Trilogy of Internal rating, Credit decision and Rating migration
by Shashikala Ramachandra, former General Manager and Chief Risk Officer, Canara Bank
Giving a thought to the path leading to the Internal Rating Based (IRB) approach of BASEL III guidelines, the first step is to understand Internal rating grades and its impact on capital. The IRB document on Risk estimates of Probability of Default (PD), Loss Given Default (LGD), Exposure at Default (EAD) and Maturity (M) for arriving at the Economic capital clearly articulates the approach.
Back to the past ; At the point of time when the BASEL guidelines on standardised approach was introduced and implementation was going on in several geographies, the BCBS committee had a view that a more risk sensitive approach was required to be introduced and this would be based on the Internal rating of the borrower which was being carried out by several banks. It was observed that there were no common standards in the internal rating system of the banks.
In the Consultative paper released by BCBS in June 1999, there was a passing reference to the internal rating based approach and a task force was constituted to study the Internal rating system in the Banks. It was in May 2001, that another consultative document was released by BCBS with an explanatory note for BASEL II guidelines. The orientation of the IRB approach was inclined towards measurement of credit risk through the Rating function and the rating models for assessing the credit risks culminating in a Risk grade, a simple notation. The exercise had its genesis in the banks which were already using Risk rating models for borrower gradations way back in the 1990s. This was also being reviewed on an annual basis and by some banks at quarterly intervals too.
Hence when the PD function as a risk estimate was thought of, the key ingredient was the Risk grades and its migration over a one year horizon. The count of those that moved to default grade was the starting point for PD models. The prerequisite for Internal Rating based approach, as the phrase suggests, is the Internal Rating itself. Even when the Banks migrate to the revised BASEL III approaches for credit risk, the reference to Internal rating continues for a segment of loan book based on certain thresholds.
In order to understand the variability that can happen in rating grades it is necessary to be aware of the underlying reasons and other factors which impact the rationale. This is possible in an internal rating exercise but not in an External rating as only the definition of the rating grade and the brief rationale can be referenced.
The internal rating is carried out under 4 broad categories for segmented analysis viz, Financial Risk, Business Risk, Management Risk and Industry Risk, a mix of quantitative and qualitative parameters scored on the basis of weightages attached to the parameters and scores to the elements within the parameters. However, these inputs could differ from one bank to another due to which rating grades of the same borrower at times may differ. The External Rating is included as a component to bring in the element of ‘Through the cycle effect’ as the Internal rating model is generally considered as ‘Point in time rating’ carried out on the basis of Financial statements.
The Internal rating model may likely have more than 100 parameters collectively under each category of risk, and each parameter having multiple attributes is to be picked and scored.
The question is how significant is this rating grade as a pointer in credit decisions, and what is actually contributing to the variability in rating grades.
There is a need to understand certain inherent factors to get answers to this question. I am just listing some significant indicative impacts ;
- The model is developed and put to use based on the historical data from the rating module of the Bank, if available, or on the basis of proxy data sourced from the rating agencies. In the 1990s the rating exercise was conducted more for Issuer rating rather than an overall borrower rating.
- The rating model was initially developed based on expert judgement and hence it’s measured as compared to a statistical model
- Initially the rating exercise was being carried out by the credit sanction team which may have certain subjective bias
- The scores and weightages were based on expert judgment
Rating grade was and is, no doubt, a tool for decision making wherein the analyst working on credit sanctions will aggregate and consider several related information before allotting a risk rating grade. However, certain factors that contribute to the rating grade may have a likely impact not just on the credit decision but also on the capital to be set aside for which rating is a key ingredient:-
- The rating exercise is an annual exercise based on the Audited financial statement of the borrower entity, whereas credit decisions may be taken at any point of time. Hence the date of a credit decision is not in sync with the date of Risk rating, as a lag in time is imminent.
- Any internal rating model has a degree of judgement in picking up the attributes by the analysts and on the future projections being scored on the ability of the borrower’s enterprise to have adequate cash flows. So also certain parameters of management risk.
- Even if review of rating is conducted at quarterly intervals, it is based on limited information available and without quarterly audited financial statement
- Quarterly profit generation is not ploughed back and hence uncertain
- Any untoward event either on account of natural calamities, procurement challenges, adverse movement of cost and price, employee unrest etc shows the impact with a lag and only after the impact is computed and /or disclosed.
