  ( r  s )T
whereas the price of a riskless security is
                                            B' (0, T )  Be  rT

We denote d R (t ) as the probability of default in year t after the bond was assigned rating R ,

given that the bond has not defaulted before that date. Seen from date t  0 , S R (t ) is the

     Vice President of Moody’s Investors Service
     see Johnson (1967)

survival probability at date t . In the event of default the investor receives a fraction  of par,

the recovery rate. S R (t ) is given by

                                                  1  d
                                                  j 1
                                                               R   ( j)

whereas the probability that the bond rated R will default in year t is given by

                                            DR (t )  S R (t  1)d R (t )

The expected value of the random flow X t received in t is such that

                                      E ( X t )  S R (t  1)d R (t ) B' (0, t )

The price of zero-coupon bond with initial rating R maturing at T is then the
sum of the expected returns in each year

                        T                                     T
         BR (0, T )   E ( X t )  S R (T ) B' (0, T )   S R (t  1)d R (t ) Be  rt  S R (T ) Be  rT
                       t 1                                   t 1

with this formula we can compute the spread s of the risky zero bond:

                                   1 T                                               
                              s    ln  S R (t  1)d R (t ) e r (t T )  S R (T )
                                   T  t 1                                           

Fons determines the term structure of credit risk by calculating the spreads for zero bonds of
every maturity T.
Fons’ model requires the value of the recovery rate of a bond, which does not depend on the
initial rating, but on its seniority and the bankruptcy laws of the issuer‘s home country.24
The figure above show the term structures of credit spreads calculated by Fons’ model, using
historical probabilities of Moody’s default database, assuming a fixed recovery rate of 48.38%
of par. Since in this model so far only the event of default is considered it belongs to the so-
called "default models".

                                    Risk-neutral credit spreads

             0,100%                                                                 A
             0,080%                                                                 Aa
             0,060%                                                                 Aaa










        Figure 4: Risk Neutral Credit Spreads, Fons(1994). Financial Analysts Journal

Although his approach is rather simple in relation to other theoretical models of credit risk,
the predicted credit spreads derived by Fons show strong similarity towards recent observed
market data.

3.2.2) Discrete-Time Markovian Model
However, not only the "worst case" event of default has influence on the price of a bond, but
also a change in the rating of a company or an issued bond. Therefore, Jarrow, Lando,
Turnbull (JLT)25 introduced a Discrete-Time Markovian Model to estimate changes in the
price of loans and bonds.
Consider a probability space (,F,P) with  being the set of all credit events during the
lifetime of a bond, i.e.:

    rating upgrade
    stable rating
    rating downgrade (with ‘default’ as a special event)
The distribution for the default time is modeled via a discrete time, time-homogeneous
Markov chain on a finite state space S   ,... k  . The state space S represents the possible
credit classes, with 1 as the highest (e.g. Aaa in Moody’s rankings) and K-1 being the lowest
(e.g. C). The last state, K, represents bankruptcy.

  see Moody’s approach of calculating recovery rates
  Jarrow, Robert A., Lando, David and Turnbull, Stuart M. (1997). A Markov Model for the Term
Structure of Credit Risk Spreads

The Markov chain X  X (t ) 0t T defined on the state space S ( X t :   S ) is

specified by a K  K transition matrix

                                       p11                p12         p1K 
                                                                                
                                       p 21               p 22       p2 K 
                                    P                                 
                                                                                
                                       p K 1,1         p K 1, 2    p K 1, K 
                                       0                                 1 
                                                           0        0           
where p ij  0 for all i, j , i  j and      p
                                             j 1
                                                    ij    1.

