CREDIT RISK MEASUREMENT AND MANAGEMENT OF THE LOAN PORTFOLIO LEARNING OBJECTIVES By the end of this chapter you should be able to: 1. 2. 3. 4. 5. 6. 7. 8. 9. describe the benefits of credit risk management explain and use Altman’s score explain how stock prices can be used to explain credit risk explain how actuarial approach can be used by examining CreditRisk+ explain how a macroprudential approach can be used by looking at CreditPortfolio view suggest how risk-adjusted return on capital can be used for portfolio purposes use the Sharpe Index for lending purposes calculate the risk of a loan portfolio using CreditMetrics™ understand the elements of loan pricing. z KEY TERMS z Altman score credit default swap credit migration default point option modelling put option Sharpe Index CreditRisk+ call option credit derivatives credit options duration pass through risk-adjusted return on capital CreditPortfolio CreditPortfolio View 375 concentration risk credit event CreditMetrics™ expected default frequency pay through securitisation zone of ignorance 376 INTRODUCTION As mentioned in the credit scoring chapter, mathematical modelling is increasingly being used for the measurement of credit risk (see Chapter 3). Originally, lending institutions did not have departments that took an institutional perspective of the lending portfolio. Loans were granted and were promptly forgotten as long as borrowers made their scheduled repayments. These practices, however, led to loan portfolios that were not close to the efficient frontier. The efficient frontier is a finance concept. In brief, it maps the return that an investor should receive for a given level of risk. In terms of lending, it means receiving an appropriate return for a given level of lending or credit risk. The aim of credit risk management is to balance between risk and return to achieve optimum profitability and efficiency. Bank managements realised that they were not doing this. Exacerbating this situation was the rise of relationship banking. This form of banking was based on a manager developing close links with an entity or individual. Often, lenders would lend too much to one entity relative to the rest of the lending portfolio. This is known as concentration risk. (We address concentration risk in Chapter 16 - Quantitative Finance.) The underlying risks may be sound, but there is a danger when the economy undergoes turmoil. Any bank that was overly exposed to the airline industry in September 2001 might have had problems due to the downturn in the airline and ancillary industries. This is the same for the mining industry leading up to 2017. Not all would agree that concentration risk is a bad thing. Some have pointed out that small financial institutions that focus their business choose to wear concentration risk because they have particular expertise. Many regional building societies, for example, have concentration risk in terms of their credit type and geography. Lenders recognised that lending needed to be conducted on a more scientific basis, rather than relying on the lender-borrower relationship. This change would remove the subjective nature of lending. Further, rather than managing loans on an individual basis, lenders needed to manage loans on a portfolio basis, just like any other investment portfolio, such as bonds or equity. Finally, banks questioned what happened when there were identified loans that were not appropriate to the statement of financial position. This question has given rise to the new credit management techniques of securitisation and credit derivatives. Credit risk management helps a bank to achieve the following objectives: (a) achieve an appropriate balance between risk and return, (b) avoid concentration risk, (c) manage loans on a portfolio basis and (d) take a group of loans off the statement of financial position. Before credit risk can be managed, it is important to know the extent to which the financial institution is exposed to it. This knowledge is achieved by measuring credit risk. As you read this chapter, keep in mind that modelling is being developed at a rapid rate; the models we are looking at today are the most popular and have regulatory approval. In the next section, we discuss the various techniques of the measurement of credit risk:. 377 CREDIT RISK MEASUREMENT Measurement of credit risk refers to the quantification of credit risk exposure of a bank. It has a number of useful benefits, such as: ■ removing the subjectivity from credit assessment ■ adding a scientific basis to the credit assessment process ■ rating loans that have no credit ratings or providing a system of grading ■ providing a mechanism for monitoring loans to ensure they are not potential problem loans. Many credit risk measurement models are available today. Many have been internally developed by lending institutions, while some are available as 'off-the-shelf solutions to lending institutions. Lending institutions often use an 'off-the-shelf solution as a check against an internal model. While there is a plethora of models, most are based on either accounting ratios or information contained in share prices. In the recent past, given the limitations of some these models, others have been proposed. These include using actuarial approaches and incorporating endogenous variables. Keep in mind that the development of all these models indicates that none of them are prefect. z In this chapter, we will examine four models that use these approaches. The Altman score was one of the earliest models used by banks in an attempt to remove the subjectivity from the lending decision. It examines the accounting ratios of two populations—failed and non-failed companies—to assess credit risk. Despite questions about its “look back” methodology, it is relatively robust and is still used used. Another methodology has been the use of information implied in share prices and option pricing theory to predict default. This approach has been popularised by KMV Corporation. More recent developments have been the use of actuarial methodologies as proposed by Credit Suisse's CreditRisk+ model and the incorporation of economic variables as developed by McKinsey's Credit Portfolio View model. ALTMAN'S z SCORE Altman's (1993) work was predicated by work by Beaver (1967), who found that firm bankruptcy could be predicted by the use of financial ratios for up to five years before bankruptcy. Beaver used univariate models that distinguished between failed and non-failed firms for periods of up to five years in advance. This mathematical method was derived from the biological sciences, which define populations with different characteristics. Table 11.1 below shows a strong relationship between credit ratings and key financial ratios. The differences between the ratios of AAA-rated organisations and B-rated organisations are quite marked. 378 To some extent, the use of accounting ratios in such analysis is a logical extension of original credit analysis where a person would be the mechanism of analysis, generating and using the ratios. Altman and others extended the work of Beaver to improve the predicative power of such models to predict default and/or bankruptcy. The analysis was improved by extending single univariate models to multivariate models that used a number of financial ratios. A survey of the research over the past thirty years has sought to identify which accounting ratios provide the best predicative power, including: ■ activity ■ liquidity ■ solvency ■ profitability ■ earnings variability ■ size. Table 11.1 Relationship between ratings and financial ratios Adjusted key industrial financial ratios - 1 US industrial long-term debt Three-year medians (1998-2000) EBIT interest coverage (x) EBITDA interest coverage (x) Free operating cashflow/ total debt (%) Funds from operations/total debt (%) Return on capital (%) Operating income/sales (%) Long-term debt/capital (%) Total debt/capitalisation (incl. short-term debt) (%) Companies (no.) Credit rating AAA AA A BBB BB B ccc 21.4 26.5 10.1 12.9 6.1 9.1 3.7 5.8 2.1 3.4 0.8 1.8 0.1 1.3 84.2 25.2 15.0 8.5 2.6 (3.2) (12.9) 28.8 34.9 27.0 13.3 55.4 21.7 22.1 28.2 43.2 19.4 18.6 33.9 30.8 13.6 15.4 42.5 18.8 11.6 15.9 57.2 7.8 6.6 11.9 69.7 1.6 1.0 11.9 8.8 22.9 37.7 42.5 48.2 62.6 74.6 7.7 8 29 136 218 273 281 22 a times number (that is, a number to multiply by) EBIT = earnings before interest and tax; EBITDA = earnings before interest, tax, depreciation and amortisation. X= Source: Standard Poor's, www.standardandpoors.com. Depending on the industry generally, each of the above ratios has strong predicative power. As a general rule, however, liquidity, profitability and earnings variability have strong predicative power. This is easy to explain. Cash repays loans, so those ratios with a strong cash component are the ones that indicate the potential for debt defaults. 379 In the remaining part of this section, we will concentrate on the development of the Altman score as a basis for considering later developments. The Altman z z score is given as: z = 1.2X, + 1.4X2 + 3.3X3 + O.6X4 + 0.999X5 where X, = the working capital divided by total assets X2= the retained earnings divided by total assets X2 = the earnings before interest and taxes divided by total assets X4 = the market value of equity divided by the book value of total liabilities X5= sales divided by total assets. The first four variables are expressed as decimals while the fifth is expressed as a times number (for example, 5 x) as opposed to a percentage or decimals. We will explain the interpretation of the result of this score later, because first it is instructive to understand why these particular variables were chosen. Altman’s initial work showed that many ratios were reasonable indicators of the potential for default. They were further categorised into the categories of activity, liquidity, solvency, profitability, earnings variability and size. Altman chose, however, to concentrate on those ratios that were well accepted in the academic literature. He reduced the number of variables to twenty-two and used the following method: 1. 2. 3. 4. Altman observed each statistical significance of various alternative functions, including determining the relevant contributions of each independent variable. He evaluated the intercorrelation among the relevant variables. He then observed of the predicative accuracy of the various profiles. He used his own judgement when finalising the function. It needs to be recognised that the resulting function, as stated above, provides a good discriminant between failed and non-failed companies but is not necessarily optimal. The categorisation of these ratios then becomes as shown in Table 11.2. The following issues should be considered in the use of this technique: 1. 2. 3. 4. It is heavily biased towards US data, so an examination of Australian data may create a different score. The function implies a heavy bias towards financial ratios that indicate the firm’s ability to create cashflow. This is obviously satisfying given that cashflow repays debt. There are obvious problems with the use of financial ratios: a. They are open to manipulation by firms. b. They are also open to interpretation by analysts. The function is independent of loan amount, which alerts us to the need for supplementary analysis of the statistical method. z 380 Table 11.2 Ratio categorisation Variable x,= ie x3 = Category Liquidity Profitability Profitability/ productivity Leverage/solvency Activity It is also instructive to picture graphically (Figure 11.1) the results of the discriminant function, with the objective of having the groups as different as possible. Figure 11.1 Multidiscriminant distributions using z scores The bounds of the two distributions are 1.81 and 2.99. A firm that scores 2.99 or above, therefore, would be considered to be to be creditworthy, while one below 1.81 would be considered to be non-creditworthy. The difficulty for lending officers is with scores that fall between 1.81 and 2.99. This is known as the zone of ignorance and represents sampling errors between the two populations. In this zone, good loans could be classified as bad, and bad loans could be classified as good. Statistically, the former is known as a type 2 error while the latter is known as a type 1 error. Each situation is an opportunity cost to the lender, although the latter has a more serious explicit 381 cost. Ways of dealing with the zone of ignorance differ among lending organisations, but the following would be considered: ■ the risk profile of the lending organisation ■ the lending organisation’s relationship with the potential borrower ■ the judgement of the analyst. The following example should assist in your understanding of this technique. Imagine you are a lending officer with XYZ Bank. You are given the following information about Company ABC: Table 11.3 Company ABC Assets Current assets Noncurrent assets Shareholders’ funds Liabilities Current liabilities Noncurrent liabilities $10 $20 $6 $9 $15 You also know that earnings before interest and tax are $2 and sales are $35. Would you lend to the company if you followed the Altman method? Table 11.4 Company ABC and the Altman method Result Variable (10-9) X, 30 6 -r30 *3 =0.037 = 0.2 2 —5— =0.067 30 6 x5 35 ... 30 =1.67 times Before providing the solution to this example, we should point out that such a calculation may involve making assumptions. This is not unusual, but care should be exercised in arriving at the 382 assumptions. In this example, we have assumed (given lack of information, which also is not unusual) that: ■ the equity on the statement of financial position is all retained earnings, and ■ the equity on the statement of financial position is also the market value of equity. We can now complete the exercise. Using the formula for Altman’s z score, we have: z = 1.2X, + 1.4X2 + 3.3X, + 0.6X4 + 0.999X5 = 1.2(0.037) + 1.4(0.2) + 3.3(0.067) + 0.6(0.25) + 0.999(1.67) = 2.36 The score falls into the zone of ignorance (between 1.81 and 2.99). Depending on XYZ Bank's tolerance in the zone of ignorance, there is a strong likelihood that you would lend to this company. If, however, you made different assumptions regarding equity (retained earnings and the market value of equity), then the score would most likely fall below the lower bound of 1.81. z z score was developed in 1968, there have been developments in its form: A private firm z score has been developed, which takes into account that the market value Since the • ■ ■ of equity is not available. The major change is in X4, where the book value of equity is directly substituted for the market value of equity. It has been recognised that non-manufacturing firms, without assets, are prejudiced against because assets are used in many of the ratios. Two major changes are thus made: X5 is dropped and the book value of equity is used for Xj. Finally, a second generation model, known as ZETA, was developed in 1977 to account for changes in business failure. The most significant change in this model is the recognition that the size of the firm makes a difference. The general result has been that the larger the firm, the less likely it is to fail. This results occurs because smaller firms are often newer and more likely to enter bankruptcy. While these newer models have different functional forms given their different variables, they follow the same principles. Finally, it should be noted that many other models follow the same techniques, particularly credit scoring, in assessing credit risk. Using stock prices The criticism of Altman-style models, which use financial ratios, is that they use data from financial reports. The data are thus 'old' and backward looking, whereas default is about looking forward to potential defaults. If we consider, however, that the market price of assets and the market value of debt are forward looking, then using this information, we may be able to calculate the default probability. This is how KMV Corporation’s expected default frequency model works, using option modelling. The lending proposition of a bank is to lend money to an organisation that uses the funds. If the firm uses the funds well, then the value of assets will rise and the borrower will repay the loan. If, however, the funds are used badly, then the asset value will fall, the value of the asset may be 383 unable to repay borrowings and a default will occur. If we graph this scenario, we find Figure 11.2. (Note that there is an upper bound on returns from lending.) Figure 11.2 A sold put option in bank lending Those who are familiar with option modelling will see that this is a sold put option on assets where the equity value will be a function of the market value of assets and its volatility, the liabilities of the firm, interest rates and the term to maturity, in terms of option pricing. The default decision becomes an issue of how close the market value of assets is to the default point. The aim then becomes to predict the probability that asset values will fall below the default point. If we look at the lending decision from the equity holder's point of view, then the pay-off diagram shown in Figure 11.3 would result. Figure 11.3 A call option representing the equity holder's interest in bank lending 384 Interpreting this graph is simple. When borrowing money, the equity holders would repay their loan if the asset value were greater than the liabilities. Again, those with a familiarity with option pricing will see that this is a call option on the assets of the company. The difficulty with the approach is that there are two unknowns in the formulation; the market value of assets and asset volatility. If we exploit the relationship between assets and equity in the option pricing formula, as well as the volatility, then we can solve the two unknowns. The functional forms would be: Price of risky debt = Option function (asset value, asset volatility, capital structure, interest rate, term to maturity) Price of risky debt = Option function (asset value, asset volatility, capital structure, interest rate, term to maturity) We are assuming that the asset value and its volatility can be inferred from the option pricing model. We can now solve the two equations and find the values attached to assets and volatility. How do we use this number? The following principles apply. The net worth of an organisation is simply the market value of assets less than the default point. Under the popular KMV Corporation method, the default point is all the current liabilities plus half the long-term liabilities. This recognises that not all debt is due at the same time; the presence of long-term debt does provide breathing space in times of financial stress. The net worth, however, should be considered in terms of business risk. In other words, some firms can afford greater leverage (that is, risk) than others can. The market of assets magnifies the effect of the volatility on the asset size. We can, however, now calculate how far we are from default: Distance of default = (Market value of assets) - (Default point) (Market value of assets) (Asset volatility) This provides the distant to default, or the number of standard deviations to default. Once we have this figure, we can estimate the default probability. It is important to understand that some of this technology is proprietary to KMV Corporation, which maintains a database of 100 000 companies and 2000 incidents of default or bankruptcy. If our distance to default turns out to be four standard deviations, for example, then the KMV Corporation database will indicate the proportion of firms with four standard deviations distance to default that actually defaulted. This proportion is known as the expected default probability: Expected default probility = Number of default firms All firms of sample 385 The following sample estimation of default is taken from a technical document made available by KMV Corporation. There are three steps: 1. 