- Measurement of distance to default would generally be more than 12 months when credit quality could be almost on the verge of deterioration, in some cases
- Influence of borrower on the rating agency
- The scores are divided into all the four risk categories and again sub-scored into the attributes. Any lower score in some of the attributes which have the potential to identify default may not significantly affect the risk grade thereby camouflaging the actual intended outcome of the model.
Hence it is very important in credit decisions to look into the rationale very closely and also take cognizance of score variation from year on year in each category to understand the reason and its impact of credit decision and pricing. It may so happen that a borrowal account may retain the same grade over two consecutive risk ratings. Even within the same grade, direction of score also needs to be examined before taking a credit decision.
In case of external rating, the breakdown of the attributes and scores as also the variation between two ratings will not be available making it challenging to rely only on the rationale. Internal rating was mooted as a better indicator of riskiness in the consultative paper of BCBS in view of the fact that the validations can be carried out to understand the power of discrimination in the internal rating models of the bank. The results of validation also give the banks insight into the modifications or updations required to fix any shortcomings in the model. This also facilitates the Banks to understand and improve their processes and credit quality to optimally use capital for healthy business growth.
It is clear that any credit decision should not be based on a single grade, a notation which can indicate an overall view and not a granular exploration of a credit proposal.
Apart from the impact on credit decisions, the basis of internal rating is also a key starting point in the journey towards a commonly used model developed for computation of Probability of default which further leads to the Expected loss and capital computation.
Certain key assumptions and interpretations that go into the rating migration analysis as a prerequisite for the purpose of either the validation of the rating model or the PD computation.
This also brings us to reference the migration analysis which is one of the tools for validating the internal rating model and also used for the purpose of PD computation in advanced approaches.
Coming to the migration analysis; Basically the PD computation commences with the count of default borrowers and the non-default borrowers over a one year horizon.
So, a matrix of Internal rating of all corporate borrowers based on transition in ratings from one base year to the other gives the number of defaults during the review period of one year.
Eg : Movement of internal rating of a borrower from beginning of a financial year to the rating as at the end of the financial year.
Looks simple. The aim is to bring into the fore the likely assumptions and the variations that can happen in this exercise from, say, from one Bank to the other.
Illustration: A borrower is rated in an Internal rating model based on the financial statement as at the end of a Financial year 31.3.2018. The rating based on the Financial statement at the end of the subsequent year 31.3.2019 may have migrated to a worse grade or even default. It is quite likely that the borrower enjoys his limits upto the availability of the rating as at 31.3.2019 as there is bound to be a lag after the end of the financial year and upto the rating exercise allotting the rating grade.
The uniformity in approach of allotting a rating grade to a borrower entity may possibly differ from one Bank to another. This could affect the uniform approach to migration analysis among banks which is a result of certain assumptions and interpretation as also practical data issues, some of which could be stated as below:
- Review period dates may be from base year end date to the next year end date
- Review period dates may be from the date of rating of the previous year (not necessarily the end of the financial year) to the date of rating of the subsequent year
- Review period dates may be from the date of quarterly review rating based on limited review and at the end of 12 months from the last review
- Treatment of unrated borrowers which go to default
- If borrowal accounts are closed, the reason to be analysed. Reasons could be the end of tenor of loan or pre closure or shifting liability to another Bank
- Treatment of new loan accounts disbursed during the year where migration cannot be seen
The PD should ultimately become the basis for Computation of Expected loss, and eventually the required capital to be set aside. However it is important to understand the variability, the assumptions and its probable impact on the PD outcome. This is the stage when the level of conservatism comes into force to get a realistic result.
The necessity to understand the aforementioned is a pointer to be pegged as PD which is a risk estimate is also a significant contributing factor for pricing of the loan appropriately to cover the credit risk being boarded into the loan book, limit setting, stating risk appetite among several other uses in risk evaluation and mitigation techniques.
In conclusion, the internal rating is no doubt a very important tool in decision making, however the credit analyst shall carry out due diligence analysing the rating rationale as well as level and direction rating grade.
Whereas, in the task of migration analysis, the assumptions are probably unavoidable, but a degree of conservatism needs to be applied and review of ratings at shorter intervals should be put in place duly factoring external environments also to get a credible movement of rating.
(Smt. Shashikala Ramachandra, Retired General Manager and GCRO from Canara Bank. Associate partner Pegasus Institute of Excellence, Director , Kogta Finance India Ltd, Blogger on Risk and Banking related subjects (https://email@example.com))