Fin 129 Financial Institutions Management Credit Risk Fin 129 Financial Institutions Management
Assessment of Credit Risk FI manager must be able to: Price Loan correctly based upon risk evaluate possibility of default and make sure total risk limits are not violated
Credit Quality Recent concern over the quality of credit: Rapid growth in loans (over 10% a year in the late 1990’s Commercial Real Estate Loans Low-quality auto loans Credit cards Impact on charge offs Since 1991, the ratio of nonperfoming loans (90 days past due) has actually increased while growth has increased.
Commercial and Industrial Loans Syndicated Loan -- provided by a group of FI’s instead of an individual lender -- spreads risk Secured Loan - backed by specific assets Unsecured Loan - only a general claim on assets Spot Loan --borrower takes entire amount at one point in time in contrast to a loan commitment Commercial paper -- short term unsecured borrowing by firms
Real Estate Loans Mainly Mortgage Loans Adjustable Rate Mortgages creates prepayment risk and default risk. Adjustable Rate Mortgages interest rate is adjusted periodically.
Consumer Loans Credit Card loans Revolving Credit- Open line of credit where the borrower can borrow and repay at will Usury Ceilings -- maximum rate that an FI can charge on consumer and mortgage debt
Credit Analysis Essentially default risk analysis investigating the borrowers Evaluating Credit risk inherent in the operations of the business (or activities of the individual)? What can be done by the borrower to lower the risk? How can the lender control and structure the risk? (execution and administration of loan)
Evaluating Credit Risk the 5 c’s +1 Character -- Capacity -- Capital -- Conditions – Collateral -- Cash -- Which of the characteristics is the most important?
Evaluating the 5 c’s Character Based upon reputation of firm and past borrowing experience with the lender. Creates an implicit contract that guarantees loans will be made and repaid Works to the disadvantage of small and first time lenders Capacity Does the representative have the legal ability to commit the firms resources
Evaluating the 5 c’s Capital Borrowers wealth position and can it withstand changes in economic conditions Look at simple ratio analysis, Debt equity ratio Volatility of earnings - more volatility implies higher probability of default (Cash) Liquidity of capital
Evaluating the 5 c’s Conditions Market Specific Factors, common to all firms Current phase of business cycle and relation to business of firm Collateral What assets can be pledged to secure the loan? Are claims on assets senior to other claims?
Stages of Credit Evaluation Historical Perspective Overview of management & operations Business and industry outlook report i.e. competitors, suppliers (conditions) Background info (character) Common size financial ratio analysis compare to industry averages (liquidity, leverage, profitability) (capital, collateral, cash) Analysis of cash flows (capital and cash) Cash based income statement, Investigate sources and uses of cash
Stages of Credit Evaluation First three provide historical perspective -- then look at future Projections of borrowers financial condition Pro forma Financial Statements Attempt to provide an objective outlook at the future prospects Run sensitivity and Scenario Analysis on the projections
Credit Execution and Administration Loan Covenants Specific requirements either party must adhere to Affirmative requires borrower to take certain actions (Maintain liquidity position) Negative -- restricts the borrower from certain actions (acquiring more debt) Credit Review Review outstanding loans and monitor problems loans
Credit Execution and Administration Default Scenarios What actions (or inactions) would constitute default Who is responsible for collection costs, attorney fees etc… What actions the are the lender legally allowed to take.
Credit Execution and Administration Documentation Collateral needs to be “perfected” -- FI wants to have senior claims Position Limits The maximum amount of allowable credit to a single borrower Risk Rating FI can grade (rank) individual loans and counter parties (we explain how soon)
The 5 bad c’s the FI should avoid Complacency -- assumes that since things were good in the past Carelessness -- Poor underwriting techniques Communication -- Loan policy needs to be communicated to loan officers and enforced Contingencies -- tendency to downplay or ignore Competition -- Following changes in competitors practices instead of following own policies.
Evaluating Credit Risk Credit Scoring Models Quantitative models that use observable characteristics to score or rank borrowers based on probability of default. Credit scoring provides a measure of the possibility of default based upon characteristics of borrowers. Characteristics cannot be prohibited info (antidiscrimination laws - sex, race, not included) and must be statistically justified in relation to default risk.
Credit Scoring Models Linear Probability and Logit Models Uses historical data to explain the repayment experience of old loans. Divide loans into two categories those that did default (prob of default =1) and those that did not default (prob of default = 0)
Linear Prob. and Logit models For the given set of variables the following regression can be estimated:
Linear Prob and Logit Models Given the estimated values of Bj, you can take the loan applicants current values often variables and estimate the expected probability of default Z.