The (i,j)th entry p ij represents the actual probability of going from state i to j in one time step,
whereas pii  1   pij  i  S .
                    j 1
                    j i

In order to use Markov chains for modeling the rating history of a bond, we have to make
several assumptions:
1. The probability for a credit rating changing from category i to category j is the same for all
    i-rated companies.
2. Markov property: A rating change from time t to t+1 is independent of rating changes
    prior to t
    PrX t 1  it 1 X t  it   Pr( X t 1  it 1 X t  it ,..., X 0  i0 )  it  S, t  N0

3. time homogeneity: The probability to change from i to j during t , t  1 is the same for
    every t  N0
    PrX t 1  j X t  i   PrX t m1  j X t m  j  m  N0

Let p ij (n) denote the n-step transition probability of going from state i to state j in n  N time

steps. Using the Markov property, we can calculate the n-step transition matrix P(n) by
computing the n-fold product of P
                                                         P ( n)  P n
    Rating to:               Aaa       Aa           A         Baa        Ba       B    Caa-C    Default     WR

                 Aaa       95.41%    2.75%    0.00%        0.00%      0.00%    0.00%    0.00%    0.00%    1.83%
                 Aa         1.99%   88.05%    5.38%        0.00%      0.00%    0.00%    0.00%    0.00%    4.58%
    Rating       A          0.00%    2.28%   89.02%        5.87%      0.11%    0.33%    0.00%    0.00%    2.39%
    from:        Baa        0.12%    0.36%    4.79%       86.11%      5.27%    1.08%    0.00%    0.00%    2.16%
                 Ba         0.00%    0.14%    0,29%        7.07%     74.46%    9,38%    0.87%    1.01%    6.78%
                 B          0.00%    0.00%    0.23%        0.35%      3,95%   77.24%    6.50%    5.46%    6.27%
                 Caa-C      0.00%    0.00%    0.00%        0.00%      0.00%    5.26%   70.33%   18.66%    5.74%

                    Source: Special Comment (2000), Moody’s Investors Service

      The Table above e.g. shows the All-Corporate Rating Transition Matrix in 1999 published
      by Moodys.
      With some further refinements – e.g. the introduction of so-called more realistic and
      empirical observable risk-adjusted pseudo-probabilities q ij (t , t  1) 26 for a movement from

      one state to another produced by a measure Q that is the unique equivalent martingale
      measure, which makes the bond prices martingales27 - JLT's approach then is able to
      determine the compute the probability of default occurring after date T, i.e. the survival
      probability until T, of a bond rated i in t:
                                                     K 1
                                    Qti (  T )   qij (t , T )  1  qik (t , T )
                                                     j 1

      Finally, the value of a risky zero-coupon bond issued by a company rated i at time t can be
      derived using the expectation under the equivalent martingale measure Q :

                                   Bi (t , T )  B ' (t , T )   (1   )Qti (  T )   
      whereas the recovery rate  is taken to be an exogenously given constant and B' (t , T ) to
      be the value of a default free zero bond.

Some versions of this model are applied in Risk Software, for example in J.P. Morgans
CreditMetrics™. Obviously, this technique could be applied also for Medium-Sized
Companies if we have information about historical default rates and transition matrices.
But still we face some problems that make it difficult to apply this model to estimating the
Credit Spreads of Medium Sized Companies:

         Empirical studies show that Credit Spread is not only default and downgrade risk but
          also due to the lack of the markets liquidity. The question rises how important for

     for further description see Jarrow, Robert A., Lando, David and Turnbull, Stuart M. (1997).
     A martingale is a zero-drift stochastic process with the following form:
                                                   dX  dZ
where dZ is a Wiener Process and  is a variable that may itself be stochastic.
A martingale has the convenient property that its expected value at any future time is equal to its value
today, i.e. E ( X t )  X 0 , t  N0

          example the fact is that credits or loans of medium sized companies will hardly be
          traded on financial markets.
         Due to lack of historical data for default rates/transition matrices – classical reduced
          form models cannot be applied yet. For the next years it will be very important to find
          methods that can deal with only sparse information about ratings and transition
          matrices but still provide reliable estimates for default and transition probabilities.
         Rating procedures are developed – but so far we do not have any information about the
          goodness of ratings of so-called ECAI for smaller companies.
Therefore, in the next chapter we will take a brief look at the recently appearing rating
companies for medium-sized enterprises and describe their methods and the current situation
in this branch in Germany.