2. 3. estimate the current value and volatility of the firm’s assets determine how far the firm is from default (its distance from default) scale the distance to default to a probability. Consider Philip Morris Companies Inc., which at the end of April 2001 had a one-year estimated default frequency of 25 basis points (0.25 percent). The calculation made is as shown in Table 11.5. Table 11.5 Sample calculation of estimated default frequency - Philip Morris Companies Inc. Variable Market value of equity Book liabilities Market value of assets Asset volatility Default point Value $110688 million $64 062 million $170 558 million 21% $47 499 million Distance to default 3.5 How calculated/information accessed Share price (accessed from the stock market) X shares outstanding (from the annual report) Statement of financial position (from the annual report) Derived from the option pricing model*, as described earlier in the chapter Derived from the option pricing model, as described earlier in the chapter Calculated as all short-term liabilities plus half the long-term liabilities, as described earlier in the chapter Calculated from the ratio of (170 - 47) (70 X 21%), which comes from the formula using the information derived above: / Distance of default = Estimated default frequency 25 basis points - (Market value of assets) (Default point) , , , , . , ... ( (Market value of assets) (Asset volatility) Empirical mapping between-distance to default and default frequency * KMV uses a standard Black-Scholes option pricing model. Students wishing to examine this model should refer to a text on pricing derivatives. Again, we note that the default probability, as in the Altman case, is independent of the loan size. Actuarial Approaches Many methodologies arise due to the limitations of prevailing methodologies. CreditRisk+ is no different, scores have the issue of back looking accounting figures and an assumption of KMV is dependent on the capital structure. The underlying assumption of Credit Suisse's CreditRisk+ is that there are no assumptions on why loans default except they do. z So, CreditRisk+ model, in its simplicity predicts whether a loan defaults or not and builds a distribution around this framework. While the statistics involved in this approach are beyond 386 this text, it is noted that the methodology of this approach is well represented by a Poisson distribution. The approach then takes a three building block approach to predicting credit default, both on an individual and portfolio basis. This approach is found in Figure 1. Figure 1 CreditRisk+ risk measurement framework (Source CreditRisk+) Input • default rates • default rates/volatilities Building Block #1 • exposures • recovery rates Building Block #2 Stage 1 Sage 2 Building Block #3 Building block number 1 identifies the average default rate for each obligor banding. Obligator bandings are normally based on credit ratings. It is obvious that average default rates have similar problems as other methodologies. Building block number 2 adjusts the average default rate for individual exposures and expected recovery rates. Building block number 3 then calculates the expected default losses. In terms of intuition, this model is attractive given it provides a yes/no answer. However, it does have its deficiencies. The main one is its strength. While it provides the yes/no result, it takes no account to credit migration. CREDITPORTFOLIO VIEW In previous chapters, we examined the business cycle and its relationship to lending. Creditportfolio View uses the assumption that lending decisions are connected to the state of the economy, that is, the business cycle. Therefore, implicit to the model, are macroeconomic variables such as: unemployment rates, GDP growth, interest rates, foreign exchange rates, government expenditures as well as savings rates. Before looking at this methodology, two comments need to be made as regulation may affect this methodology. Firstly, Basel 3 has through the cycle debt provisioning where probabilities are assigned for the possibility of debt default. Secondly, the rise of macro-prudential supervision means that regulators are acknowledging the effect that these types of variables have on lending. 387 Again, the statistics from this methodology are beyond the scope of this text, however, it is easily described. Default is modelled on a logit function where the dependent variable is a risky default with the independent variable an index based on current and lagged macroeconomic variables. The index itself is estimated by multi-factor model. The logit model is then “corrected” for transition probabilities indicating when the business cycle changes. PORTFOLIO MANAGEMENT While much of this book is devoted to the analysis of a single loan or a number of loans to the same borrower, attention has been given in recent years to the effect (in terms of the risk-return pay-off) of adding loans to an existing portfolio of loans. Concentration risk has gained much attention in this regard. A portfolio of loans can be viewed as a portfolio of assets, and the proposition has been that managers can use modern portfolio theory to manage the loan portfolios. The issue becomes how to manage new loans that are highly positively correlated to the rest of the portfolio. In good economic times, this is not a problem; if there is an economic downtown, however, the majority of the loan portfolio could be exposed to the same factors. In particular, smaller financial institutions that operate in a limited geographical area are subject to this risk. Portfolio management of loan portfolios is nevertheless a relatively new phenomenon and there are questions about the relevance of modern portfolio theory to lending portfolios. ■ • ■ Are the assumptions under which modern portfolio theory operates applicable to loans? ■ Modern portfolio theory assumes that the distribution of returns is normal. Loans, however, are characterised by having a one-sided distribution because the minimum return is assumed to be zero in default. • Modern portfolio theory assumes that all the assets in the portfolio can be revalued. This is simple in equity markets because shares operate on a share market. For bank loans, however, this is difficult because the loans are not traded on an exchange. Is the environment under which modern portfolio theory operates the same as that for lending portfolios? In terms of equity portfolios, for example, investors have the option of purchasing or not purchasing a share at a given price and return. Lending portfolio managers do not have such luxury, because they inherit loans approved by lending managers. Is concentration risk a problem? It could be argued that concentration risk is a result of concentrating lending in expert areas and that diversification could be conducted in different ways, such as by geographic region rather than by industry or company type. There are a number of different models for portfolio management, including the risk-adjusted return on capital, the Altman Sharpe Index approach and CreditMetrics™. The next sections are not designed to illustrate the flaws of existing methods, but will explain the approaches used. There is no doubt that better methods will become available. 388 Risk-adjusted return on capital One of the earliest attempts to address the problem of approving loans was prompted by the introduction of capital adequacy. While lending institutions are required to set aside capital for each class of loan for capital adequacy purposes, not all loans are equal. If two loans—one housing and the other corporate, for example—have the same interest rate, then the lending institution would prefer the housing loan because less capital would need to be put aside under the current capital adequacy guidelines (although changes to the guidelines are proposed). Bankers Trust (BT) originally developed the risk-adjusted return on capital. It recognised that capital was put aside for each loan and suggested that a hurdle rate for loans needed to be achieved before a lender added the loan to the portfolio. The approach is a return on equity approach, rather than portfolio approach, but has been used for portfolio purposes. It can be justified on the basis that loans of various risk classes are added to the portfolio if they provide the appropriate return. This can be expressed as: Risk adjusted return on capital = Income from the loan for one year Capital at risk The formula is deceptively easy but has many treatments. The following questions are among the issues: What income from the loan should be included? If the loan attracts further business, should this income be included? How is capital at risk defined? BT uses a duration number (see below), but many other methods can be used. A more accurate representation that could be used is the value-at-risk calculation derived by CreditMetrks™. ■ ■ BT uses a simple duration number to define the capital at risk as follows: AL ar where AL L = the percentage change in the market value of the loan over the period -D'l = Macauley duration (which is the weighted average receipts on a security or loan, where the weights are present value) = the maximum discounted change in the credit risk premium over the period' 389 It can be re-arranged to be: AL = Dl X L X AR 1+Rt where AL = the capital at risk. A simple example is as follows: the base rate for a loan is 8 percent and the loan also has various AR fees that total 0.5 percent. The duration of the loan is two years - ' p — and is 100 basis points. 1+rl The loan amount is $100 000. The income is the total of the various fees applied to the loan amount: 0.5% X $100 000 = $500 The capital at risk is as follows: AL = 2 X 100 000x 1% = 2000 therefore: T,. , ,. . , . .. , 500 Risk - adjusted return on capital = ■ - =25% 1 F 2000 As long as the lending institution’s hurdle rate (return on equity) is less than 25 percent, then the loan would be added to the portfolio. Hurdle rates are normally defined by lending institutions as the return on equity. These processes have been enhanced over the years. The most used risk-adjusted return on capital method was developed by Bank of America. There major departure Bank of America make from the BT methodology is the recognition that the capital at risk is unexpected losses. This is calculated by Bank of America as: CI X ơ. X LGD. X Exposure Where ■ CI is the confidence interval used by the financial institution; ■ ơ| is the standard deviation of similar types of loans; • ■ LGD. is the loss given default for the loan type; and exposure is the principal. Bank of America uses a confidence interval of six. However, rating agencies, who have stated a preference for the Bank of America approach, consider ten to be the appropriate number to obtain the top credit rating. 390 The last issue that we need to consider is that of scarce capital. A financial institution cannot continually lend, because capital is finite. What does a lender do if, for example, capital is exhausted and a loan application exceeds the required hurdle rate? The obvious answer would be to raise new capital, but this takes time. This is where the portfolio approach applies. Most financial institutions would allocate marginal capital for each loan, in recognition that it takes time to raise new capital for lending purposes. Altman'S Sharpe Index approach Altman's approach to portfolio management is characterised by the Sharpe Index where the portfolio return is adjusted by the portfolio risk. This approach is not unlike the risk-adjusted return on capital for a single loan. Altman's approach is simple: optimise the Sharpe Index subject to various constraints. It is important then to identify the various components. The portfolio return is the weighted average return of each asset in the portfolio. This recognises that the risk that a loan brings to a portfolio depends on the amount of the loan relative to the portfolio. The return of each asset is defined as the promised yield to maturity, less expected annual loss. There are many ways in which to calculate the expected annual loss, with many financial institutions using their historical experience for each loan type. To calculate annual losses, Altman uses an insurance concept called mortality rates and losses. This is the risk premium implied by the default experience of the credit rating of the loan. Financial institutions, therefore, would need to develop ratings for those loans without a credit rating. Using Altman’s terminology, the problem can be broken into steps. The first step is to calculate the return on the portfolio, using the following equation: N R= p £ i= X,EAR, i i 1 where R = the return on the portfolio X. = the proportion of each asset invested EAR, = the expected annual return. The next step is to calculate the variance of the portfolio using the following equation: N N z z i= 1 Wm j. 1 where V. = the variance of the portfolio. 391 Then, maximise the following relationship, which is the Sharpe Index: where N s x-1 i= 1 Rp > the target return X< the individual bond investment limit. Care must be taken when using this and other techniques, to ensure adequate data are available, particularly for portfolio variances. Otherwise, other forms of variance would need to be used. CreditMetrics™ CreditMetrics™ is a portfolio method that is growing in popularity. Its great attraction is that it incorporates the fact that credit risk changes over time. The technique would be familiar to those who have an exposure to JP Morgan's RiskMetrics™ and value-at-risk methods. CreditMetrics™, also developed by JP Morgan, seeks to model portfolio risk by tracking value changes in lending assets by assessing the probability of credit changes. CreditMetrics™ is best understood by using an example. The following example is taken from a JP Morgan technical document (available for free on the website www.riskmetrics.com.au). Suppose we have a BBB-rated bond with a maturity of five years. The bond has a coupon of 6 percent. The credit rating of this bond over a period of time can rise to AAA, fall to ccc or default. Each rating change has a probability, as shown in Table 11.6. Table 11.6 One-year transition matrix Initial rating AAA ' AA A BBB BB B ccc AAA 90?81 0.70 0.09 0.02 0.03 0.00 0.22 AA 8.33 90.65 2.27 0.33 0.14 0.11 0.00 A R.a ting at y ear end (i<') - BB_ . .. " 0.68 7.79 91.05 5.95 0.67 0.24 0.22 Source: Standard Poor's 1996, CreditWeek, 15April. ...»