Abank s risk management system is in jeopardy when its

Size: px

Start display at page:

Download "Abank s risk management system is in jeopardy when its"

Ferdinand Christian Elliott
5 years ago
Views:

1 COMMUNITY BANKING Risk Ratings Revisited by John E. McKinley Abank s risk management system is in jeopardy when its risk-rating system is substandard. Citing data culled from Beating the Odds... A Community Banker s Guide to Risk Management, the author contends that many banks risk-rating systems need some improvement. Both a model portfolio risk-rating system and a loan decision matrix are presented for banks to consider in upgrading their own risk-rating process. In 1994, RMA published Credit Risk-Rating System, a risk-rating guidance based on industry best practices. The publication raised banker awareness of the need for effective risk-rating systems and provided a uniform rating approach for banks of all sizes. However, as former RMA Chair Paul Dorfman suggested in the August 1998 issue of The Journal of Lending & Credit Risk Management, there is still much room for improvement in rating systems. In addition, there is room to improve their usage in risk management. Many banks currently are tapping into the potential offered by more sophisticated rating approaches. However, as discovered in the recently released Beating the Odds...A Community Banker s Guide to Risk Management, the industry is not taking full advantage of rating systems. Some of the shortcomings that limit the effectiveness of risk ratings include: Insufficient risk-rating categories to effectively differentiate risk. All pass credits regarded as sound, therefore, not requiring differentiation. Inadequte rating discipline. Borrowers are rated, rather than transactions. Insufficient linkage of risk-rating and loan approval processes. These statements are generalizations and may apply fully, partially, or not at all in individual 1998 by RMA. McKinley, who in 1996 retired as principal operating officer and chief credit policy officer of Bank South Corporation, is president of McKinley Consulting, Atlanta, Georgia. 36 The Journal of Lending & Credit Risk Management November 1998

2 THE WATCH RATING CATEGORY CONTAINS COMPANIES WITH DETERIO- RATING FINANCIAL TRENDS OR OTHER NEGATIVE FACTORS THAT RAISE CON- CERNS ABOUT THE COMPANY S FUTURE. THIS FORCES A DISCIPLINE THAT REQUIRES UP OR OUT DECI- SIONS WHILE THE GET OUT OPTION STILL EXISTS. institutions. However, they are seen often enough to warrant examination. Insufficient risk-rating categories to effectively differentiate risk. There has been a slow evolution from an approach of one or a few pass categories to a more advanced approach of four to six pass categories. The current community bank average is 3.59 pass categories. The key to effective differentiation is the number of categories between the average rating and the OAEM rating. At least three ratings, including a true watch rating, should precede OAEM. The existing OAEM definition is too broad and, therefore, does not provide an effective early warning signal. Assuming that the #3 rating represents average risk, the ratings between 3 and OAEM would be as follows: 4 Acceptable risk. 5 Marginally acceptable risk. 6 Watch. These categories are chosen because of their importance as predictors of future portfolio migration and, therefore, credit volatility. The number of categories needed above the average risk rating depends on the individual bank s needs. A large bank with major-sized companies in the portfolio would require more categories than a small bank with a portfolio consisting entirely of small to medium-sized companies. A large portion of most small banks commercial portfolios falls within the acceptable risk category and include sound borrowers with a financial weakness(es) that excludes them from an average rating. The marginally acceptable category includes start-ups, other highly leveraged borrowers, or those with generally insufficient financial strength to qualify for a higher rating. Within the watch rating are companies with deteriorating financial trends or other negative factors that raise concerns about the company s future. This category forces a discipline that requires up or out decisions while the get out option still exists and reflects the notion that migration of problems must be stemmed early, to avoid costly and time consuming work-out. These six categories offer a basic approach to risk ratings for those banks that have not yet expanded their number of pass categories. Even smaller banks would have five or six pass and four criticized rating categories. The usual reaction from bankers is that this is too complicated and confusing. Certainly, it requires more training and discipline than having just one or two pass categories. The benefit, however, is greater clarity for lenders and risk managers, better loan decision-making, and improved portfolio management. Later in this article, a refinement of this system is introduced that provides even more risk differentiation with significantly more rating categories. Rather than increasing complexity, however, it actually simplifies the rating process and is easier for lenders to understand and to use in making loan decisions. All pass credits regarded as sound. Seventy-eight percent of community bankers surveyed said that no further differentiation of pass credits was needed. This logic overlooks the increasing likelihood of downward migration for weaker pass credits. These marginally sound assets will migrate faster, farther, and in greater numbers than their stronger brethren in the higher quality categories. In fact, poor pass credit risk distribution is likely to be the best predictor of potential problems in the next credit down cycle. High levels of competitive pressure coupled with a general indifference to how pass credits are distributed within the portfolio lead to banks aggregating large concentrations of credit in the marginal pass categories. The solution is to create a risk-rating system that provides management with the ability to determine and manage risk distribution within the pass categories. Prevention of problem loans results from an effective pass loan management mentality. With a rating system that has adequate rating categories, a model portfolio can be created with target ranges for the acceptable levels of outstandings within rating categories (see Figure 1). Devising such a model requires careful thought because it must be designed to reflect the bank s risk tolerance. 37