4) The Rating Process – Requirements and Structure

In the last two chapters we showed that both for medium sized companies and for banks trying
to model the default risk of such companies there is need for appropriate ratings. In this
chapter we will come up with functional requirements of such ratings and further give a brief
overview about the most important rating agencies in Germany and some of their methods.

4.1) Rating Agencies in Germany

In the last three years Rating Agencies with special focus on medium sized companies. In this
section we will give a very brief overview of these companies.

         URA Unternehmens Ratingagentur AG (1998): URA was the first rating agency
          explicitly specialized on SMEs. It was founded in July 1998 in Munich on initiative of
          the BBW (Bayerisches Bildungswerk: an association in charge of education in
          accordance with the existing social and economic system) and a private auditing
          company. In the middle of 1999 the BBW drew back from the project.The agency’s
          target group are firms showing a turnover of more than 10 million DM. The price span
          of the rating products offered by URA ranges from 18 to 36 thousand DM.28

         R@S Rating Services AG (1999): After the BBW had drawn back from URA, it
          founded the R@S -also located in Munich, in summer 1999. They define their main

     See Becker 2000: p. 92

           target group by a minimum turnover of seven million DM and demand – dependent on
           the customers’ size and turnover – a price spectrum between 15 and 40 thousand DM.
          Euro Rating AG (1999, founded by a former Moody‘s analyst): Founded by a
           former Moody’s analyst in summer 1999, EuroRatings is based in Frankfurt. In
           comparison to the above introduced agencies, their rating idea rather resembles the
           philosophy of the international agencies: A rating is the estimation of a company’s
           future ability to honor their interest and redemption obligations.29 This different
           approach also can be seen, when comparing their prices.

Since 1999 some more agencies with the same focus were founded since 1999, among them
              Much-Net (1999)
              GDUR (2000)
Besides these professional companies a friendly society Rating Cert e.V., a friendly society –
tries to set standards for Rating culture and deals with the identification of the principles of a
company rating, with concepts for an analyst’s qualification, general rating methods, as well
as the certification of agencies and analysts.

4.2) Functional Requirements
A rating product targeting small and medium-sized companies (SMEs) as customers has to
fulfill different functions than the traditional products do. The big international agencies on
the sector of industrial enterprises so far restrict themselves mostly to the evaluation of exactly
defined loans. Certainly, this rating also takes into consideration the firm’s overall credit
status, but usually the company itself is not the object of the rating and the future outlook and
ongoing projects are not noticed by it. As SMEs are not in the situation in which they easily
can acquire equity, but on the other hand are in need of capital to finance investments to
enlarge their business to a bigger extent than big firms, qualitative aspects regarding the
company’s future developments should be considered in order to draw conclusions on the
firm’s ability to pay and to find the correct rating. Additionally, to improve the enterprise’s
chances on the equity market not only the credit standing should be evaluated, but also a
chance and risk contemplation should influence the result of the rating.30

     See EuroRatings brochure: „Verbesserte Markttransparenz“, p. 13
     See Becker 2000, p. 90f

As German entrepreneurs proverbially are frightened of imparting intimate information about
their company, ratings do not enjoy a big acceptance among medium-sized companies. In
order to make leaders of small and medium-sized companies understand, that it is necessary to
remove the informational asymmetry between them and the capital markets, incentives to use
this chance to acquire capital have to be created. One aspect is to demand a for the target
group bearable price. The big agencies take a relatively high price, which is eventually not
affordable for smaller companies or at least they estimate the benefit lower than the costs.
Therefore the new established rating agencies URA Unternehmens Ratingagentur AG, R@S
Rating Services AG and EuroRatings AG decided on lower entry prices (figure 1).
Another aspect is an agency’s independence of a bank’s capital market strategy. As mentioned
above, results of internal ratings by banks are often object to the bank’s intention and,
therefore, not useful for an objective judgement about a firm’s actual situation. In addition, if
an enterprise intends to get a bank credit and is rated by the bank’s own agency, it might hold
back negative information in order to get better credit conditions. Those factors decrease the
rating’s quality.31 On the other hand, results of private rating agencies have to be comparable
and as a consequence should agree on using the same scale and — if not the same
standardized methodology, so at least systems with high conformity. Furthermore, to achieve
the market’s acceptance the continuity of the rating system and the permanence and good
reputation of the agencies have to be guaranteed.32 Without these requirements, a company
cannot be sure if a rating will be still valid in the future so that it is not willing to spend
money on it.