... BBB 0.06 0.64 5.52 86.93 7.73 0.43 1.30 ’ 07’12 " Ó.OÓ 0.06 0.74 5.30 80.53 6.48 2.38 0.14 0.26 1.17 8.84 83.46 11.24 ccc Default 0.0Õ 0.00 0.02 0.01 0.12 1.00 4.07 64.86 0.00 0.06 0.18 1.06 5.20 19.79 392 Table 11.6 shows that the most obvious scenario will be that the bond stays at the same rating, with little probability of it moving higher. (These credit migration probabilities are available from Standard & Poor's.) Given a yield curve and assuming that each credit rated bond has a coupon of 6 percent, we can calculate the value of each bond as shown in Table 11.7. Using a zero coupon yield curve, we can calculate the zero coupon rates as shown in Table 11.8. Table 11.7 Distribution of value of a BBB par bond in one year Year end rating AAA AA A BBB BB B ccc Value (s) Probability (%) 109.37 0.02 0.33 109.18 108.66 5.95 107.55 86.93 102.02 5.30 98.10 1.17 0.12 83.64 Source:JPMorgan 1997, CreditMetrics™—TechnicalDocument, www.riskmetrics.com. Table 11.8 Zero coupon rates, by credit rating category Category AAA AA A BBB BB B ccc 1 Year 1 (%) 3.60 3.65 3.72 4.10 5.55 6.05 15.05 Year2 (%) 4.17 4.22 4.32 4.67 6.02 7.02 15.02 Year3(%) j Year 4 (%) 4.73 5.12 5.17 4.78 4.93 5.32 5.25 5.63 7.27 6.78 8.03 8.52 14.03 13.52 Source: JPMorgan 1997, CreditMetrics™—Technical Document, www.riskmetrics.com. The above interest rates are taken from the corporate bond market for each of the credit ratings. Note that zero coupon rates are used rather than par coupon rates. While standard discounted cashflow method is used to reach the above values, you need to note the following treatments used by CreditMetrics™: ■ The first year's cashflow is not discounted. ■ The subsequent cashflows are discounted on an annual basis, despite most bonds being semi-annual in nature. ■ The value of the bond in default is not a discounted cashflow; rather, it represents a recovery 393 rate of 51.13 percent. This demonstrates that not all cashflows are received and that some need to be estimated. The value of the BBB-rated bond, therefore, will be: 6 6 6 107.55 = 6 + ——— -■ + ~1---- _ __ 2 + (1__ ______ 3 + (1 + 0.041) (1 + 0.0467)2 (1 + 0.0525)3 6 (1___ ______ _ (1 + 0.0563)4 The previous equation is simply the normal discounted cashflow formula that is used in finance: Cashflow from bond (1 + Zero coupon rate)" where n = the year. Figure 11.4 shows how the distribution appears visually. Figure 11.4 Distribution for a five-year BBB-rated bond in one year Frequent V 0 '.IM p ■I; ftp > 0 !Tó I.I.iljO - li t hoikoỉl Source: JPMorgan 1997, CreditMetrics™—Technical Document, www.riskmetrics.com. You will notice that the distribution is very skewed, not normal. This presents some statistical issues, but we will assume that the distribution is normal. Completing the calculations, we can find the standard deviation of the distribution, as shown in Table 11.9. The following are explanations of the terms in Table 11.9: • year-end rating: the credit rating at the end of year one ■ probability of state: the transition probability indicated by Standard & Poor’s as per Table 11.6 ■ new bond value plus coupon: the value of the bond as per the yield curve in Table 11.8 394 ■ ■ ■ probability-weighted value; the transition probability multiplied by the value of the bond. (The sum of this column is the transition-weighted value mean.) difference of value from mean; the new bond value less the value-weighted mean probability-weighted difference squared; the difference squared multiplied by the transitional probability. (The sum of this column is the value-weighted variance and the square root is the standard deviation.) Table 11.9 Standard deviation calculation - calculating volatility in value due to credit quality changes Year end rating AAA AA A BBB BB B ccc Default Probability of state (%) 0.02 0.33 5.95 86.93 5.30 1.17 0.12 0.18 New bond value plus coupon ($) 109.37 109.19 108.66 107.55 102.02 98.10 83.64 51.13 Mean = $107.09 Probabilityweighted value ($) 0.02 0.36 6.47 93.49 5.41 1.15 1.10 0.09 ProbabilityDifference of weighted difference value from mean ($) squared 0.0010 2.28 2.10 0.0146 0.1474 1.57 0.1853 0.46 1.3592 (5.06) 0.9446 (8.99) (23.45) 0.6598 (55.96) 5.6358 Mean = 8.9477 Standard deviation = $2.9900 Source: fP Morgan 1997, CreditMetrics™—Technical Document, www.riskmetrics.com. The standard deviation is the stand-alone credit risk of the BBB-rated bond. It is the unexpected loss of the distribution and, technically, it is the capital that a lending institution should put aside. Here, we are concerned about the capital at risk due to credit risk changes. This is measured as the amount that can be lost depending on the number of standard deviations. A 1 percent value at risk is one standard deviation (1.65), while a 5 percent value at risk is two standard deviations (2.33). (1.65 and 2.33 are the number of standard deviations under the normal curve.) In terms of the calculations, the capital at risk becomes as follows: ■ 5 percent value at risk: 1.65 X 2.99 = $4.93 ■ 1 percent value at risk: 2.33 X 2.99 = $6.97 In the next chapter, we will deal with capital adequacy. It is sufficient to say here that capital adequacy deals with unexpected losses and would require that banks put aside $8 million. (Under capital adequacy guidelines, this loan would have a 100 percent risk weighting, requiring 8 percent to be put aside as capital.) Under a more 'scientific' method, we see that less capital could be put aside. 395 Moving from the stand-alone risk to a portfolio risk becomes a more difficult proposition because we now need to deal with joint probabilities. This is beyond the scope of this text, but we can describe the problem if we imagine adding another bond to the portfolio. The following issues then need to be considered: ■ As seen in the stand-alone example, there are eight different possible values. With a second bond in the portfolio, there would be sixty-four different values (eight multiplied by eight possibilities of new values). ■ We then need to calculate from the transitory probabilities the sixty-four probabilities. Given the correlations among the credit correlations, it is not a matter of multiplying the probabilities as if they are independent. ■ We then calculate the mean and variance of the value of the distribution. As loans are added to the portfolio, the calculations become more complicated. Now we have moved through the computational issues, we can recap with the following procedure to calculate the portfolio risk of a lending portfolio for a credit-rated bond. 1. Define the portfolio as individual assets 2. For each asset, define the cashflows and calculate the present values for each potential state, using a zero coupon yield curve 3. Using a transition matrix, calculate the probability-weighted present value and standard deviation. These three steps could be construed as one stage because they calculate the stand-alone risk of each loan 4. The next stage is to calculate the portfolio risk by executing the above procedure for the joint probabilities for a loan in the portfolio to derive the standard deviation of the portfolio. While the CreditMetrics™ approach overcomes some of the stated objections to using modern portfolio theory (i.e. the assumptions of normality of distributions), it has its own concern: like many other methods, it always involves the problem of valuing bank loans. MANAGING THE PORTFOLIO In this chapter, we have moved from analysing the single loans to analysing the overall portfolio risk. The question now becomes: what does the lending manager do if he/she finds unacceptable concentrations of risk? To put this into more familiar terms, what does the portfolio manager do if he/she finds that the lending assets fail to sit on the efficient frontier? In these circumstances, the lending portfolio manager would need to shed some of the assets that cause some of the problems. For many years, the only technique was to sell the loans off the statement of financial position. Known as securitisation, this technique was limited to various selected assets. The development of the financial markets, however, has given rise to a new class of derivative instruments - credit derivatives. We will now discuss both techniques. 396 Securitisation Securitisation is a method of packaging the cashflow from an asset into an investment (not collateral) security and selling it to investors. The most recognised securitisation structures are those that package home loans, but many other structures are available. These include packaging cashflows from: ■ utility bills such as electricity and water bills ■ royalties such as those of recording artists • car loans and leases ■ rentals from large, well-tenanted commercial buildings. Financial institutions securitise for different reasons and are not restricted to credit risk management. This is particularly the case in Australia, although lenders in countries such as the United States use securitisation or similar techniques such as loan sales to remove risk concentration. In Australia, securitisation is also used for: ■ capital management ■ liquidity management ■ interest rate risk management. The usual reasons for securitisation in Australia are capital and liquidity management. Note that securitisation, contrary to appearances, does not mean selling the lending off the statement of financial position. The two main types of securitisation structure are: 1. pass through structures, and 2. pay through structures. The type of structure used depends on the purpose of securitising lending assets. Before discussing these structures, it is important to recognise the characteristics of the following types of asset that can be securitised: ■ The assets must have high-quality cashflows; in other words, there must be a low probability of default. This makes home loans an attractive asset because they have historically low default rates. ■ The lending assets have to be homogenous, which means they have the same risk profiles. Again, home loans have a similar risk profiles. ■ Credit rating agencies also impose conditions on structures when they apply ratings. ■ Individual lending assets must be seasoned (which usually means that at least one repayment has been made) and must have insurance (which usually means, for home loans, that the loan is mortgage insured). ■ On a portfolio basis, structures will normally be penalised if there is geographic risk. Pass through structures The major characteristic of pass through structures is that the lending assets are completely sold off the statement of financial position. Those financial institutions that are using securitisation 397 for capital management purposes favour these structures, which have the following component (Figure 11.5): ■ the owner of the assets ■ a special-purpose vehicle through which the assets are sold ■ a trustee/manager that manages the assets and their cashflows. This involves receiving the cashflow from the lending assets—for example, the principal and interest repayments that would be received via the lending institution—and passing them onto the investor. The trustee/manager is also responsible for investing surplus cashflow received to maintain the value of the asset. ■ investors who buy the security. In terms of the asset pool, there are two types of structure: 1. static pools 2. dynamic pools. A static pool has a fixed number of assets in the pool. To maintain the value of the pool, the pool is either insured for the value or overcollaterised. Overcollaterisation means a greater value of assets is put into the pool than the value for which the pool is to be sold. This is obviously at a penalty to the lender wishing to securitise assets. Both measures are important to investors, who would be concerned that loans would lose their value through pre-payments or default. Figure 11.5 Process for pass through structures Dynamic pools have assets with maturities that are shorter than the actual securitisation structure. As cashflows come in from the lending asset, the trustee/manager re-invests the proceeds to maintain the value of the pool. This often requires the trustee/manager to carry out risk management or purchase guaranteed investment contracts (known as GICs in the market), which guarantee a yield. Pay through structures Pay through structures are not much different from pass through structures, particularly in terms of the cashflows. The one major difference is that the lending institution does not sell the 398 assets off the statement of financial position; rather, it packages the cashflows into the special­ purpose vehicle. Pay through structures are more popular with institutions that are not regulated by the Australian Prudential Regulation Authority. The reason is that if the purpose of the securitisation were capital management, then the regulator would not consider the asset as having been removed from the statement of financial position using this structure. The most popular assets that are securitised through these structures are credit card receivables from retailers and car lease receivables from nonbank lenders. Securitisation and credit risk management The question now becomes: how does securitisation help credit risk management? The most obvious purpose for using securitisation for credit risk management is to sell off concentrations of credit risk. Financial institutions, having identified a concentration of lending assets, can sell off these assets through securitisation vehicles. At this stage, the assets that can be securitised are limited to housing loans, credit card receivables and car lease receivables, but the overseas experience reminds US that the variety of assets that can be securities will slowly rise over time. Three other issues need to be kept in mind when securitising via pass through structures: 1. 2. 3. It is important that there is no recourse, either legal or moral, to the lender if the securitised loan defaults. The Australian Prudential Regulation Authority has guidelines regarding recourse. Apart from the regulatory requirements, the lending institution will not have removed the credit risk if investors are able to claim recourse. An often-overlooked issue is that when credit risk is sold off, fair value for the credit should be received. This may seem to be a liquidity issue, but it becomes a credit issue because the reward-risk issue arises when assets are sold at deep discounts. Securitisation also brings up a relationship issue for the lender and borrower. When a lending asset is securitised, the institution may be required to inform the borrower that its loan is being sold. This may be viewed negatively by the borrower, particularly in the corporate lending market. Much of this problem is overcome by assigning the asset to special-purpose vehicles, which means that the title remains with the lending institution. There is a growing tendency in the corporate lending markets to use credit derivatives to lay off credit risk, because this does not disturb the relationship between the lending institution and the borrower. We will now deal with these instruments. Credit derivatives Credit derivatives are a set of financial instruments that allowparticipants in the financial markets to either assume or remove credit risk to a portfolio without buying or selling the lending asset. Credit derivatives have a major benefit over securitisation. With most securitisation structures, the lender loses the relationship with the borrower, which is unacceptable with many loans, 399 particularly with corporate customers. Credit derivatives allow the removal of credit risk without breaking the relationship. The financial institution that lends money but wishes to lay off the credit risk is known as the protection buyer, while the institution that assumes the credit risk is known as the protection seller. In many respects, the financial institution is insuring the statement of financial position against credit risk, much like insuring the statement of financial position against changes in interest rates. Many credit derivatives have a corollary in the interest rate derivatives market. There are many different credit derivatives, but they fall into three main categories: (1) credit default swaps; (2) total return swaps; and (3) credit options. A credit default swap acts like an insurance policy: for a periodic fee, a lender can hedge a loan with a protection seller, who would pay an agreed amount on the instigation of a credit event. The types of credit event covered are normally negotiated between the protection seller and buyer, and could include: ■ bankruptcy ■ credit rating change ■ capital structure changes ■ default on a loan ■ changes in credit spreads above an agreed level. Figure 11.6 shows how the credit default swap works. The residual payment is normally defined as the face value of the lending asset less the market value (or recoverable value) of the loan. Figure 11.6 Credit default swap While the credit default swap structure looks like an insurance policy, the total return swap looks more like an interest rate swap, as shown in Figure 11.7. Under this swap, the protection seller pays the protection buyer any losses in market value on the reference loan (or, in the case of the market value of the asset increasing, the protection seller receives the increased value). In return, the protection buyer pays a funding rate such as the London Inter Bank Offer Rate (LIBOR) or BBSW.1 The asset on the statement of financial position of the financial institution maintains its value regardless of any change in credit rating (up or down). 1 BBW is a page on the Reuters information system that provides average bank bill rates over a range of maturities to 180 days. 400 Figure 11.7 Total return swap Some credit options operate as normal options where the strike price is based on credit spreads widening. They are not unlike a credit default swap, except the fee is paid in full at the beginning of the derivative rather than over the life of the instrument. If the fee is paid upfront, then the derivative is generally referred to as an option; otherwise, it is a swap. The protection sellers tend to be a limited number of institutions, mainly banks (both domestic and international). In offshore markets, many insurance companies are now protection sellers, gaining diversification benefits by moving into new markets. Australia is expected to move in the same direction. In the near term, however, given that banks are using only credit rated bonds, the market is expected to be limited as financial institutions attempt to lay off the same credit risk. This will be a problem until the market deepens. The following are among the other major problems: ■ Good-quality pricing data are not available. • The pricing models have not been tested during periods of financial distress. ■ Where numerous credit derivatives have been written on a particular asset, if that asset defaults, then the value could paradoxically be higher than the default value because the asset (loan) would be deliverable. ■ Credit derivatives, operating in an illiquid market, cannot be easily valued. ■ With the risk now removed from the statement of financial position, the incentive for the financial institution to monitor the reference asset is diminished. ■ At this stage, the Australian Prudential Regulation Authority penalises any hedge using a credit derivative if it does not match the maturity of the underlying reference asset. Many protection sellers, being risk averse, will write protections for maturities less than those of the reference assets. INDUSTRY INDIGHT Banks' ability to assess iendsng risk A New York-based risk analysis company has found that banks in Australia are less proficient than their American counterparts in being able to assess commercial lending risks. The analysis of loan book samples at two of the five biggest Australian banks by Zeta Financial Services showed that up to 17 percent of their loans carried a high propensity to default in the next five years. 