3 Once in place, the model provides a benchmark to alert management when danger levels are being approached for various rating categories. For example, when categories 5 and 6 in Figure 1 reach 13%, this should trigger a review to determine where the growth is coming from. Could it be new loans or migration from higher quality rating categories? Once the cause is determined, appropriate strategies can be devised. The objective of the model portfolio is to ensure that the risk distribution of the bank is consistent with its risk tolerance. The process of determining the appropriate distribution is a risk management discipline that will benefit any bank since it requires consideration of how risk distribution correlates with total portfolio risk. Without a tool such as this, banks might fail to realize that portfolio riskiness is growing beyond acceptable levels. Inadequate rating discipline. To be effective, rating systems must be highly accurate and reliable. This requires a priority on getting the rating grade right at loan inception and keeping it up-to-date during the life of the loan. The minimum standard for accuracy should be 95%, with 97% being the ultimate target. Most banks have an effective discipline on criticized and classified credits because they receive so much attention from the regulators. However, that discipline does not hold on pass credits. In fact, many banks are little concerned whether their pass ratings are accurate since they regard all pass credits as sound and therefore, not requiring differentiation. Borrower ratings versus transaction-ratings. Seventy percent of community bankers rate borrower and loan together. Rating borrowers rather than transactions creates builtin inaccuracies in portfolio risk information. The logic in favor of transactions is that borrowers often have multiple transactions with differing risk characteristics. By lumping all transactions into the borrower rating, these differences are obscured. Consequently, the result is a ratings hash that can significantly misrepresent the risk in the portfolio. For example, a single high-risk transaction within a borrower s total outstandings may be insignificant. But if there are numerous such transactions being hidden within the portfolio, there may be a high level of imprecision in portfolio risk information. Rating transactions and not borrowers offers the ability to more effectively accumulate like-rated transactions into useful risk data. Figure 1 Model Portfolio Risk Rating Guidance Ranges 1& % % % 5& % % Rating Grades 1 minimal risk 2 modest risk 3 average risk 4 acceptable risk 5 marginally acceptable risk 6 watch 7 OAEM 8 substandard 9 doubtful 10 loss Insufficient linkage of risk-rating and loan approval processes. A traditional approach to risk ratings has been to make the loan decision and then have Credit Review or a credit officer assign the risk rating. This practice is still followed in many banks and is considered a key ingredient in their risk management process. It certainly has merit since objective parties are less likely to be influenced by personal bias. However, the limitations to this approach seem to outweigh its benefits. The accurate determination of risk is the major ingredient in the loan decisioning process, to include not only the go or no go decision but also the amount to be lent, the terms and covenants and the pricing. Since lenders have responsibility for making, structuring, and pricing loans, it seems logical that they should also be the ones to risk-rate them. What is needed is a transactionbased rating approach that can be an integral part of the loan underwriting and decisioning processes. In other words, the process of evaluating and rating the transaction risk and the process of underwriting are fully integrated. Once the rating is determined, the loan decision follows as a byproduct. To ensure accuracy, the rating system must be further refined and guidance provided to correlate risk ratings with positive and negative loan decisions. Loan Decision Matrix At Bank South, modification of an eight-point rating scale resulted in a 38 The Journal of Lending & Credit Risk Management November 1998