                                                          23223            23434


An important influence on the latter criterions for a successful introduction of a SME rating is
                                       If one agency sets the center of gravity on the shareholder
the equality of the evaluation target. EuroRatings
        U.R.A.           R@S                          Fitch IBCA        Moody's         S & P's
                          Figure 5: Entry Prices for Ratings in Euros (Date 01/2000)
                                                                    Source: aims on 01/2000, p.
value, another one focusses on the value of the debt and again another one IWK-Studie the future79

     See Wagner 1991, p. 118
     See Wagner 1991, p. 122

outlook, despite congruent rating scales there is no sence in a comparison of the results.33 So
there is a demand for transparency of the methodology of the rating products.
In Germany a friendly society, RatingCert e.V., tries to cope with that task and wants to
develop common quality standards for a rating of medium-sized enterprises and make them
public around Europe. RatingCert e.V. sees its main proposition in identifying and securing
minimal requirements to credit agencies in Europe without hindering the competition among
A further important point is sufficient qualification of an agency’s analysts. In order to
evaluate the future prospects and current position on a certain market, the employee of a rating
agency must at least have insights into the current technical status quo of the respective
branch. Especially smaller companies, as e.g. component suppliers for bigger firms, compete
on markets, where high technical knowledge is necessary to differentiate the products of the
manufacturers. So, to give a proper and realistic valuation of a firm’s standing on its market,
in addition to economically educated analysts, there is a need for specialists having a more
technical background of the different industry branches in a rating agency. This requirement
seems to be quite difficult to achieve, as the SME segment is very inhomogeneous35 and a
very broad range of specialists for lots of different fields is needed. Thus, we will now have a
brief look at the Eligibility Criteria for an ECAI according to Basel II. There are six criteria

         Objectivity: "The methodology for assigning credit assessments must be rigorous,
          systematic, and subject to some form of validation based on historical experience."
         Independence: An ECAI has to be independent, the assessment process should be as
          free as possible and there should be no political or economic pressure
         International access/ Transparency: Since the assessment should be available also
          to foreign institutions, the rating and the methodology should be published in an
          international accessible form.
         Disclosure: : "An ECAI should disclose qualitative and quantitative information
          about their rating methodology and information as set forth below" to avoid a so-called
          "assessment shopping" for institutions which give more favourable ratings.

   See Becker 2000, p. 90
   See RS Rating Services AG brochure 01/2000, p.12

        Resources: The ECAI should have sufficient resources to carry out appropriate credit
        Credibility: This criterion is very important to prevent the misuse of confidential
         information. In order to be eligible for recognition, according to Basel II, an ECAI
         does not have to assess firms in more than one country.

4.3) An Exemplary Rating Process
Although it was stated above that the rating procedure of the traditional agencies has to be
adapted to the requirements of middle-class enterprises, the rough structure will be similar.
Therefore, this structure shall be introduced in the following (see figure, p. 23).
The process starts with a company’s request of a rating being handed in to the respective
agency. In a first meeting the enterprise is informed about which information is needed to
generate the rating. While the company is collecting this information, the agency appoints a
team of analysts.36
First, by taking into account the history as well as the social, political and legal situation in the
country the enterprise is located at, the so-called country risk is determined.37 The next step is
the analysis of branch specific characteristics and competitive analysis. This “industry
outlook”38 is evaluated by contemplation of indicators like market growth, profitability,
market concentration39, dependency of the level of economic activity, eventual barriers to
entry or technological changes40. Then the company’s market position is determined and a
quantitative analysis, including a contemplation of key figures as the return on investment or
the capital ratio, as well as a qualitative analysis, taking into account organizational factors
like business management, future strategies or financial flexibility, follow.22
By an aggregation procedure, a result is extracted and passed to the rating committee
consisting of a heterogeneous group of experienced analysts.41 This rating committee
discusses the outcome of the rating and decides whether there are any tasks to be redone or