401 Zeta Financial Services is the Australian arm of Zeta Services Inc., which has been working in the area of credit risk rating for the past seventeen years. Its software is now used in 30 percent of the top fifty US banks and Zeta is a consultant to the US Federal Reserve Board. A director of Zeta Financial Services, Mr Graham Soper, said the sample studies conducted by his firm had highlighted the need for banks to have an independent assessment of their loan books. He said the sample survey results showed a major flaw in the Reserve Bank's proposed new regulatory requirement that chief executive officers of banks sign off on the adequacy of internal risk management systems. bank management is going to sign off a declaration that their risk management systems are inadequate', he said. ‘No ‘What is needed is an external' objective test of their credit risk management. 'The auditors can't do it because they don't have the software to be able to take a sample of a loan book and give them a rating.' Mr Soper, who spent twenty years in insolvency and corporate reconstruction work before starting a company called Corporate ScoreCard' said much had been written about the risks to banks from derivatives while ignoring the dangers from 'the common old garden corporate loan'. seem to have ignored the enormous potential for banks to lose money in corporate lending', he said... the top six banks alone wrote off SI7.45 billion between 1991 and 1994, and hit a peak of bad loans in 1992 of $25.7 billion. ‘People Mr Soper said the pilot studies of two of the top five Australian banks had found that only 65 percent of loans matched the Zeta credit risk ratings whereas in the United States 70 percent of loans matched the Zeta ratings. Of the remainder of the loans in the Australian samples, half were rated higher than the Zeta ratings and half were lower. • The chairman of Zeta, Mr Brian Wright, said banks that failed to assess accurately the credit risk of customers ran the risk of either overcharging or undercharging for risk. Mr Wright' who had a long experience of dealing with small to medium-sized enterprises [SMEs] when he was head of the Commonwealth Development Bank' said Zeta's work had shown that up to 15 percent of SMEs were far stronger than the banks thought they were. Zeta was started in 1979 by Professor Ed Altman and Mr Bob Haldeman. Professor Altman, who is professor of finance and chairman of the MBA program at the New York University Business School' developed the idea that default or credit risk could be measured directly from financial statement data. Zeta Services Inc. in New York licensed the Australian and New Zealand rights to its software in 1993. Source: T Boyd 1995, ‘The art of picking losers', Australian Financial Review, 6 November, p. 28. 402 The above article highlights the veracity of the credit risk management system. It makes a very important point: differing credit risk management systems provide different answers. In this case, the divergence is between in-house systems designed by lending institutions and those available 'off the shelf ’. Is the comparison valid? It depends. Systems designed in-house have the advantage of including those characteristics peculiar to the lending institution. They can be difficult to keep current, however, in terms of financial market developments. This iầ where off-the-shelf systems are valuable. There are two points to the article. First, how are lending institutions kept accountable for the way in which they manage credit risk? Second, the article implies that banks may experience bad debts five years from 1995, and we are now starting to observe spectacular corporate collapses that affect lending institutions. LOAN PRICING The discussion on credit risk analysis and portfolio management would be incomplete without a discussion of the elements of loan pricing. We mentioned earlier in the chapter- that lending as a whole rarely lies on the efficient frontier, inferring that the interest rate charged on loans does not reflect its risk. This does not necessarily reflect all lending products, but portfolios as a whole. In this section, we will highlight the elements that should be considered in loan pricing: 1. 2. costs of the statement of financial position non-credit risk costs. Costs of the statement of financial position Given that the loan is funded from the statement of financial position, a number of costs that are generated from the statement of financial position need to be considered: 1. capital costs 2. liquidity costs 3. the cost of funds. Capital costs Lending institutions charge capital to loans using differing methods. As a base method, they would set aside capital based on the capital adequacy guidelines. This would mean that some capital would need to be used when funding a loan. Investors would require a return on the equity they provide. This cost to the lending institution is determined by the board of directors. Liquidity costs The element of liquidity is often forgotten in the lending equation. While the Australian Prudential Regulation Authority has relaxed the formal guidelines for liquidity management, a prudent lending institution would hold a portion of its assets in liquid securities or cash. This 403 approach implies that liquidity should be held against every lending asset, depending on the policy. Cost of funds When we take the above elements into account, we can calculate the cost of funds. The required capital and deposits fund the lending and liquid assets. The following returns would be fixed or set by the market: ■ The board sets the return on equity. ■ The market sets the return on liquidity. ■ The market also sets the cost of deposits. ■ The only variable is the return on the loan, which is the balancing figure. This should become clearer in an example later in the section. Non-credit risk costs Lending also involves risks that are not directly related to the statement of financial position but affect the cost of the loan and thus its price. These risks include: 1. 2. 3. interest rate risk pre-payment risk origination costs. The first two risk types occur generally when the loan is a fixed interest loan, while the third occurs with all loans. Interest rate risk Many financial institutions provide the majority of their loans on a variable basis and fund these loans with at-call deposits. As interest rates rise, many borrowers tend to want to switch to fixed interest rate loans. In a rising interest rate environment, the costs of deposits also rise. If the lending institution does not take this into account, then a loss could arise. There are a number of solutions to this scenario. The institution can fund the loan from term deposits (or similar instruments) of the same maturity, obviating the interest rate risk. The longer the maturity, however, the more difficult it is to raise term deposits. If term deposits can be raised, then that would be the cost of these funds. If funds cannot be raised, then the alternative would be to execute a derivative that would fix the rate. Likewise, the rate at which the derivative is fixed would be the cost of funds. Pre-payment risk The opposite situation to interest rate risk is pre-payment risk. Again, this is a risk only for fixed rate loans, but in this instance when interest rates fall. When interest rates fall, many borrowers of fixed rate loans seek to refinance their loans at lower rates. If successful, the financial institution 404 incurs a loss because it is unable to reinvest the funds at the same rate. Further, if the cost of funds has been fixed by term deposits or a derivative, then this arrangement might have resulted in a relatively high cost of funds. There are two alternative solutions to this phenomenon. The lending institution can estimate the average cost of pre-payment risk and add this cost to the loan (although this may make the loan uncompetitive). More often, the lending institution specifies a penalty that should cover the majority (if not all) of the costs. This penalty is easy to calculate and easy to understand. The economic solution, however, is to calculate the present value of the cashflow on pre­ payment, with the lending institution receiving a penalty or paying a benefit depending on the interest rate of the loan and current interest rates. Originating and operating costs The cost of marketing and then monitoring the loan needs to be incorporated into the loan. The more complex and risky the loan, the more monitoring it requires. This effort should result in a higher interest rate being applied. Credit costs We have deliberately not addressed credit costs first. Hopefully, you will observe that the margin over base rates is more than simply credit risk. Credit risk, nevertheless, makes up an important element in loan pricing. The two elements in default risk pricing are: 1. expected losses 2. unexpected losses. Expected losses Lending institutions always expect some loans to default. We cannot predict the future, and a loan that seems good today can default later as a result of unforeseen events. Lending institutions, given their experiences, expect some loans to default. They can price their expected losses into the loan price using the following formula: Expected losses = Default probability X (1 - Recovery rate) Both the default probability and recovery rate for the loan type would be determined from the lending institution's experience (or, in some cases, credit rating agencies). Unexpected losses It is more difficult to deal with unexpected losses. These losses are generally said to reflect the volatility of expected losses and thus change from period to period. The tail of the CreditMetrics™ distribution shown in Figure 11.3 could indicate unexpected losses. Unexpected losses cannot be priced. The correct way to deal with them is to set capital aside (much like capital adequacy) and price unexpected capital as a return-on-equity issue. 405 Loan pricing - an example Now we have identified all the elements of a loan, we can consider the following example. We will also examine what occurs when the wrong relationships are assumed. The price of a $150 000 five-year housing loan needs to be considered. The lending institution determines that there needs to be 5 percent liquidity against lending assets, which currently earn 4.9 percent. At-call deposits are priced at 3.5 percent and five-year swap rates are at 5 percent. The annual cost of operating and managing the loan is $1000. The lending institution’s default probability for housing loans is 2 percent, while its recovery rate is 95 percent. It considers its unexpected losses in terms of losses that are outside the capital put aside under capital adequacy guidelines. Home loans have a small probability of loss and the lending institution puts an extra 1 percent aside. To deal with pre-payment risk, the bank charges an extra 30 basis points. What would the price of the loan be if investors require an after-tax return on equity of 20 percent and have a tax rate of 30 percent? It is important to view the pricing of the loan in stages. Our first step is to calculate the return on equity. Under capital adequacy guidelines, the amount of capital required is: $150 000 X 8% X 50% = $6000 The above 50 percent is as per the capital adequacy guidelines. The after-tax return on equity becomes: $6000 x 20% = $1200 We can also work out the amount of liquid assets, using simple algebra required under the 5 percent policy, as shown on the next page. Liquid assets = Assets X 5% . where Assets = (Loans + Liquid assets) = ($150 000 + Liquid assets) Liquid assets = ($150 000 + Liquid assets) X 5% = $7500 + 0.05 Liquid assets Liquid assets - 0.05 Liquid assets = $7500 0.95 Liquid assets = $7500 $7500 Liquid assets ■ • ■ ■— = $7895 4 0.91 We can now construct a first-stage simple statement of financial position as follows. 406 Table 11.10 Statement of financial position Assets $150 000 Loan Liquid assets Total $7 895 $157895 Liabilities Deposits Equity Total $151 895 $6 000 $157 895 You will note that the deposits are the balancing figure. Given that we know all the costs above (except the loan price), we can make the first calculation, using at-call deposits. Tabid l.ll Fừst calculations Loan Liquid assets Less interest on deposits Profit before tax Financial performance ($) 6 643 .74 386.86 5 316.32 1 714.28 Less tax at 30 percent Profit after tax 1 514.28 200.00 Yield (%) 4.43 4.9 3.5 Balance ($) 150 000 7 895 151 895 20.0 6 000 The balancing number is loan financial performance, from which we can infer the loan rate. It is not immediately obvious that to find the answer we need to work backwards from the profit after tax financial performance to obtain the balancing figure of $6 643.74, which gives a yield of 4.43 percent. The other issue to be considered here is that the loan is fixed for five years. If interest rates rise, then investors will not obtain their 20 percent return on equity. In theory, lending institutions should fund these Ioans at the five-year rate and transfer price the deposit at this rate, as we do below in Table 11.12. In practice, this is difficult to do. Table 11.13 shows the effect. Table 11.12 First calculations Loan Liquid assets Less interest on deposits Profit before tax Financial performance ($) 8 92217 386.86 7 594.75 1 714.28 Less tax at 30 percent Profit after tax 1 514.28 200.00 Yield (%) 5.95 4.9 5.0 Balance ($) 150 000 7 895 151 895 20.0 6 000 407 Table 11.13 The effect Loan Liquid assets Charge for pre-payment risk Less interest on deposits Less monitoring cost Less expected losses Profit before tax Financial performance ($) 10 050.75 386.86 450.00 7 594.75 1 000.00 150.00 2 142.86 Less tax at 30 percent Profit after tax 1 Yield (%) 6.70 4.90 0.30 5.00 Balance ($) 150 000 7 895 150 000 151 895 0.001 150 000 20.00 7 500 642.86 500.00 We can now deal with the remaining issues and complete a new table (as above). ■ The cost of monitoring is $ 1000. ■ Expected losses are as follows: Expected loss = Default probability X (1 - Recovery rate) ■ =2% X (1 - 95%) = 0.001 ■ Extra capital of 1 percent changes the $6000 capital to $7500. ■ There is a 30 basis point charge for pre-payment risk. Practical loan pricing If this theoretical approach were adopted, then the price of loans would undoubtedly be higher than what is observed in the market. Why? There are two major reasons: 1. Competition often pushes down the price of loans. 2. Lending institutions price the borrower on the whole connection. The fees on one facility (such as transaction fees on cheque accounts) are often expected to offset a lower interest rate. Lending institutions often lose money using this’scenario because some of the expected income does not materialise. Lending institutions need to exercise caution so their good credits do not subsidise the lesser credits. A DAY IN THE LIFE OF... A credit risk analyst 8.15 a.m. I work for the Commonwealth Bank Group as a credit risk analyst. I read the Australian Financial Review and Sydney Morning Herald as a daily routine. 408 9.00 a.m. A submission for the April meeting of the bank's risk committee is due today. The risk committee is chaired by Mr David Murray and convened every second month. Issues discussed at the meeting arise from the challenges of integrated risks (credit, operational, market and liquidity) that the bank is facing in the short to medium term. My part of the submission concerns the concentration level of the bank's lending to the Australian commercial portfolio, as well as the quantitative measure of the risk-return profile of each industry grouping in the Australian commercial portfolio. (The return measure is the Sharpe ratio and the risk measure is the portfolio unexpected loss. At individual asset level, risk is measured by the contribution of risk to the portfolio unexpected loss, accounting for the diversification of the portfolio.) The submission is a summary result of a routinely run model, which takes two to three weeks to process. The model produces vast amounts of information but space reserved in the risk committee submission is normally 300 words and one attachment. 10.00 a.m. attend a portfolio management team meeting on the subject of portfolio stress testing. I have to break off work on the risk committee submission because the meeting is pre-arranged. The executive credit committee and risk committee requested the portfolio management team to perform portfolio stress testing at the middle of 2001, in the wave of apparent credit down cycle and big corporate collapses, which caused the bank to increase its loan loss provision substantially. This task was given more priority as the result of the 11 September 2001 events in the United States. The format of the meeting is brainstorming. One team member is not entirely happy with the current method of using a credit migration matrix derived from the Early Warning System (a forecast model that I run quarterly, to predict the portfolio expected loss in one year), because gradual migration does not constitute a stress event. Echoing the concern from the capital management team, one other team member questions the validity of stress test in reference to its usefulness in capital allocation. He argues that the stress test should only help the senior management to understand the risks faced by the bank. The chief manager of the portfolio management team decides to continue the current method subject to further discussion. I The official meeting finishes at 11.00 a.m. I spend the next half hour talking about the weekend footy results, the euphoria of the Corporate Games and gossip such asjustin Timberlake dumping Britney Spears. 1 finish work on the risk committee submission by 12.30 p.m. I have lunch at my desk, while surfing on the Internet, then walk to the Domain after lunch. 