A Credit-risk Rating Framework (CRF) is necessary to avoid the limitations associated with a simplistic and broad classification of loans/exposures into a "good" or a "bad" category. The CRF deploys a number/ alphabet/ symbol as a primary summary indicator of risks associated with a credit exposure. Such a rating framework is the basic module for developing a credit risk management system and all advanced models/approaches are based on this structure. In spite of the advancement in risk management techniques, CRF is continued to be used to a great extent. These frameworks have been primarily driven by a need to standardise and uniformly communicate the "judgement" in credit selection procedures and are not a substitute to the vast lending experience accumulated by the banks' professional staff.

Broadly, CRF can be used for the following purposes:

1. Individual credit selection, wherein either a borrower or a particular exposure/ facility is rated on the CRF

2. Pricing (credit spread) and specific features of the loan facility. This would largely constitute transaction-level analysis.

3. Portfolio-level analysis.

4. Surveillance, monitoring and internal MIS

5. Assessing the aggregate risk profile of bank/ lender. These would be relevant for portfolio-level analysis. For instance, the spread of credit exposures across various CRF categories, the mean and the standard deviation of losses occurring in each CRF category and the overall migration of exposures would highlight the aggregated credit-risk for the entire portfolio of the bank.

Basic Architecture of CRFs

The following elements outline the basic architecture and the operating principles of any CRF.

Grading system for calibration of credit risk

1. Nature of grading system

2. Number of grades used

3. Key outputs of CRF

Operating design of CRF

1. Which exposures are rated?

2. The risk rating process

3. Assigning and monitoring risk ratings

4. The mechanism of arriving at risk ratings

5. Standardisation and benchmark for risk ratings

6. Written communications and formality of procedures

CRFs and Portfolio Credit Risk

1. Portfolio surveillance and reporting

2. Adequate levels of provisioning for credit events

3. Guidelines for asset build up, aggregate profitability and pricing

4. Interaction with external credit assessment institutions

The architecture and operating principles are discussed in detail in the ensuing paragraphs.

Grading System for Calibration of Credit Risk

The grades (symbols, numbers, alphabets, descriptive terms) used in the internal credit-risk grading system should represent, without any ambiguity, the default risks associated with an exposure. The grading system should enable comparisons of risks for purposes of analysis and top management decision-making. It should also reflect regulatory requirements of the supervisor on asset classification (e.g. the RBI asset classification). It is anticipated that, over a period of time, the process of risk identification and risk assessment will be further refined. The grading system should, therefore, be flexible and should accommodate the refinements in risk categorisation.

Nature of Grading System for the CRF

The grading system adopted in a CRF could be an alphabetic or numeric or an alpha-numeric scale. Since rating agencies follow a particular scale (AAA, AA+, BBB etc.), it would be prudent to adopt a different rating scale to avoid confusion in internal communications. Besides, adoption of a different rating scale would permit comparable benchmarking between the two mechanisms. Several banks utilise a numeric rating scale. The number of grades for the "acceptable" and the "unacceptable" credit risk categories would depend on the finesse of risk gradation. Normally, numeric scales developed for CRFs are such that the lower the credit-risk, the lower is the calibration on the scale.

Illustration

A rating scale could consist of 9 levels, of which levels 1 to 5 represent various grades of acceptable credit risk and levels 6 to 9 represent various grades of unacceptable credit risk associated with an exposure.

The scale, starting from "1" (which would represent lowest level credit risk and highest level of safety/ comfort) and ending at "9" (which would represent the highest level of credit risk and lowest level of safety/ comfort), could be deployed to calibrate, benchmark, compare and monitor credit risk associated with the bank's exposures and give indicative guidelines for credit risk management activities. Each bank may consider adopting suitable alphabetic prefix to their rating scales, which would make their individual ratings scale distinct and unique.

Number of Grades Used in the CRF

The number of grades used in the CRF depends on the anticipated spread in credit quality of the exposures taken by the bank. This, in turn, is dependent on the present and the future business profile of the bank and the anticipated level of specialisation/ diversification in the credit portfolio. CRFs with a large number of levels/ grades on the rating scale are, as evident, more expensive to operate as the costs of additional information for (very) fine gradation of credit-quality increase sharply. A bank can initiate the risk-grading activity on a relative smaller/ narrower scale and introduce new categories as the risk-gradation improves.

