Weights assigned to Rating Factors

[Angshuman]

Hi David,
Rating agencies assigns weights to different rating factors that are considered for ratings. My question is on what basis they assigns weights to different rating factors? Can you please throw light on the same?

Regards,

 

[mdecav]

Since I work for one of the agencies, can you give more detail as to your question? For more transparency, for some lines of business there has been a trend toward a more "scorecard" type of analysis. While the 'scorecard' is a more transparent way of showing the quantitative analysis/aspects of a credit, there are still qualitative factors that could override the scorecard.

Hope this helps.

 

[Angshuman]

Hi,
For example, in case of methodology adopted by Moody's in 'Independent Exploration and Production Industries' ,one of the several factors is 'Reserve and Production Characteristics' which is given a weight of 40% .My question is, is it based on some study or it is judgemental?

Second Question is in case of 'Equity Investments' what is the methodology that is followd for grading such investments? Is it same as the 'Rating' methodology? If you provide any doc/refer any book that will be of great help.

Regards,
Angshuman

 

[mdecav]

Most likely the percentage is judgmental.

Ratings agencies do not rate equity investments, although I know that some equity funds in Latin America are rated based on performance measures (Sharpe and information ratio), but this is mainly due to the regulatory environment there.

The promise of interest payments and the return of principal of a fixed income security is the underlying criteria by which the agencies rate; the forward-looking view of probability of default on that promise.

 

[mlp12]

A rating/scoring model can be developed based on,

a) Statistical Analysis (if sufficient data, especially with regard to defaults, is available), or,

b) Expert Judgement (if sufficient data, especially with regard to defaults, is not available).

It is also pertinent to mention here that a rating/scoring model developed on (a) or (b) is to be validated on regular basis to fine tune the parameters and also their weightages.

mlp12

 

[Scorer]

It's logical to assume that they process all data they have to find out statistics-based scorecards and after this if the quality of this rating model (scorecard) is not good enough (that is usually the truth) they accompany this model with judgmental factors. Again, I'm sure they verify judgmental factors with simulations. Some data cannot be easily or unambiguously quantified or data set can contain opposite patterns, in this case judgmental in inevitable. But I think they try as hard as possible to avoid judgmental that means "without solid proof".