2 p.m. deal with an e-mail from the Credit Management Unit at the Customer Services Division regarding the impact of Colonial integration on its agriculture portfolio. It is putting together a working paper to the board, outlining the current status of agribusiness in Australia and the bank's strategy in dealing with the new challenges in this area. It is concerned with the current economic environment and the potential negative impact on its agriculture portfolio. I 409 3 p.m. attend a scheduled meeting with two consultants from Risk Solution of Standard & Poor’s regarding the proposal to build a national database on loss given default. The primary purpose of this database is to make the Australian Prudential Regulation Authority comfortable and may be used to satisfy the Basel II requirement for advanced status. Today’s meeting is the follow-up of a Melbourne meeting of the major banks. A further meeting is scheduled to customise the database structure and definitions. The meeting finishes in just under one hour. I 4 p.m. I receive a telephone call from the financial control of institution banking to voice concern with the assignment of R-square (a correlation parameter) to the business-in-government sector in the running of KMV Portfolio Manager™. A meeting is arranged for 10.30 a.m. on Thursday. I have a chat with my work partner about a wide range of subjects, while flipping through some industry/economic magazines. 4.45 p.m. Just as I am getting ready to go home, I receive a request to assess a syndicated credit from the office of the bank’s chief credit officer. The portfolio management team does not assess credit generically (that is, on an individual client’s credit quality); rather, we put the applicant into a subportfolio and examine its impact in terms of change to the risk-return profile of the portfolio. It takes 45 minutes to finish trimming the relevant data, which are then input into our portfolio model. I run the subportfolio overnight. Analysis of the preliminary result is the work of another day. Source: Mr Harvey Yu, Portfolio Analyst, Portfolio Management, Group Risk Management, Commonwealth Bank of Australia, 2001. It can be easy to think that the task of a credit portfolio manager is boring and mechanical. Most of the tasks carried out by someone in this position are involved and time consuming. A particular problem is that requests can appear 'out of left field'. Given that an incorrect credit decision may have an adverse impact on the lender's profits, the credit portfolio manager’s task requires insight and accuracy of judgement. SUMMARY 1. What is the main benefit of credit risk management? The major benefit of credit risk management is that it removes the subjectivity from lending decisions. Ultimately, loans can be assessed on a scientific rather than human basis. 410 2. What is Altman’s z score? How is it used? Altman used multidiscriminant analysis to distinguish between two populations: good borrowers and bad borrowers. A score is constructed using accounting ratios, then this score is measured against a benchmark. 3. How can stock prices be used to explain credit risk? The problem with Altman’s z score is that it is considered by many to be backward looking. KMV Corporation, via its expected default frequency method, used share prices to predict default. This assumes that share prices imply all future possibilities regarding the share. 4. Explain how actuarial approach can be used by examining CreditRisk+ This approach does not take into account the reason for default but uses an actuarial approach to credit risk 5. Explain how a macroprudential approach can be used by looking at Creditportfolio view This approach uses a macroeconomic approach which connects credit risk to the business cycle. 6. How can risk-adjusted return on capital be used for portfolio purposes? Given that loans require capital, it has been suggested that a portfolio could be constructed on the basis of the return on capital adjusted for risk. The method also recognises that capital is a finite resource and marginal capital can be used to fund a loan. 7. How can the Sharpe Index be used for lending purposes? The Sharpe Index measures the risk-adjusted return, like the return on equity method, using traditional portfolio management. Lenders can maximise their returns using the method, with suitable constraints. 8. Why is CreditMetrics™ useful for calculating the risk of a loan portfolio? Again, the Sharpe Index can be considered to have problems because some of the assumptions do not equate well to lending. CreditMetrics™ recognises this problem, as well as the fact that credit risk is not static. 9. What are the elements of loan pricing? The elements of loan pricing include the consideration of costs of the statement of financial position, as well as credit risk costs and costs that are not of the statement of financial position. 411 DISCUSSION QUESTIONS 1. 2. 3. Outline the problems of traditional lending methods and possible solutions. Are there any problems with the solutions? Compare and contrast the approach of the score model and the KMV expected default frequency model. Use the following extracts from the Harvey Norman 2000 annual report to calculate the Altman score. z z Statement of financial position as at 30 June 2000 Consolidated 2000 ($’000) Parent entity 1999 ($’000) 2000 ($’000) 1999 ($’000) ■ ! Current assets 37 385 3 147 - 476 077 358 477 151 669 61 001 24 599 - 13 552 6 - 588 015 9616 395 839 151 675 174 576 Receivables 9 067 8514 - ị Investments 10 396 37 881 55 596 55 592 ị Property, plant and equipment 547 129 388 560 - > Intangibles 590 — - J Other Total noncurrent assets 3 747 1 131 950 569 749 438 702 56 727 157 764 834 541 208 402 56 542 231 118 312 124 216 373 77 64 ị Cash ị I Receivables Inventories • Other • J Total current assets Noncurrent assets 1 J ị ị Total assets Current liabilities ị Accounts payable . .. . 2 567 1 • 174 576 ; 1 Borrowings 33 591 12 401 - - Ị Provisions 58 115 48 547 21 517 26 085 403 830 277 321 21 594 26 149 1 Total current liabilities 412 Parent entity Consolidated ■ 2000 ($’000) 2000 ($’000) 1999 ($’000) 1999 ($’000) ; ị Noncurrent liabilities Borrowings j Provisions 1 Total noncurrent liabilities 152 151 - 388 238 - 203 608 152 389 - 21 594 186 808 26 149 i 204 969 ị ff 142 869 ; Total liabilities 607 438 429 710 1 Net assets 550 326 404 831 i Shareholders' equity Share capital Reserves Retained profits ; Total shareholders' equity _ Ị 203 220 i 1 - ! — ■Íi I 187 792 142 869 142 869 83 551 58 614 - 278 983 203 348 43 939 550 326 404 831 186 808 i i Ị — it 62 100 J 204 969 ỉ Statement of financial performance for the year ended 30 June 2000 J 1 ““ Ị i ; Operating profit before abnormal item ■ Abnormal item ; Operating profit before income tax ! Income tax attributable to operating profit ; Operating profit after income tax Retained profits at the beginning of the financial year ! Total available for appropriation ■ ị Dividends provided for or paid ; Retained profits at the end of the financial year ! Basic earnings per share (cents per share) ị Parent entity Consolidated ị 2000 ($’000) 1999 ($’000) 2000 ($’000); 173 897 136 843 - 7 374 28 881 ị — Ịi 173 897 129 469 62 626 49 344 111 271 80 125 203 348 153 895 17 475 ; 62 100 í 314619 234 020 79 575 ị 35 636 30 672 35 636 • 278 983 203 348 43 939' 10.83 7.89 28 881 ỉ 11 406 • -■ f Source: Harvey Noman 2000, Haney Norman Holdings Limited and its Controlled Entities—Annual Report 2000. 413 Would you lend to Harvey Norman? What is the difficulty in using the Altman z score? 1. Under what circumstances does KMV's expected default frequency model not work correctly? 2. Explain the limitations of the concept of the risk-adjusted return on capital. 3. What financial basis does Altman use to construct his portfolio management model? Why does he use constraints in maximising the return on the portfolio? 4. In the example given for CreditMetrics, we calculated the stand-alone risk for a BBBrated bond. Using the same data, calculate the stand-alone risk of a five-year AA-rated bond with a coupon of 5 percent. Is there anything about your answer that you find unusual? 5. Explain the circumstances in which you would use securitisation and the circumstances in which you would use credit derivatives. 6. Is securitisation a credit risk management tool? 7. If a protection seller under credit derivatives is assuming the risk of a loan, why does the protection seller not just provide the loan? REFERENCES AND FURTHER READING Altman E 1968, 'Financial ratios and discriminant analysis and prediction of corporate bankruptcy’, Jour­ nal of Finance, 23, pp. 189-209. Altman, El 2000, 'Predicting financial distress of companies: revisiting the Z-score and Zeta models', New York University Working Paper, New York. Beaver, w 1967' 'Financial ratios as predictors of failure’, Journal of Accounting Research. Caoeutte, J' Altman, E & Narayanan, p 1998' Managing Credit Risk, John Wiley & Sons, Toronto. Croughy, M Galai' D & Mark, R 2000' 'A comparative analysis of current risk models’, Journal of Banking and Finance, 24' pp. 59-117 Crosbie, p. & Bohn, JR 2001, 'Modelling default risk’, www.moodyskmv.com , accessed May 2001. Morgan JP 1997, CreditMetric™—Technical Document, www.riskmetrics.com. Saunders, A 1999, Credit Risk Measurement, John Wiley & Sons, Toronto. White, G' Sondhi' A & Fried' D 1998' The Analysis and Use of Financial Statements, John Wiley & Sons, To­ ronto. CREDIT RISK FROM THE REGULATOR'S PERSPECTIVE LEARNING OBJECTIVES By the end of this chapter you should be able to: 1. understand the issues of credit risk from the perspective of the regulators 2. relate capital adequacy to credit risk considerations 3. express the issues of large exposures 4. identify securitisation issues for regulators 5. identify credit derivative issues for regulators 6. describe the credit rating process 7. discuss the new capital adequacy guidelines. KEY TERMS Basel II clean sale moral hazard tier 1 capital Basel III credit rating prudential regulation tier 2 capital 415 capital adequacy large exposures recourse 416 INTRODUCTION Until the 1980s, the then regulator (the Reserve Bank of Australia) had a large role to play­ in credit risk issues. The regulator would often dictate maximum lending rates and direct financial institutions to sectors of the economy that the government considered needed lending support. While it is difficult to assess the effects of these actions, the following conclusions can be drawn: * The sectors that were not specified for lending assistance would have been subject to credit rationing. ■ Lending institutions might have been subject to concentration risk. • Given the directives of the regulator, banks might have approved marginal lending applications in the directed sectors. From the 1980s, however, successive governments dismantled this regime and moved to prudential regulation. The strictures on interest rate ceilings and lending strictures were removed and replaced with the less prescriptive requirements of capital adequacy, large exposure, concentration risk, bad debts and credit issues for securitisation as well as credit derivatives. Regulation was simply established to reduce the possibility of insolvency of financial institutions, with the onus being on lending institutions to comply. The global financial crisis has accelerated prudential regulation reform. Banks now need to consider compliance with Basel III requirements as well as meet the requirements of credit licensing under the National Consumer Credit Protection Act. In terms of Basel III, the main focus is on liquidity risk and therefore not in the purview of this text. However, the guidelines do address the treatment of bad debts, and that is commented on later in this chapter. Since 1998, the regulator for credit risk issues for financial institutions in Australia has been the Australian Prudential Regulation Authority, which supervises institutions known as approved deposit-taking institutions (ADIs). ADIs include banks, building societies and credit unions. Most issues canvassed in this section are based on standards or guidelines that are maintained by the authority. The Australian Prudential Regulation Authority does not directly address credit risk. Given that capital adequacy has become the catch-all for credit risk, however, this discussion will focus on the way in which the authority addresses credit matters via capital adequacy. Capital requirement is an important tool used by regulators to assess credit risk, so it is important that you have full understanding of the present capital adequacy regulation. Australian Securities Investments Commission (ASIC) The failure of some banks to manage their retail lending activities properly during the global financial crisis has hastened the Federal Government to adopt the National Consumer Credit Protection Act to protect retail borrowers. The main points of this legislation address: 417 1. 2. 3. 4. licensing by ASIC, the rights of the borrower, the obligations of the lender, and the nature of the contracts. These laws are designed to protect the consumer from undertaking loans that they essentially or potentially cannot pay back. Banks are prohibited from creating onerous conditions for their borrowers. CAPITAL ADEQUACY Note: this chapter has been written during the transition of Basel 2 to Basel 3. While there are dates that this transition will be completed, given events in other jurisdictions such as Europe, the final transition is unclear. Recent history has shown that financial institutions normally fail as a result of poor credit decisions. The responsibility for credit decision-making rests with the board of directors of a financial institution. Given that shareholders elect the board, this means that shareholders explicitly assume the credit risk decisions of the institution. Credit risk would not be a concern if the bank's board and shareholders invested only their own funds, but the board invests the funds of other people. These 'other people' are depositors, who keep funds in banks but have no control over the way in which these funds are invested. The control is ultimately with the board and senior management, and decisions could be taken that are not in the interest of the depositors. This is known as moral hazard. Depositors need to be protected against poor decisions by management— this realisation led to the birth of capital adequacy regulations. Under the base capital adequacy, financial institutions are required to put aside capital for each credit risk exposure, whether it is on or off the statement of financial position. Under capital adequacy, financial institutions are required to put aside a minimum of 10.5% plus a countercyclical buffer of up to 0.25% of capital for risk-weighted assets. (If the Australian Prudential Regulation Authority believes that a financial institution has a high risk profile, then it will impose a higher benchmark.) This benchmark may be increased when Basel III is finally adopted. The term 'risk-weighted assets’ needs explanation. Under the capital adequacy regulation, certain types of loan are considered to be safer or less risky than others; for example, capital adequacy considers home loans to be less risky than business loans. (Note that this is a generalisation and the reverse could easily be true in specific situations.) The risk weight or the capital that needs to be put aside for a housing loan will therefore be less than that for a business loan for a similar amount. The Basel Committee of the Bank for International Settlements decides the risk weight for each type of asset. These recommendations have been mostly adopted in Australia and elsewhere in the world. 418 Under Basel II, prudential regulation is split into three pillars. Pillar 1 deals with the capital that needs to be put aside for credit, market and operational risk. Under this pillar, there are a number of methodologies to calculate credit risk. For the purposes of this chapter we look at the standardised version and touch on other methodologies in the chapter on quantitative finance (Chapter 16). Basel III is a direct response to the Global Finance Crisis of 2007 and 2008. While the major issue of this crisis was not capital rather than liquidity, capital has been strengthened. Pillar 2 is essentially a catch-all of all other risks and is managed under the process known as the Internal Capital Adequacy Assessment Process (ICAAP). ICAAP is a document that outlines all the risks outside Pillar 1, and how they are identified, measured and assessed for capital requirements. It is a mistake to think that it has no credit implications whatsoever. The most obvious credit risk that should be assessed under Pillar 2 is concentration risk. Concentration risk is the risk created by lending too much to either a counterparty or sector. Concentration risk is difficult to measure, and Basel II has some controversial assumptions as well. These issues are dealt with later under quantitative finance. Pillar 3 is known as market discipline and refers to transparency in reporting. This is an all encompassing requirement, but its importance can be seen in that banks now provide reports on capital adequacy and impaired loans. The formula for capital adequacy is: , Total capital (Tier 1+Tier 2) Risk-based capital ratio = ------ Z7-"—— ——————-------r Risk-adjusted assets As of 2016, this ratio is 10.5% Capital is much more than ordinary equity and includes instruments that perform like equity. Tier 1 capital exhibits permanence like capital, while tier 2 capital has the ability to absorb credit losses but is not permanent. Tier 2 capital cannot make up more than 50 percent of overall allowable capital. Table 12.1 indicates the various capital classes. Common Equity Tier 1 Capital consists of the sum ofl: (a) paid-up ordinary shares issued by an ADI that meet the criteria in Attachment B; (b) mutual equity interests issued by a mutually owned ADI that meet the criteria in Attachment K; (c) retained earnings; (d) undistributed current year earnings (refer to paragraphs 20 to 24); (e) accumulated other comprehensive income and other disclosed reserves (refer to paragraphs 25 and 26); 1 For fuller explanations, refer to the guidelines. 