4 loan decision matrix that accomplished the bank s objective of correlating the underwriting, rating, and loan decisioning processes. To begin using the matrix, a rating was split into its major components with individual ratings assigned to each component: Borrower. Transaction. Industry/Market. The borrower was rated on a scale of 1 to 8, with 1 being the strongest. In Bank South s credit underwriting analysis, the elements of borrower riskiness were rated, including financial condition and trends, character and management, capacity, and company size. The process was automated with weightings for each element so that an ultimate borrower score was derived by the computer. Analysis was highly focused with comments on how each element was rated but with very limited dialogue. For the transaction, collateral, guarantors, and loan term were rated within three categories strong, average, and weak. A rating guidance was provided with rating guidance for all descriptions of collateral type based on quality, marketability, and margins. For example, highquality accounts receivable with normal margins would have been found within the average rating category. The resulting grade was assigned as S, A or W to avoid confusion from having numerical ratings for all three components. The industry/market also was rated on a scale of 1 to 3, based on judgment of the strength and weaknesses associated with the industry or market. These ratings were provided from a centralized credit analysis unit, with Grade 1 being the strongest. Because the borrower analysis was automated, a detailed guide for rating transactions was provided, and the industry was analyzed centrally, the process became consistent and reasonably easy for the lender. Bank South also facilitated the rating process on new transactions by providing guidance that all 1-rated borrowers were acceptable risks and all 5-rated borrowers (watch) were unacceptable, unless there were strong mitigating factors. This narrowed the decision process to borrowers rated in the 2, 3, and 4 categories. To provide further guidance on borrowers in the 2-4 categories, a matrix was created (see Figure 2) with all possible combinations of rating components (borrower, transaction, and industry). Following the basic premise of expert systems, the best credit and lending people in the bank were asked to determine Figure 2 Loan Decision Matrix which ratings were clearly acceptable risks and which were clearly unacceptable. There were only five remaining categories which were borderline. The conclusion was that these borderline rated categories were where the emphasis should be placed in the loan decisioning process. Additional guidance was developed for borderline credits. The loan decision usually turned on judgment as to the relative strength or weakness of the individual components of the rating. For example, a borderline composite rating was 3A3 an average borrower within an average industry and with weak collateral. This loan decision turned mostly on the strength of the borrower with some consideration given to the relative weakness of the collateral. A weak 3 borrower rating usually meant a no decision. A strong 3- rated borrower was a yes decision. B T I B T I B T I 2 S 1 yes 2 A 1 yes 2 W 1 yes 2 S 2 yes 2 A 2 yes 2 W 2 yes 2 S 3 yes 2 A 3 yes 2 W 3? 3 S 1 yes 3 A 1 yes 3 W 1? 3 S 2 yes 3 A 2 yes 3 W 2? 3 S 3 yes 3 A 3? 3 W 3 no 4 S 1 yes 4 A 1 yes 4 W 1 no 4 S 2 yes 4 A 2 yes 4 W 2 no 4 S 3 yes 4 A 3? 4 W 3 no B = Borrower T = Transaction I = Industry/Market? = Borderline Decision Note: This is a sample matrix. Each bank should determine its own matrix based on its unique risk tolerance. 39

5 Using the matrix, lenders could make lending decisions based on the bank s risk tolerance that was built into the matrix. Additionally, there was consistency in decision-making since individual biases were replaced by the collective judgment of the bank s best credit people, who constructed the matrix. Of course, it was necessary to have great confidence in the ability to rate credits accurately and that required extensive testing of the system and training of the lenders. Fortunately, there was already a strong rating discipline in place. The end result of this approach was that skilled lenders those who could accurately risk rate credits could be provided much more lending authority. The credit officers and managers could then focus their attention on the borderline decisions, thereby placing the crucial resources where they were most needed. Credit ratings and loan decisions were checked after the fact to ensure system integrity and good decisionmaking. This system requires excellent rating discipline with authority and responsibility placed squarely on the shoulders of the lending officers. Because risk rating becomes such a high priority, better decisions are ensured and the quality of portfolio information is improved. Banks that utilize this approach usually decide to correlate the component ratings in the matrix with traditional single digit ratings for example, 1-8 at least as an interim step. Sometimes it is done because loan systems cannot adapt to three digits; other times it is to allow for a transition to the new system. Another consideration is regulatory acceptance. The benefits from following this rating approach proved to be numerous, including: It is an effective means for determining and communicating the bank s risk tolerance. It is easy for lenders to understand and use. Borrower and industry ratings are reusable for new borrowings with an existing customer. Because the bank s credit expertise was built into the model, less time is required in the credit screening process. Lenders have a clear idea what is and is not acceptable and so waste less time chasing unacceptable deals. Response time on loan requests can be significantly reduced. The matrix allows credit experts to concentrate their energy and time on borderline credits and allows lenders more decision making authority. It has proven to be a more accurate and effective rating methodology. The matrix is designed to provide decision guidance based on risk alone. There often are other factors which will override the matrix decision. Summary Individual transaction decisions are still the basis for a sound portfolio. These decisions depend upon a thorough and accurate evaluation of risk. The result of that analysis should be a risk rating that expresses the risk accurately and becomes the basis upon which the loan decision is made. If that rating is accurate for as long as the loan is an asset of the bank, then the rating becomes the primary raw material for portfolio credit risk management. When the rating system also provides sufficient risk differentiation, the bank has the ability to create portfolio management data and to measure total transaction risk within the portfolio. This article has outlined several uses for risk ratings loan decisions, model portfolios, problem loan identification, and pruning. There are many additional uses, such as risk/return analysis and pricing, migration analysis and loan loss reserve allocations, and lending guidance limits. No bank today can afford not to have effective risk management systems and these systems cannot function without an adequate risk-rating process. John McKinley can be contacted by telephone at or by e- mail at jcmckin2000@aol.com. 40 The Journal of Lending & Credit Risk Management November 1998

Effective Credit Risk Management with ErmsCo Dual Risking Rating System

Executive Brief Effective Credit Risk Management with ErmsCo Dual Risking Rating System 2007-2012 ErmsCo Inc Intellectual Property. All Rights Reserved. 103 Introduction to ErmsCo About ErmsCo ErmsCo is