Linear Prob and Logit Models Strengths Weaknesses Use of logit solves this
Liner Discriminate Credit Scoring Models Divides borrowers into risk classes based upon aggregate score, does not estimate For each observable variable, a weight is determined based upon past experience of loans. Then an aggregate value is calculated and the loans are separated by the high or low probability of default.
A second version Points assigned based on characteristic and past experience (For example length of time in current job (more than a year add 5, less than a year add 2). Then Aggregate score is calculated.
Fico Scores Most credit scores in the US are calculated by software developed by Fair Issac and Company There are three main providers of credit scores: Equifax, Experian and TransUnion Most lending institutions will obtain scores from multiple services when evaluating credit risk.
Distribution of Fico Scores* Source www.Fico .com
Fico Score Comparison Score Rate Spread Additional Cost 720-850 5.783 $150,000 30 year fixed rate mortgage national averages Score Rate Spread Additional Cost 720-850 5.783 700-719 5.903 .125 $4,308 675-699 6.446 .663 $23,139 620-674 7.596 1.813 $64,870 560-619 8.531 2.74 $100,138 500-559 9.289 3.506 $129,139 source www.fico.com
Payment History Payment Info on specific types of accounts Public Records (bankruptcy, suits, wage adjustments, past due items, etc.) Severity of past delinquencies Time since delinquency or poor public rec. Number of
Amounts Owed Total amount owed on accounts Amount owed on specific accounts Lack of a specific type of balance Proportion of
Length of Credit History Time since account opened Time since account opened, by type of account Time since account activity
New Credit Number of recently opened accounts and proportion of accounts recently opened by type Number of recent credit inquiries Time since recent account openings by type Time since credit inquiry Re-establishment of positive credit history following past problems
Type of credit used Number of (presence, prevalence, and recent info on) various types of accounts (credit cards, installment loans, mortgage etc)
Note The same factors may impact different applicants in different ways. Some factors may be more or less important depending upon the other factors. No one piece of info will determine your score.
Not in your credit score Race, age, religion, nationality, sex, marital status (Regulation B) Salary, Occupation, title, employer, date employed, employment history Location Rates charged on outstanding credit Child/family support obligations and rental agreements.
Average Credit Statistics Number of Obligations: 11, 7 credit cards, 4 installment loans Past Payment Performance: 4/10 30 days late or greater 2/10 60 days late or later, 85% never had loan 90+days overdue. Credit Utilization: Credit Cards 48%<$1000 54% < 10% > $10,000 Total (less mortgages) 54% < $5,000, 30% > $10,000
Average Credit Statistics Total Available Credit: $12,190 combined on all credit cards, 1/8 uses more than 80% of available credit, over 50% use less than 30% of available credit Length of Credit History: average 12 years1 in 5 have histories over 20 years, 1 in 20 shorter than 2 years Inquires: one a year average, less than 7% have more than 4 inquiries in past year.
Problems with Credit Scoring Models in general Limited number of cases (in the extreme only two are considered - default no default) No obvious economic reason that future will reflect the past. Need to adjust the model constantly to account for possible changes Ignores some relationships such as borrower lender relationships. No centralized database of business loans to provide measure of market risk.
Newer Models of Credit Risk Term Structure Approaches Mortality Rate Approaches RAROC models Option Models Credit Metrics Credit Risk +
Term structure approaches The goal of a term structure approach is to derive the probability of default from observed differences in yield (risk premiums). Construct zero coupon treasury yield curves and zero coupon corporate curves for similar rated debt. Look at risk premium for a given maturity.
Term structure approaches Assume that the yield on the low grade bond is 8% and the yield on the treasury is 5%. If an investor is indifferent between the two options it implies that the expected return on the risky asset equals the return on the risk free asset or p(1+k) = (1+i) where p = probability of no default, k = return on risky asset, i = return on treasury
Term structure approaches Assume that k = 8% and i = 5% then the implied probability of no default, p, is found by:
Term structure approaches If the loan is perceived to have a 1-.9722 = .0277 or 2.77% probability of default it should have a risk premium set equal to 3%. Can be extended to account for a partial recovery of principle and interest in the case of default. Let g represent the proportion of the loan that is recoverable (1-p)g(1+k) +p(1+k) =(1+i) This can also be extended to the case of multi period debt instruments
Term structure approaches For multiperiod debt instruments you need to calculate the marginal probability of default for each time period then aggregate this into a cumulative probability of default. Using the forward rate and the cumulative probability, an expected probability (or implied) probability of default can be found.