419 (f) minority interests (calculated in accordance with Attachment C) arising from the issue of ordinary shares to third parties by a fully consolidated subsidiary[3] included in the Level 2 group where: (i) the shares giving rise to the minority interest would, if issued by the ADI, meet the criteria in Attachment B; and 1. (ii) the subsidiary issuing the shares is itself an ADI or an overseas deposit-taking institution that is subject to equivalent minimum prudential requirements and level of supervision as an ADI To qualify as Tier 2 Capital, an instrument must satisfy the following minimum criteria: (a) the instrument must be paid-up and the amount must be irrevocably received by the issuer; (b) the instrument represents, prior to any conversion to Common Equity Tier 1 Capital the most subordinated claim in liquidation of the issuer after Common Equity Tier 1 Capital instruments and Additional Tier 1 Capital instruments^?]; (c) the paid-up amount of the instrument, or any future payments related to the instrument, is neither secured nor covered by a guarantee of the issuer or related entity[50l, or other arrangement that legally or economically enhances the seniority of the claimt511. The instrument may not be secured or otherwise subject to netting or offset of claims on behalf of the holder or issuer of the instrument; (d) the principal amount of the instrument: (i) has a minimum maturity of at least five years; and (ii) is only recognised in Tier 2 Capital (and so in Total Capital) in the five years prior to maturity on a straight-line amortised basis (refer to paragraph 2 of this Attachment); (e) the instrument contains no step-ups or other incentives to redeem. The issuer and any other member of a group to which the issuer belongs must not create an expectation at issuance that the instrument will be bought back, redeemed or cancelled before its contractual maturity. The contractual terms of the instrument must not provide any feature that might give rise to such an expectation^; (f) the instrument may only be callable at the initiative of the issuer and only after a minimum of five years from the date on which the issuer irrevocably receives the proceeds of payment for the instrument. The issuer: (i) must receive prior written approval from APRA to exercise a call option. For instruments issued by subsidiaries not regulated by APRA included in a Level 2 group, prior written approval from APRA must also be obtained; 420 (ii) must not do anything that creates an expectation that a call will be exercised; and (iii) must not exercise a call unless: (A) the issuer, prior to or concurrent with the exercise of the call, replaces the instrument with a capital instrument of the same or better quality, and the replacement of the instrument is done at conditions that are sustainable for the income capacity of the issuer; or (B) the ADI meets the requirements relating to reductions in capital in APS 110. An instrument may provide for multiple call dates after five years. However, the specification of multiple call dates must not act to create an expectation that the instrument will be redeemed upon any call date; (g) issuers must not assume, or create market expectations, that supervisory approval will be forthcoming for the issuer to redeem, call or purchase an instrument; (h) the instrument must confer no rights on holders to accelerate the repayment of future scheduled payments (coupon or principal) except in bankruptcy (including wind-up) and liquidation. Wind-up of the ADI must be irrevocable (that is, either by way of a court order or an effective resolution by shareholders or members). The making of an application to wind-up or the appointment of a receiver, administrator, or official with similar powers, including the exercise of APRA’s powers under section 13A(1) of the Banking Act, must not be sufficient to accelerate repayment of the instrument; (i) the instrument must not provide for payment to investors other than in the form of a cash payment; (j) the instrument cannot have a credit sensitive distribution/payment feature (i.e. a distribution/payment that is based in whole or part on the credit standing of the issuer or the group or any other member of the group to which it belongs). However, an instrument may utilise a broad index as a reference rate for distribution or payments calculation purposes. Where an issuer is a reference entity in the determination of the reference rate, the reference rate must not exhibit any significant correlation with the issuer's credit standing. APRA will not allow inclusion of an instrument as part of Tier 2 Capital where it considers that the reference rate is sensitive to the credit standing of the issuer; (k) the instrument is directly issued by the issuer, and, except where otherwise permitted in this Prudential Standard, the issuer, any other member of a group to which the issuer belongs, or any related entitỵ[531, cannot have purchased or directly or indirectly[5411 funded the purchase of the instrument; (l) the instrument has no features that hinder recapitalisation of the issuer, or any other members of the group to which the issuer belongs. This includes features that require 421 the issuer to compensate investors if a new instrument is issued at a lower price during a specified timeframe; (m) where the terms of the instrument provide the ability (even in contingent circumstances) to substitute the issuer (i.e. to replace the ADI with another party), the relevant documentation must set out the mechanism to ensure that there will be a simultaneous capital injection into the ADI to replace the transferred capital instrument. Any replacement capital injection must occur at least simultaneously with the substitution and must be unconditional. The capital injection must be of equal or better quality capital and at least the same amount as the original issue, unless otherwise approved by APRA in writing; (n) the instrument does not contain any terms, covenants or restrictions that could inhibit the issuer■' s ability to be managed in a sound and prudent manner, particularly in times of financial difficulty, or restrict APRA’s ability to resolve any problems encountered by the issuer; (o) the rate of dividend or interest on the instrument, or the formulae for calculating dividend or interest payments, must be predetermined and set out in the issue documentation; (p) where an issuer defaults under the terms of the instrument, remedies available to the holders must be limited to actions for specific performance, recovery of amounts currently outstanding or the winding-up of the issuer. The amounts that may be claimed in the event that the issuer defaults may include any accrued unpaid dividends and interest, including payment of market interest on these unpaid amounts. All such unpaid dividends and interest must be subordinated to the claims of depositors and other creditors of the issuer; (q) the instrument must not provide for payment of a higher dividend or interest rate if dividend or interest payments are not made on time, or a reduced dividend or interest rate if such payments are made on time; (r) where an issue of an instrument involves the use of an SPV, the issue of the instrument is subject to Attachment I; (s) the instrument includes provisions addressing loss absorption at the point of non­ viability in accordance with Attachment J; (t) the instrument is clearly and separately disclosed in the issuer’s financial statements and, at Level 2, in any consolidated financial statements; and (u) issue documentation clearly indicates: (i) the subordinated nature of the instrument, and that neither the issuer nor the holder of the instrument is allowed to exercise any contractual rights of set-off; 422 (ii) the application of requirements relating to loss absorption at the point of non­ viability under Attachment J; and (iii) the instrument does not represent a deposit liability of an issuing ADI. The amount of the instrument eligible for inclusion in Tier 2 Capital is to be amortised on a straight-line basis at a rate of 20 per cent per annum over the last four years to maturity as follows: 2. Years to Maturity More than 4 Less than and including 4 but more than 3 Less than and including 3 but more than 2 Less than and including 2 but more than 1 Less than and including 1 Amount Eligible for Inclusion in Tier 2 Capital 100 per cent 80 per cent 60 per cent 40 per cent 20 per cent The ratio of capital for each Tier, in summary is: Common equity is 4.5 96 Tier 1 Capital is 6.0% Total capital ratio is 8% plus the conservation buffer of 2.596 of Tier 1 capital. Table 12.1 Various capital classes This is a very generalised table and more specific rules can be gained from the guidelines. In particular, there are various conditions now surrounding perpetual debt and other non­ permanent forms of equity. The amount of capital that is allocated for each loan depends on the type of loan, its risk categories and the credit rating, as perceived by the regulators. These risk categories (a selective list, based generally on loans) are as follows: Table 12.2 Risk categories Claim Credit rating grade Risk rate Class I - Cash items 1. 2. Notes and coins. All Australian dollar balances and other Australian dollar claims on the Reserve Bank of Australia 0 0 423 3. 4. 5. 6. Gold bullion held in the ADI's own vaults or on an allocated basis by another party to the extent that it is backed by gold bullion liabilities. (Gold bullion held on an unallocated basis by another party, though backed by gold liabilities, is weighted as a claim on the counterparty unless a lower risk-weight is approved in writing by APRA.) Cash items in the process of collection (e.g. cheques, drafts and other items drawn on other ADIs or overseas banks that are payable immediately upon presentation and that are in the process of collection). Class II - Claims on Australian and foreign governments and central banks All Australian dollar claims on the Australian Government. Claims on overseas central governments and state or regional governments, State or Territory Governments in Australia (including State or Territory central borrowing authorities), central banks (including the Reserve Bank of Australia) and foreign currency claims on the Australian Government.... 0 20 1 2 3 4.5 6 Unrated 13. Class VI - Past due claims The unsecured portion of any claim ... that is past due for more than 90 days and/or impaừed: a) 14. 15. 16. 0 0 20 50 100 150 100 where specific provisions are less than 20 percent of the outstanding amount of the past due claim or impaired asset; or b) where specific provisions are no less than 20 percent of the outstanding amount of the past due claim or impaired asset. Refer to the risk-weighttng... for loans and claims secured against eligible residential mortgages that are past due for more than 90 days and/or impaired. Class VII - Other assets and claims Claims (other than equity) on Australian and international corporate counterparties (including insurance and securities companies) and commercial public sector entities. Alternatively, if an ADI has obtained approval in writing from APRA, it may risk-weight all claims (other than equity) held on the banking book on Australian and international corporate counterparties (including insurance and securities companies) and commercial public sector entities at 100 percent. If an ADI has obtained approval in writing to use a 100 percent risk-weight for these claims, it must do so in a consistent manner and not using any credit ratings for any of these claims. All claims (other than equity) on private sector counterparties (other than ADIs, overseas banks and corporate counterparties). 150 100 1 2 3.4 5.6 Unrated 20 50 100 150 100 100 424 Table 12.3 Risk-rates for residential mortgages LVR (%) Standard eligible mortgages Risk-weight Risk-weight (no mortgage (with at least 40% of insurance) the mortgage insured % with an acceptable LMI) _ . —... _ Non-standard eligible mortgages Risk-weight Risk-weight (with at least 40% of (no mortgage insurance) the mortgage insured with an acceptable % LMI) % 0-60 60.01 - 80 80.01 - 90 90.01 - 100 > 100.01 35 35 50 75 100 % 35 35 35 50 75 35 50 75 75 100 50 75 100 100 100 Where the credit ratings are given as: Table 12.4 Credit ratings Credit rating grade (short-term claims on corporates, ADIs and overseas banks) Risk-weight (%) 1 £2 2 4 20 50 100 150 To be used in conjunction with the following credit rating grades: Table 12.5 Recognised long-term ratings and equivalent credit rating grades Credit rating grade 1 2 3 4 Standard and Poor's Corporation AAA AA+ AA AAA+ AA- BBB+ BBB BBBBB+ BB BB- Moody’s Investor Services Aaa Aal Aa2 Aa3 Al A2 A3 Baal Baa2 Baa3 Bal Ba2 Ba3 Fitch Ratings AAA AA+ AA AAA+ A ABBB+ BBB BBBBB+ BB BB- 425 5 6 B+ B B- Bl CCC+ Caal Caa2 Caa3 Ca ccc ccccc c D B2 B3 c B+ B B- CCC+ ccc ccccc c D LARGE CREDIT EXPOSURES An 'exposure' under prudential regulations generally means a potential for loss under a finance facility that has been provided. In terms of credit exposure, when an ADI provides a large loan to a business, it exposes itself to that business in the event that it defaults. A large exposure means that the capital base of the ADI is exposed. The Australian Prudential Regulation Authority regulation for large credit exposures is found under Prudential Standard APS 221, Large Exposures. This guideline states the objective as follows: This standard aims to ensure that locally incorporated ADIs implement prudent measures and limits to monitor and control risk of concentrations in respect of large credit exposures to individual counterparties or groups of related- counterparties on a consolidated group basis. The obvious issue here is that the Australian Prudential Regulation Authority feels that large exposures have the potential to affect ADI solvency. The standard states that credit risk exposure increases when spread through only a small number of lending counterparties. Under the standard, a large exposure is defined as an exposure to an individual or group of counterparties that exceeds 10 percent of the consolidated capital base. The capital base is that specified in the capital adequacy guidelines and includes both tier 1 and tier 2 capital. Note, however, that it is not illegal to have exposures greater than 10 percent as long as the Australian Prudential Regulation Authority is notified. In these circumstances, the authority may choose to place additional requirements on an ADI (such as a higher capital adequacy benchmark ratio) to address such risks. The Australian Prudential Regulation Authority requires that each ADI report its large exposures (10 percent and above) quarterly. This highlights the need for ADIs to specify correctly the relationships between common counterparties, to ensure all exposures are recognised. 426 SECURITISATION Securitisation is a good credit risk management tool in certain circumstances. As with large exposures, however, the Australian Prudential Regulation Authority imposes some guidelines in allowing securitisation to be effective as a technique. The issues for securitisation are to be found in the Prudential Standard APS 120, particularly, Guidance Note AGN 120.3. Guidance Note 120.3 is expansive and we cannot canvass all issues in this section. It works around the concept of a clean sale, which: ■ ■ absolves the financial institution from any legal recourse from the sale of loans results in the financial institution not holding capital against the loan. The above can be likened to the sale of a motor vehicle. Once the sale has been effected, the seller is not liable for any defects or accidents. If, however, there is an obligation held by the securitising financial institution, then the Australian Prudential Regulation Authority will deem that the credit risk remains with the financial institution. How the obligations are removed is summarised in the following sections. Clean sale supply of assets This section of the prudential note deals with the sale of loans that are not revolving, such as home loans. The following rules govern a clean sale: ■ ■ ■ ■ There should be no beneficial interest in the sold assets and absolutely no obligation to the financial institution. There should be no recourse (including costs) to the lending institution. In addition, there should be no obligation for the lending institution to re-purchase the lending assets. The amount paid for the loans should be fixed and should be received by the time the assets are transferred from the lending institution. Any assets that are provided to the special-purpose vehicle as a substitute or provided at below book value are not considered as relieving credit risk. The above guidelines have not had the benefit of substantial testing and the concern is not about legal recourse; rather, the concern is about moral recourse. The selling of financial assets is always subject to the problems of asymmetric information where the seller knows more about the loan than the buyer. There is a concern that in the event of economic downturn, when defaults become more frequent, sellers of assets maybe under a moral obligation to re-purchase assets. The moral obligation occurs because there might have been something defective about the loan that was not obvious at the time. Under these circumstances, the financial institution would have been carrying the implicit credit risk. 427 Revolving facilities Credit card receivables are the most common revolving facility that has been securitised. The conditions for a clean sale of these assets are different, given that the lending assets can have redraw facilities attached to them. The following is a summary of conditions: ■ The rights, details and obligations of each party must be clearly specified, including the distribution of cashflows. * As with asset securitisation, the financial institution cannot supply additional assets to the pool. ■ Liquidity shortfalls for the financial institution share must not exceed the interest receivable. ■ The financial institution always has the right to cancel any undrawn amounts on the revolving facilities. ■ Again, like normal lending securitisation, the financial institution must be under no obligation to re-purchase assets that have defaulted. While there will be a new Standard as of 2018, it does not substantially change the management of credit risk. CREDIT DERIVATIVES Chapter 11 discussed the innovation of credit derivatives, which had the ability to dramatically alter the credit profile of the statement of financial position of a financial institution. Given the infancy of the credit derivatives market, however, the Australian Prudential Regulation Authority acknowledges the benefits of these instruments but has imposed conservative guidelines. The guidelines for credit derivatives are divided into Guidance Note AGN 112.4, which deals with credit derivatives in the banking book, and Guidance-Note AGN 113.4, which deals with credit derivatives in the trading book. We are dealing with the management of credit risk (that is, the management of an underlying loan), so our focus here will be on the former guideline. For the purposes of the discussion, we will also assume that the financial institution is a protection buyer (that is, it is protecting a credit position). This does not negate that the possibility that a protection seller is assuming a credit position for the purposes of managing the concentration risk of a portfolio. We have already addressed large exposure issues in a previous section. Note that large exposures include credit derivative transactions. The guidance note, however, is not clear as to whether large exposures include those credit derivatives that result in the acquisition of credit risk or all credit derivatives. Given the current conservative stance taken by the Australian Prudential Regulation Authority, it may be monitoring the market as a whole. 