   See Becker 2000, p. 62f
   See Becker 2000, p. 64
   See Wagner 1991, p. 93
   See Wagner 1991, p. 152
   See Becker 2000, p. 65
   See Becker 2000: p. 66

the rating can be accepted. If it is not content with the results, the quantitative and qualitative
analysis is repeated. Usually country and branch risk is not calculated for every single rating,
but regularly and then dropped into the rating. The by the committee acknowledged rating is
then passed on to the requesting company which again has the possibility to accept it or not. If
not, new meetings are scheduled in which the business management can hand in new or
additional internal information, resulting in a qualitative reanalysis. Finally, having accepted
the rating, the company now can opt if the rating is published or not. Eventually, the company
is observed over a period of time after the first rating and, if necessary, an upgrade or a
downgrade may be executed.

4.4) The Selection and Aggregation of Criterions

A crucial point, where a middle-class rating has to differ from the traditional ratings for large-
scale enterprises, is the choice of criterions which are investigated and the weights, they are
awarded with. Obviously, there are different aspects which influence the status of a medium-
sized company more or less than criterions one would concentrate on, when rating a larger
firm; e.g. as a matter of fact and as a consequence of the rather bad standing middle-class
companies have on the equity market (compared to stock-joint companies), they have a much
smaller capital ratio. So a comparison between a larger and a medium-sized company by
measuring the capital ratio with the same weight might be somewhat unfair. So, a specific
weighting system has to be developed by middle-class rating agencies.
But this is only one point. The other and even more important change, which has to be made,
compared to traditional rating schemes, is that the selection of criterions itself, which serve as
the basis of the rating, has to be adapted. It is self-evident, that there are specific aspects
which are typical for medium-sized companies and not observable in large-scale enterprises
and vice versa. An example: In a private owned middle-class company the entrepreneur has to
cope with the problem of appointing an adequate successor or substitute, when he plans to
back out or is temporarily indisposed. Such an event might lead to a totally different business
strategy, but does not occur in that form in a stock-joint company, which typically is a larger
enterprise. Firms listed on the stock market are monitored by their supervisory boards
guaranteeing a certain grade of strategic persistence. So, the eventual contractual successor
and substitute arrangements must be taken into account.42 On the other hand, a medium-sized
firm might have e.g. a less elaborated budgeting system, resulting in some key indicators not
being available.43 Differing from traditional rating systems, which are capital-market-oriented,
a SME rating should evaluate a well-balanced mixture of past as well as future-oriented
aspects, seeking for a rating system, which should be more stakeholder-focussed. This often is
a problem which has to be solved separately for every rating request, but with the goal of
standardizing the process as far as possible, it stays an issue.
A formal conception of a system of criterions and their aggregation to a final rating result was
developed by Wagner [1990] (see figure 7, p. 24). He calls certain aspects, such as finance
and accounting or business and personnel management, components, each of which forms an
integral part of the rating.44 This system of components is individually filled with criterions

   See Wagner 1991, p. 174
   See Wagner 1991, p. 150
   See Wagner 1991, p. 155

for each rating process, depending on the type of company and the provided information.
These criterions are assigned to the respective component.
Taking into consideration possible autocorrelations, a quite sophisticated aggregation concept
and a linear regression model then is applied to obtain an adequate rating.
However, it seems that so far in practice most methods rather work with more qualitative than
quantitative analysis and rather simple methods to obtain a rating. Figure 9 e.g. shows the
fields (corresponding to the components in Wagner’s terminology), URA investigates and
calculates values for, by using it’s own “analysis tool”. Those fields are measured by
investigating sub-criterions, which in the case of field two “human resources” – as an example
– are “person–nel planning”, “workforce” and “human resources development and policy”.
Despite there seems to be quite a good balance between the purely capital-market- and past-
oriented field “corporate finance”, and the other at least partially qualitative and future-
oriented components, the rating methodology has still space for improvements.