Marginal Mortality Rate The marginal mortality rate of the loan is the probability of the of the loan defaulting in a given year of issue
RAROC Models Risk Adjusted Return on Capital Models Income should be adjusted for fees and interest spread
RAROC The most difficult part of the analysis is finding the capital at risk. One approach would be to use duration.
CreditMetrics and Credit Risk + Uses value at risk to find the possible loss on the loan portfolio given the assumption of “a bad” outcome over the given time period. Credit Risk + Similar to Value at Risk Calculating the FI’s required capital reserves to meet losses above a given level.
Loan Portfolio and Concentration Risk
Migration Risk The risk that the quality of a loan or portfolio of loans will decrease (credit risk increase) If the credit rating of a given industry or group of loans declines then lending in that industry will decrease.
Migration matrix Presents the probability of a loan being upgraded or down graded over a given period of time. Generally the matrix represents a given industry or region, but could be more widespread.
Rating Migration Corporate Debt
Using the matrix If the FI realizes that a larger portion of loans in a category has been downgraded it can chose to reallocate its loans moving some into a higher rating class.
Concentration Limits Limits placed externally on the total amount of credit placed with a given borrower. The concentration limit is a function of the maximum loss as a percent of total capital and the amount lost in the event of default.
Concentration Limit Assume that the Maximum loss is 10% Loans in the sector on average loose 40% of capital in the event of default
Concentration limit con’t If the firm has $100 million in total capital it would be willing to place 25% of its capital in this sector. If all $25 Million defaulted, there is a loss rate of 40% (60% gets recovered) or an expected loss of $25 Million (.4) = $10 Million which is 10% of total capital
Modern Portfolio Theory Basics of diversification (review) By increasing the number of assets in the portfolio we can eliminate systematic risk. The amount of risk that can be eliminated depends upon the correlation of the assets. As long as the assets are not perfectly correlated there is a gain to diversification. The same idea can be applied to loans.
Expected Portfolio Return The combined return of two assets is simply the weighted average of their returns
Variance of Returns as long as the correlations are less than one (preferably some being negative) the portfolio variance will be reduced.
Efficient Frontier By changing the weights in a portfolio you get different return and risk combinations. It is often possible to rearrange a portfolio and produce a higher return without changing the risk. The efficient frontier provides the set of portfolios that produces the highest return at each level of risk.
Efficient Frontier Given four assets, the next slide shows a graph of 76 different portfolios created by changing only the weights in the portfolio. The vertical axis is the return on the portfolio, the horizontal axis represents the standard deviation of the portfolio. The efficient frontier is the set of points that provides the highest return for each level of risk.
Available combinations Given the efficient frontier you can increase return for a given level of risk. You can also decrease risk to find the minimum risk portfolio.
Problems with MPT and loan management Loans and many other assets held by FI’s are often:
KMV Portfolio Manger Since many assets are non traded it is difficult to calculate the inputs in used in modern portfolio theory Three inputs are needed: return, variance of each asset and correlations (or covariances)
KMV portfolio manager Return on Loan AISi = All in Spread = Spread on the loan + fees E(Li) = expected loss on loan EDFi = Expected default frequency LGDi = Loss given default
KMV Portfolio Manger Risk of the loan ULi=“unexpected” loss on the loan Assumes defaults are binomially distributed.
KMV Portfolio Manager Correlations are based upon the default risk correlation of the firms assets over time. The correlations are therefore relatively low ranging from .002 to .15.
KMV Portfolio Manager Given the individual returns, variances, and correlations the portfolio returns can be calculated using the standard portfolio theory.
Loan Volume Based Models Concentration risk can be analyzed based upon data established for large volumes of data for example: Commercial bank call reports: reports to the Federal Reserve loan allocation among different asset classes. Can be aggregated and used as a benchmark. Shared National Credits - based on SIC codes. Again can be used for benchmarking
Factor based analysis Based on the systematic loan loss for a given sector Bi is the systematic loss sensitivity of the ith sector. Where the entire portfolio has a B =1
Regulatory models General limits are often also set by regulatory agencies. For example life-health insurers can have at most 3% of their portfolio in a single issuer or security.