428 The effectiveness of credit derivatives becomes an issue of how well the instrument reduces the requirement of capital adequacy. Although a credit derivative may reduce the overall credit exposure, its effectiveness may be reduced if capital relief is not forthcoming. The Australian Prudential Regulation Authority will only provide regulatory capital relief if, for example, it is satisfied that the set of credit events is not restrictive and allows the transfer of sufficient risk. It is important, therefore, that it is made clear that the credit events clearly and unambiguously transfer credit risk to the protection seller. The Australian Prudential Regulation Authority has noted that it expects one of the credit events to be bankruptcy to allow capital relief. Further, note that some materiality thresholds (an amount that is lost before the credit event is triggered) may disallow relief. The final issue in the regulatory treatment of credit derivatives is that of mismatches, which are discussed in terms of asset mismatches and maturity mismatches. A credit derivative is deemed to afford protection if the physical settlement has a deliverable obligation. If the settlement is in cash, however, then the following requirements must be met for capital recognition: ■ ■ ■ The underlying and reference assets are the same. The underlying asset is an obligation under the terms of the contract. An obligation is defined as a financial obligation. The reference asset ranks lower than the underlying asset. In terms of maturity, a financial institution is deemed to have full protection if the maturity of derivative equals the maturity of the underlying asset. If the maturity of the derivative is shorter than the underlying asset, then the residual term of the underlying asset does not count for capital adequacy. If, for example, a loan has a term of five years and the credit derivative has a maturity of four years, then only 80 percent of the exposure is counted for regulatory relief. Other issues such as currency mismatches and in-built options on credit derivatives are outside the scope of this chapter. For those interested, see the guidance note for discussion. DEVELOPMENTS IN REGULATION Regulation is now an ongoing conversation. While Basel III has a final deadline of 2019, there is no doubt that the ongoing problems in Europe, where there are shortages of both capital and liquidity, will compel regulators to continually review prudential regulations in light of developments in financial markets. Credit ratings The credit rating agency discussed here is Standard & Poor's, which is a private company owned by the McGraw publishing group. Its main business is to rate debt issues or actual borrowers. In defining its rating objectives, Standard & Poor’s (2000) state that A credit rating is Standard 429 & Poor's opinion of the general creditworthiness of an obligor, or the creditworthiness of an obligor with respect to a particular debt security or other financial obligation, based on relevant risk factors'. The ratings are generally divided into two types: short term and long term. There are other types but they are beyond the scope of this chapter. The definitions of these ratings are found in Table 12.6. Table 12.6 Standard & Poor's credit ratings Rating I AAA AA A BBB * BB B ccc cc c D’ . Description Long term This is the highest rating. The obligor’s capacity to meet its financial commitment on the obligation is extremely strong. An obligation rated AA differs from the highest rated obligations only to a small degree. The obligor’s capacity to meet its financial commitment on the obligation is very strong. An obligation rated A is somewhat more susceptible to the adverse effects of changes in circumstances and economic conditions than obligations in higher rated categories. The obligor’s capacity to meet its financial commitment is still strong, however. An obligation rated BBB exhibits adequate protection parameters. Adverse economic conditions or changing circumstances, however, are more likely to lead to a weakened capacity of the obligor to meet its financial commitment on the obligation. An obligation rated BB is less vulnerable to non-payment than other speculative issues, but it faces major ongoing uncertainties or exposure to adverse business, financial or economic conditions that could lead to the obligor’s inadequate capacity to meet its financial commitment on the obligation. An obligation rated B is more vulnerable to non-payment than obligations rated BB, but the obligor currently has the capacity to meet its financial commitment on the obligation. Adverse business, financial or economic conditions will likely impair the obligor’s capacity or willingness to meet its financial commitment on the obligation. An obligation rated ccc is currently vulnerable to non-payment, and depends on favourable business, financial and economic conditions for the obligor to meet its financial commitment on the obligation. In the event of adverse business, financial or economic conditions, the obligor is not likely to have the capacity to meet its financial commitment on the obligation. An obligation of cc is currently highly vulnerable to non-payment. The c rating may be used to cover a situation where a bankruptcy petition has been taken but payments on this obligation are being continued. The D rating, unlike other ratings, is not prospective; rather, it is used only when a default has occurred. 430 A-l A-2 A-3 B c D Short term The short-term obligation of A-l is rated in the highest category by Standard & Poor's. The obligor's capacity to meet its financial commitment on the obligation is strong. A short-term obligation of A-2 is somewhat more susceptible to the adverse effects of changes in circumstances and economic conditions than obligations in higher categories. The obligor's capacity to meet its financial commitment on the obligation is satisfactory, however. A short-term obligation of A-3 exhibits adequate protection parameters, but adverse economic conditions or changing circumstances are more likely to lead to a weakened capacity to meet its financial commitment on the obligation. A short-term obligation of B is regarded as having significant speculative characteristics. The obligor currently has the capacity to meet its financial commitment on the obligation, but it faces major ongoing uncertainties that could lead to the obligor's inadequate capacity to meet its financial commitment on the obligation. A short-term obligation of c is currently vulnerable to non-payment and depends on favourable business, financial and economic conditions for the obligor to meet its financial commitment on the obligation. See the definition of D under the long term ratings. * This credit rating starts what is known as speculative grades. Note: Plus and minus signs are used to modify the ratings AA to ccc to indicate relative standing in the rating category. Source: Standard & Poor’s 2000, '2000 Corporate Ratings Criteria', www.standardandpoors.com, accessed 15 July 2002. If credit ratings are to be the basis of alteration to capital adequacy, then it is important to understand the rating process. The process starts with a meeting with the company under consideration, to help the company know what to expect and what will be required of the rating process. The corporate credit risk factors are divided into two: business risk and financial risk. Both risks are further divided, as we will consider in the following sections. Business risk When measuring the business risk of a company, a credit rating agency will evaluate the following factors. Industry characteristics An assessment is made of the prospects and risks attached to the industry in which the business operates. This process will also indicate the cap of ratings for the business within the industry. The riskier the industry, the lower is the cap. Issues that would be addressed include: 431 ■ ■ ■ key rating factors for that industry, such as profitability factors and risk vulnerability diversification factors for businesses that have exposures to different industries size, and geographic and market dominance. Management evaluation The management of the business will be assessed for their ability to plan and implement their business plan. An opinion is formed of management's appetite for risk, and the inclusion of past business successes would be considered. There is also a focus on the structure of the organisation and any undue reliance on one person. Industry-specific factors For many industries, Standard & Poor’s will provide specific factors that it will consider. The following is a summary of these considerations: ■ industry regulations • markets ■ operations (revenue and costs) ■ competitiveness. Financial risk The following factors are considerations when measuring financial risk: ■ Accounting quality. Given that the qualitative assessment of the business is based on the audited accounts, financial reporting quality starts as a base. • Financial policy. An assessment of the firm’s financial policy depends on whether such a policy exists, how it is used, how it assesses risk and how it proposes risk mitigation. ■ Profitability and coverage. Given that earnings and cashflow repay debt, this issue is very important in Standard & Poor’s analysis. Some measures that are examined are: ■ pre-tax pre-interest return on capital • ■ operating income as a percentage of sales ■ earnings on business segments assets. A number of issues are also examined in this area, including: ■ trends ■ an analysis of historical trends and the reconciliation to projected earnings ■ earnings in relation to fixed charges. A number of adjustments are made to ensure accounting issues do not cloud the earnings power of the business. ■ Capital structure/leverage and asset protection. The following ratios are considered by Standard & Poor’s: ■ total debt divided by (total debt plus equity) 432 (total debt plus liabilities off the statement of financial position) divided by (total debt plus liabilities off the statement of financial position plus equity) ■ total debt divided by (total debt plus market value of equity). ■ Asset valuation. In terms of capital structure, the value of the business’s assets can make a material difference to viability. Financing off the statement offinancial position. Standard & Poor’s factor the following into leverage considerations: ■ operating leases ■ debt of joint ventures and unconsolidated subsidiaries ■ guarantees ■ take or pay contracts and obligations under throughput and deficiency agreements ■ receivables that have been factored, transferred and securitised ■ contingent liabilities. Preferred stock. The characteristics of the preference shares are examined to ascertain whether they exhibit the features of debt or equity. Redeemable preference shares, for example, would be considered to be debt. Cashflow adequacy and ratios. Cashflow, not accounting earnings, is what repays debt. It is the most critical factor in Standard & Poor’s assessment. The ratios used are: ■ funds from operations divided by total debt ■ earnings before interest, tax and depreciation divided by interest ■ (free operating cashflow plus interest) divided by interest ■ (free operating cashflow plus interest) divided by (interest plus annual principal repayment obligation) ■ total debt divided by discretionary cashflow ■ funds from operations divided by capital spending requirements ■ capital expenditure divided by capital maintenance. The need for capital. Standard & Poor’s examine a business’s requirements for equity and working capital, relative to its capital works needs. Financial flexibility. Standard & Poor’s examine issues such as overreliance on any one finance source. ■ ■ ■ ■ ■ ■ Much more could be added to the above, but these are the major issues that are considered in the rating process. If you are interested in more details, the full process is available on the Internet (www.standardandpoors.com). How Standard & Poor's combine the above analysis is a trade secret. All that is publicly known is the set of elements that are considered in rating an issuer or obligation. Knowing these elements, we can move onto the new capital adequacy proposals. 433 INDUSTRY INSIGHT Credit and equities: Banks ponder a double-whammy Just when Australia’s banks were back in favour with investors, a leading bank director has raised the spectre of a double-whammy from stronger credit growth and a greater investment in equities. There was nothing alarmist about the warning from National Australia Bank executive director, finance, Mark Joiner. He pointed out that the big four banks would be forced to have a greater reliance on offshore wholesale funding markets when deposits shifted out of the banking system. Joiner is flagging an issue that was a top priority for policymakers in the two years after the global financial crisis but has since slipped into obscurity. The issue can be summed up with the following questions: Is Australia content to keep funding the economy through the big four banks in offshore markets? What needs to be done to encourage market-based financing of business investment? How can banks lock in deposit funding? What impact will new capital rules have on the ability of banks to fund credit growth? Joiner’s comments were made at the release of NAB’s s 1.4 bihion cash profit for the three months to March alongside chief executive Cameron Clyne. They are timely and worthy of broader debate for two reasons. They come amid a groundswell of academic and political comment about the need for a financial system inquiry. Those calls have been backed by shadow treasurer Joe Hockey. A second reason is that the global banking system is in the midst of implementing new capital rules called Basel III that will force banks to match the maturity of their loan assets with deposits of similar time frame. Joiner has tried hard to get a discussion going about the role banks play in the financial system. He has been a strong advocate for developing a corporate bond market so that superannuation savings can be recycled more efficiently and reduce Australia's reliance on banks. Joiner thinks the war for deposits will continue for five years but he can see the day when the money on deposit will shift elsewhere. He is not alone in worrying about this. Credit rating agencies have warned that stronger economic growth will result in Australia's banks being forced to draw more heavily on offshore markets for wholesale funding;. Moody’s Investors Service cut the ratings of the big four banks about a year and a half ago because of concerns that a stronger domestic economy would increase the need for offshore funding. The Australian Prudential Regulation Authority (APRA) keeps a close eye on the deposit and wholesale funding structures of the banks. It made historic concessions to allow the banks to diversify their funding through the issue of covered bonds, which have a higher credit worthiness than bank bond issues. 