                                        Factor                        Weight
                            Management and organization                 0.2
                                  Human resources                       0.15
                                  Corporate finance                     0.4
                                Products and markets                    0.15
                      Production- and Information-technology:           0.05
                           Facility location(s) and ecology             0.05

              Figure 9: Risk Factors and their weights according to URA Rating
For refinements it seems to be necessary also to include more statistical and mathematical
methods like e.g.:
      Multiple Regression Models
      Multivariate Techniques (Factor, Correspondence and Discriminant Analysis)
      Neural Networks
      Fuzzy Logic
      Expert Systems
or hybrid models using several of these techniques at once. But to use such more quantitative
techniques there is need of reliable numbers from the companies and also historical data. As
long as medium sized companies provide rather sparse data it will be very difficult to use
these methods adequately for rating procedures.

5.) Conclusion
In the last chapter we saw that the repeatedly mentioned rating culture is currently developing.
On last year’s October,14, ELSA, an internet access provider and supplier for computer
graphic solutions, listed on the German Neuer Markt, published their BBB+ rating, which was
furnished by URA. On November, 29 ELSA published their year-end results, which
confirmed the positive rating, and ELSA was able to resist the general downward trend in the
Neuer Markt.45
This example could eventually be a first break through concerning the establishment of an
investor-related as well as an enterprise-oriented rating market. What still is missing is the
general acceptance on both sides, as a survey in Becker’s IWK-Studie 01/2000 shows. This,
however, is only natural as the market does not have evidence, that the applied rating
procedures are valid, and their results correspond to reality, but with several examples like
ELSA this problem should be solvable.
We further showed that despite an important orientation on the classical rating system for
large companies (e.g. done by S&P, Moody‘s, KMV etc.), the models for estimating default
risk of medium sized enterprises have to be adjusted. Especially due to probably higher
volatility of a medium sized companies assets and the lack of data for default rates and
transition matrices – the classical structural and reduced form models cannot be applied yet.
Also there is no information about historical default rates for companies with various ratings,
so still we do not know enough about the goodness of ratings for medium sized companies in
general nor for certain rating companies, respectively.
We also described the need for more technical methods, like neural networks and fuzzy logic
techniques, in order to create general accepted standardized tools, whose validity can be
proven. It is not enough, to just come to a rating result by a discussion on a round table of
experts, to convince entrepreneurs and investors of the reliability of the rating result. But – as
stated in chapter three – these efforts are already taken. Additionally, the Basel Committee as
well as Rating Cert e.V. take care of eventual necessary certification methods, or – in case
they decide on the free market system securing the quality of the offered ratings – of a
framework of principles to be fulfilled by a rating agency they support. So we can conclude
that there is still need for improvement but the first steps for adequate rating procedures and
modeling default risk also of medium sized companies are definitely already taken.

     See URA-Newsletter 12/2000, p. 1/2

6) References

Anders, Ulrich (1997): Statistische neuronale Netze, Verlag Franz Vahlen GmbH, Munich

Baetge, Jörg, Interview (1997): “Nur ein erster Schritt”. In: Manager Magazin 09/1997

Becker, Helmut (2000): Rating in Deutschland, eine kritische Marktuntersuchung, IWK-
Studie 01/2000. Antonym-Verlag, Munich 2000

Berblinger, Jürgen (1996): Ansätze der internationalen Ratingagenturen. In: Handbuch
Rating, published by Büschgen/ Everling, Gabler Verlag, Wiesbaden, p.21-110

Campbell, John Y., Lo, Andrew W. and MacKinlay, A. Craig (1998). The
Econometrics of Financial Markets. Princeton University Press, Princeton, New

Cantor, Richard, Packer, Frank and Cole, Kevin (1997). Split ratings and the pricing
of credit risk. Research Paper, Federal Reserve Bank of New York.