Key Outputs of the CRF

The calibration on the rating scale is expected to define the pricing and related terms and conditions for the accepted credit exposures. It is possible to define broad pricing bands and directly link the band with the calibration on the rating scale. Further refinement in the pricing proposal would be based on the bank's judgement of the prominent elements of the loan proposal and the relationship with the borrower.

In addition to the pricing related decisions, the calibration on the rating scale would allow prescription of limits on the maximum quantum of exposure permissible for any credit proposal. The quantum (or amount of facility sanctioned) would depend on the credit-score on the CRF. These limits could be linked to an internal parameter (viz. a certain percentage of bank's capital funds) or could be linked to an external parameter (viz. a certain percentage of the total debt required by the borrower). This would help in a larger dispersion of risk amongst lenders and limit risk concentration in moderate credit-quality projects.

The rating scale could also be used for deciding on the tenure of the proposed assistance.

The rating scale could also be used to decide on the frequency/ intensity of monitoring of the exposure. Banks may also use the rating scale to keep a close track of deteriorating credit quality and decide on the remedial measures which the deterioration may warrant. For instance, the frequency of surveillance on category 5 exposures could be kept at quarterly intervals, while those on category 3 loans could be half-yearly. More importantly, movement of an existing exposure to the "unacceptable" category of credit risk (grades 6 to 9) should directly identify the extent of provisioning (loan loss reserves) that need to be earmarked for expected losses.

Though loss-provisions are often specified by the regulator (e.g. the RBI provisioning norms), banks should develop their own internal norms and maintain certain level of "reasonable over-provisioning" as a best practice. Specifically, while the credit exposure/ asset is clearly facing rapid/ steady erosion and is on the downhill transition path, anticipatory provisioning can be done based on the calibrations on the risk-rating scale. These provisions could be in the nature of general provisions.

Which Exposures are Rated?

The first element of the operating design is to determine which exposures are required to be rated through the CRF. There may be a case for size-based classification of exposures and linking the risk-rating process to these size-based categories. The shortcoming of this arrangement is that though significant credit migration/deterioration/erosion occurs in the smaller sized exposures, these are not captured by the CRF. In addition, the size-criteria are also linked with the tenure-criteria for an exposure. In several instances, large-sized exposures over a short tenure may not require the extent of surveillance and credit monitoring that is required for a smaller sized long-tenure exposure. Given this apparent lack of clarity, a policy of 'all exposures are to be rated' should be followed.

The Risk-Rating Process

The credit approval process within the bank is expected to replicate the flow of analysis/ appraisal of credit-risk calibration on the CRF. As indicated above the CRF may be designed in such a way that the risk rating has certain linkages with the amount, tenure and pricing of exposure. These default linkages may be either specified upfront or may be developed with empirical details over a period of time. The risk rating assigned to each credit proposal would thus directly lead into the related decisions of acceptance (or rejection), amount, tenure and pricing of the (accepted) proposal.

For each proposal, the credit/ risk staff would assign a rating and forward the recommendation to the higher level of credit selection process. The proposed risk rating is either reaffirmed or re-calibrated at the time of final credit approval and sanction. Any revisions that may become necessary in the risk-ratings are utilised to upgrade the CRF system and the operating guidelines. In this manner, the CRF maintains its "incremental upgradation" feature and changes in the lending environment are captured by the system. The risk-rating process would be equally relevant in the credit-monitoring/ surveillance stage. All changes in the underlying credit-quality are calibrated on the risk-scale and corresponding remedial actions are initiated.

In conventional banks, the practice of segregating the "relationship management" and the "credit appraisal" functions is quite prevalent. One of the variants of this arrangement is that responsibilities for calibration on the risk-rating scale are divided between the "relationship" and the "credit" groups. All large sized exposures (above a limit) are appraised independently by the "credit" group. Generally, the activities of assigning and approving risk-ratings need to be segregated. Though the front-office or conventional relationship staff can assign the risk-ratings, the responsibilities of final approval and monitoring should be vested with a separate credit staff.