Caouette, John B., Altman, Edward I. and Narayanan, Paul (1998). Managing Credit
Risk, John Wiley & Sons, Inc.

Carty, Lea V. (1997). Moody’s Rating Migration and Credit Quality Correlation,
1920-1996. Moody’s Investors Service

Carty, Lea V. and Fons, Jerome S. (1993): Measuring Changes in Corporate Credit
Quality, Moody’s Investors Service

Crosby, Peter J. (1999). Modelling Default Risk, KMV Corporation.

Duffee, Gregory R. (1999). Estimating the Price of Default Risk. The Review of
Financial Studies, Vol. 12, No. 1, pp. 197-226. Federal Reserve Board.

Duffie, Darrel and Singleton, Kenneth J. (1999). Modeling Term Structures of
Defaultable Bonds. The Review of Financial Studies Special, Vol. 12, No. 4, pp. 687-

Fons, Jerome S. (1994). Using Default Rates to Model the Term Structure of Credit
Risk. Financial Analysts Journal, 25-32

Gauthier, C., de la Noue, P. and Rouzeau, E. (1998). Analyzing Corporate Credit
Risk: A Quantitative Approach. Quants No. 29.

Gerke, Wolfgang et al. (1995): Probleme deutscher mittelständischer Unternehmen
beim Zugang zum Kapitalmarkt. Nomos Verlagsgesellschaft, Baden-Baden 1995.

Gupton, Greg M., Finger, Christopher C. and Bhatia, Mickey (1997). CreditMetrics –
Technical Document, J.P. Morgan.

Jarrow, Robert A., Lando, David and Turnbull, Stuart M. (1997). A Markov Model
for the Term Structure of Credit Risk Spreads. The Review of Financial Studies, Vol.

10, No. 2, pp. 481-523.

Kealhofer, Kwok and Weng (1998). Uses and Abuses of Bond Default Rates, KMV

Keenan, Sean C. (2000). Moody’s Special Comment: Historical Default Rates of
Corporate Bond Issuers, 1920-1999. Moody’s Investors Service

Kijima, Masaake (1998). Monotonicities in a Markov Chain Model for valuing
Corporate Bonds subject to Credit Risk. Mathematical Finance, Vol. 8, No. 3, 229-

Knoch, Martin and Rachev, Svetlozar (1999). Credit Risk: Recent Advances, Part 1.
Chair of Econometrics and Statistics, Department of Economics, University of

Knoch, Martin and Rachev, Svetlozar (1999). Credit Risk: Recent Advances, Part 2.
Chair of Econometrics and Statistics, Department of Economics, University of

Krahnen, Jan Pieter and Weber, Martin (2000): Generally Accepted Rating
Principles: A Primer. Published by the Center for Financial Studies under:

Lando, David (1998). On Cox processes and credit risky securities. Department of
Operations Research, University of Copenhagen

Loges, Hans J. (2000): Rating, Working Paper No.1. Published by Hans J. Loges

Molé, Matthew C. (1994). Re: Concept Release NRSROs, Moody’s Investors

R@S Rating Services AG (2000): Brochure: Rating für den Mittelstand 01/2000.
Published under:

Samson, Solomon B. (2000). Corporate Ratings Criteria, Standard & Poor’s.

Vasicek, Oldrich Alfons (1984). Credit Valuation, KMV Corporation.

Wagner, Wolf-Christof (1991): Rating mittelständischer Unternehmungen. Verlag Peter Lang
GmbH, Frankfurt am Main 1991, 04/2001, 04/2001, 04/2001, 04/2001, 04/2001, 04/2001, 04/2001


Shared By:
fanzhongqing fanzhongqing http://