434 While it is true that the banks have diversified their wholesale funding by geography, type and quality, they continue to rely heavily on short-term deposits. Joiner says the level of deposits in the banking system from Australian companies is at a record level. The APRA data makes it clear that if it were not for this business conservatism, the overall deposit to loan ratio would be a lot worse than it first appears. This is evident from an examination of loan to deposit ratios using data published by APRA. Household loan to deposit ratios are 10 times worse than those of big business. The shortfall in household deposits relative to loans as of March this year was: ANZ Banking Goup $112 billion, Commonwealth Bank of Australia $173 billion, NAB $126 billion, Westpac Banking Corp $186 billion. The shortfall between non-household loans and non-household deposits was: ANZ $28 billion, CBA $5.7 billion and NAB $24.8 billion. Westpac had deposits in excess of loans of $6 billion. The slump in credit growth in Australia, which has been welcomed by Reserve Bank of Australia governor Glenn Stevens, has allowed the banks to fund their lending with deposits. Stevens has said that the reduction in credit to sustainable levels is positive for Australia and he doubts if there will soon be a return to the levels of credit growth seen over the 20 years leading up to 2008. However, Joiner is more concerned with retail depositors turning to other asset classes, and businesses taking their cash to invest in growth. Deposits are usually a short-term liability, unlike most banking assets. Rating agencies like to talk about the behavioural characteristics of assets and liabilities. They say an asset like commercial paper, which is usually issued for three to six months, has much better behavioural characteristics than a balk deposit, which can often be at call. The regulators have recognised the problems that can be caused when there is a mismatch of assets and liabilities in institutions that are operating at very high leverage. Before the financial crisis, Australian banks had leverage of about 20 to 30 times their capital. Banks overseas such as Deutsche Bank and UBS had leverage more than double those levels. The new capital rules are designed to lessen that problem by forcing banks to match deposits and liabilities. But that regulatory approach will penalise banks with assets that have long-term maturities relative to banks with short-term asset maturities. The heavy proportion of Australian lending on mortgages, which have terms of 20 to 30 years, means they will be penalised relative to a commercial bank lending in syndicates or through commercial paper for periods of three to five years. One way to lessen the reliance of Australian banks on mortgages is to encourage securitisation of mortgage loans. This is one of the features of the Canadian financial system, which has a government guarantee of mortgage insurance. The Canadian model for securitisation has gained strong support in Australia from various commentators and academics and is said to be attractive to Hockey. Australian Securities and Investments Commission chairman Greg Medcraft threw his support behind it before be joined ASIC. 435 However, the Canadian model of guaranteeing mortgage insurance so that mortgage securitisation programs carry a government guarantee has several drawbacks. Canada's home loan default rates are double that of Australia, which suggests that government intervention has increased the level of imprudent lending. The Canadian Mortgage and Housing Corporation had accumulated total liabilities of SC570 billion at the end of March this year. One positive side effect of the Canadian approach is that Canadian banks can use a zero risk weight for securitised mortgage loans compared with a 50 percent risk weight for Australian mortgages on the books of Australian banks. Securitised Australian mortgages could be sold to Australian superannuation funds and thereby lessen the need for banks to tap offshore markets. The issues and questions raised by Joiner do not need to be answered immediately. It will take many years for Australians to recover from the financial crisis. Deposits will remain popular. But deposits are not sticky. They can move fast when interest rates fall. Australian banks will probably continue to source about $80 billion a year in funding from offshore markets, which is manageable. But those markets are a lot less stable and less liquid than they were before the crisis. The deferral of the TRUenergy initial public offering is not unexpected considering the state of equity markets. The joint lead managers were happy to be on tap for a deal but were aware the listing could slip into next year. The second half of this year was looking a little crowded on the IPO front, despite the deferral of the Genworth IPO earlier this year. TRƯenergy was in a pipeline for 2012 that included the $3 billion Coates Hire IPO and the McAleese Transport float. The first half of next year could be a boomer for investment bankers and the various IPO advisers. But that does not help with the here and now. There is a possibility that 2012 will end with an IPO bang of sorts but that will require a strong profit reporting season with plenty of positive surprises. He said regional banks would come under pressure to invest heavily in advanced risk management systems to avoid having to meet more onerous capital obligations than the major banks. • Souice: Boyd, T 2012, 'Credit and equities: banks ponder a double-whammy’, Australian Financial Review, 15 August, p. 42. When capital adequacy was introduced in 1980, it was a simple calculation that was designed to be comparable across all banks, regardless of jurisdiction. However, its simplicity was also its weakness - for two reasons. The risk weighting of 50 percent for home loans, regardless of LVR or mortgage insurance and 100 percent for corporate risk, pushed banks to heavily weight their lending books toward mortgages. Secondly, the 100 percent risk weighting for corporates failed to differentiate the various risks. This was an added cost of regulation. These issues were resolved in Basel II, introduced in 2004, whereby home loans were better articulated and corporate credit risk was risk weighted by credit rating. However, it should be mentioned that these new measures were not successful in averting the global credit crisis. The 436 major issue here was liquidity, and this has been addressed in Basel III. But this has led to the joining of credit and liquidity issues in Basel III, and this issue is the basis for the above article. The major import of this article is that prudential regulation, while providing safety, also has a cost. Under Base III considerations, assets and liabilities maturities might have been matched. This points out the ever increasing layers of regulation. And, as the article points out, it would come at a cost, as regulation increases costs. But it also points to another observation - that credit behaviour is now being influenced by regulation and is becoming more complex. SUMMARY 1. What are the issues of credit risk from the perspective of the regulators? Regulators do not directly regulate the credit risk exposures of ADIs. A number of regulations, however, are in place primarily to protect the interests of depositors. 2. How is capital adequacy related to credit risk considerations? For every loan that a lending institution makes, the institution is obliged to set aside capital. This is to protect depositors in the event of impaired assets. The minimum benchmark for capital adequacy is 8 percent. 3. What are the issues of large exposures? Internationally, many financial institutions have failed because they have lent too much to a single entity. To ensure this is not the case in Australia, the Australian Prudential Regulation Authority requires that a lender report any exposure in excess of 10 percent of its capital. 4. What are the securitisation issues for regulators? Lenders use securitisation for many reasons. The credit risk factor, however, is of most concern to the Australian Prudential Regulation Authority. If ADIs use securitisation for capital purposes, which is the most common reason, then the authority requires a clean sale which means that the ADI cannot assume the credit risk of a securitisation structure in the event of the default of an underlying security. 5. What are the credit derivative issues for regulators? Credit derivatives are an effective tool for dealing with credit risk. The Australian Prudential Regulation Authority requires, however, the direct matching of terms for the regulatory relief for these instruments. 6. What is the credit rating process? Credit rating is a process whereby an independent body assesses the probability that debt issues will be repaid in a timely manner. The two most well-known credit rating agencies are Standard & Poor's and Moody's. Such agencies will assess both the business risks and the financial risks of a company that desires a credit rating. Each agency's 437 exact process of assessment, however, is a trade secret. The credit rating process has become important because it will be the basis of new capital adequacy guidelines. 7. What are the new capital adequacy guidelines? One major criticism of capital adequacy is that it does not distinguish the various credit risks of business. Under the original guidelines, a company nearing default could have the same profile as those companies with AAA ratings. The new guidelines seek to differentiate these types of credit risk. DISCUSSION QUESTIONS Explain how the capital adequacy guidelines deal with the regulator’s concern for credit risk. 2. Westpac’s 2001 statement of financial position is presented opposite. Calculate the capital adequacy ratio. 3. Discuss the shortcomings of the current capital adequacy guidelines. 4. How do the proposed capital adequacy guidelines deal with the shortcomings that you noted in question 3? 5. Referring to the Westpac financial statement again, what difficulties do you encounter if you need to calculate capital adequacy under the new guidelines? 6. Should all financial institutions be able to use internal ratings? 7. What would be the difficulty in identifying large exposures? 8. Discuss the advantages and disadvantages of concentrated credit portfolios. 9. From a regulatory point of view, what are problems with securitisation as a credit risk tool? 10. Credit derivatives are an effective credit risk tool. Why are the regulators concerned about them? 11. Read the ‘Industry insight' (Regional banks and the Basel II capital standards’. Consider which sections of a regional bank’s lending portfolio are riskier than those of a major bank’s lending portfolio. Then, assess what you consider to be an appropriate capital adequacy provision for regional banks. You should consider the difficulty of distinguishing between regional banks and major banks. 12. What are the difficulties with using credit rating agencies in the due regulatory process? 1. 438 Statement of financial position as at 30 September 2001—Westpac Banking Corporation and its controlled entities Consolidated 2000 2001 (Sm) (Sm) Assets Cash and balances with central banks 1 079 836 Due from other financial institutions 5 094 3 325 10 629 7 174 2 960 2 731 122 250 107 533 15 700 15 665 7 352 7 547 482 620 Goodwill 1 501 1 535 Fixed assets 1034 1 Trading securities Investment securities Loans Acceptances of customers Life insurance investment assets Regulatory deposits with central banks overseas 175 441 467 21 323 19010 189 845 167 618 5 954 3 972 Deposits and public borrowings 96 157 89 994 Debt issues 27 989 19 203 Acceptances 15 700 15 665 706 651 Life insurance policy liabilities 7 123 6 991 Provisions 1038 989 20 635 175 302 15 999 153 464 Subordinated bonds, notes and debentures 4 045 4 175 Subordinated perpetual notes Total loan capital 793 4 838 717 4 892 180 140 158 356 9 705 9 262 Deferred tax assets Other assets Total assets Liabilities Due to other financial institutions Tax liabilities Other liabilities Total liabilities excluding loan capital Loan capital Total liabilities Net assets 439 Consolidated 2001 2000 ($m)(Sm) Equity Share capital 2 233 2 258 465 465 Reserves 2 819 3 099 Retained profits Equity attributable to equity holders of Westpac Banking Corporation 4 174 3 435 9 691 9 257 Outside equity interests in controlled entities 14 9 705 5 Trust originated preferred securities Total equity 9 262 Note: The above statement of financial position should be read in conjunction with the accompanying notes and discussion and analysis in the following source. Source: Westpac Investor Relations. See Westpac Banking Corporation 2001, www.westpac.com.au, accessed July 2002. REFERENCES AND FURTHER READING Australian Prudential Regulation Authority 1999, Capital Adequacy of Credit Derivatives, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 120.3, Purchase and Supply of Assets (including securities issued special-purpose vehicles), September, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.1, Impaired Asset Definitions, September, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.2, Security Valuation and Pro­ visioning, September, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.3, Prescribed Provisioning, September, Canberra. Australian Prudential Regulation Authority 2000, Prudential Standard APS 120, Funds Management and Securitisation, September, Canberra. Australian Prudential Regulation Authority 2000, Prudential Standard APS 221, Large Exposures, Sep­ tember, Canberra. Basel Committee on Banking Supervision 1999, A New Capital Adequacy Framework, Bank for International Settlements, Basel. (Re-issued for comment May 2001.) Standard & Poor's 2000, accessed 15 July 2002, 2000 Corporate Rating Criteria, www.standardandpoors. com. PROBLEM LOAN MANAGEMENT LEARNING OBJECTIVES By the end of this chapter you should be able to: 1. outline why loans default 2. highlight the extent of problem loans 3. explain why the business cycle is important for problem loans 4. define problem loans, provisions and regulatory issues 5. discuss the capital issues of problem loans 6. define ‘structure dynamic provisioning' 7. restructuring problem loans 8. illustrate a case from law. KEY TERMS bad debt write-off dynamic provisioning liquidation provisions business cycle general provisions mild financial distress severe financial distress 441 coordination problem impaired assets moderate financial distress specific provisions 442 INTRODUCTION For most of this book, our focus has been on the assessment and approval of loans. Lending is clearly a risky activity, however, and lending institutions occasionally grant loans that incur a loss. The loss may occur as a result of many factors, from poor management of the borrower to the timing of the business cycle. Bad loans, known commonly known as Non Performing Loans (NPLs), have the following effect on lending institutions (Alvarez and Marsel): ■ Reduction in net interest income; ■ Increase in impairment costs; ■ Additional capital requirements for high risk weighted assets; ■ Lower ratings and increased of funding adversely affecting equity valuations; ■ Reduced appetite for new lending; and ■ Additional management time and servicing costs to resolve the problems. The first course of action, therefore, is not to foreclose, but to manage the asset or firm. In more recent times, analysts have also viewed bad loans as a risk to the profitability of lenders and their ability to pay dividends. In the eyes of the regulators, credit default is a serious issue and is now being viewed carefully. While we make the occasional comment about regulatory issues, it is considered more fully in the chapter on credit risk from the regulator's perspective (Chapter 12). CAUSES OF DEFAULT A default is defined here as a loan for which the repayments are overdue. Lending institutions may experience defaults and problem loans for the following reasons (Golin 2001): ■ lack of compliance with loan policies ■ lack of clear standards and excessively lax loan terms ■ inadequate controls over loan officers ■ overconcentration of bank lending ■ loan growth in excess of the bank’s ability to manage ■ inadequate systems for identifying loan problems ■ insufficient knowledge about customers' finance ■ lending outside the market with which the bank is familiar. All these reasons for default are found within a lending institution. Many problem loans could be avoided by better lending procedures and policies. Also important is credit culture. This refers to the culture of personnel in dealing with procedures and policies. Often, when a loan becomes bad, it is because the policies and procedures are either circumvented oi ignored. While this may 443 not be a problem in the short term, when loan portfolios become stressed, credit culture becomes exposed. This was a common characteristic in American banks during the global financial crisis. Credit risk is never static, however', and many loans that were validly granted can become bad for many different reasons. Two examples are when a recession affects firms that rely on cashflow or when firms wind up because their products have become outdated. The issue then becomes how best to monitor these situations. Monitoring is easier said than done. While it may be easy to monitor a small portfolio of loans, the situation becomes more complex as the financial institution becomes larger. This complexity introduces higher and higher costs for monitoring. To ensure efficiency, indicators (such as consecutive missed payments) are normally implemented. These indicators are normally ' noisy', however, which means that they do not present a clear picture of the situation or indicate remedial action. In many cases, these indicators highlight a problem loan when it is too late, resulting in a less than optimal situation for the lending institution. In Chapter 11, we noted that one function of default models is that they can provide early warnings of developing problem loans. This can be helpful for monitoring purposes. THE EXTENT OF PROBLEM LOANS Table 13.1 shows the experience of Approved Deposit Institutions since 2004 with the value of impaired assets (bad loans) they are managing. From lows in the early 2000s, the value of impaired assets increases sharply from the September 2008 quarter when the global financial crisis hit. September 2004 December 2004 March 2005 June 2005 September 2005 December 2005 March 2006 June 2006 September 2006 December 2006 March 2007 June 2007 September 2007 December 2007 Impaired Assets 5,113 4,146 4,045 4,030 3,685 3,587 3,610 3,423 3,386 3,732 3,807 4,121 4,289 4,414 Total assets 1,714,182 • 1,788,696 1,778,739 1 828,435 1,874,146 1,937,173 2,045,476 2,095,849 2,163,357 2,257,753 2,329,638 2,469,109 2,658,223 2,753,381 % Impaired 0.30% 0.23% 0.23% 0.22% 0.20% 0.19% 0.18% 0.16% 0.16% 0.17% 0.16% 0.17% 0.16% 0.16% 444 March 2008 June 2008 September 2008 December 2008 March 2009 June 2009 September 2009 December 2009 March 2010 June 2010 September 2010 December 2010 March 2011 June 2011 September 2011 December 2011 March 2012 June 2012 September 2012 December 2012 March 2013 June 2013 September 2013 December 2013 March 2014 June 2014 September 2014 December 2014 March 2015 June 2015 September 2015 December 2015 March 2016 June 2016 September 2016 December 2016 March 2017 June 2017 8,339 8,817 13,323 20,941 24,991 29,214 29,103 31,020 32,743 33,057 32,386 31,301 30,084 30,208 30,467 30,295 29,260 28,660 29,185 26,672 26,396 25,772 24,687 22,367 21,657 19,874 17,513 15,896 15,228 14,381 13,751 13,756 14,574 15,011 15,231 15,306 13,480 13,219 2,852,091 2,904,256 3,119,934 3,343,185 3,215,760 3,152,011 3,104,346 3,156,131 3,142,801 3,264,926 3,263,358 3,278,906 3,278,423 3,350,943 3,542,142 3,452,212 3,480,149 3,594,762 3,608,774 3,637,839 3,632,929 3,827,501 3,797,762 3,950,661 3,947,367 4,039,336 4,145,486 4,324,342 4,468,795 4,414,940 4,569,924 4,569,792 4,520,480 4,639,799 4,511,346 4,628,430 4,531,575 4,634,819 0.29% 0.30% 0.43% 0.63% 0.78% 0.93% 0.94% 0.98% 1.04% 1.01% 0.99% 0.95% 0.92% 0.90% 0.86% 0.88% 0.84% 0.80% 0.81% 0.73% 0.73% 0.67% 0.65% 0.57% 0.55% 0.49% 0.42% 0.37% 0.34% 0.33% 0.30% 0.30% 0.32% 0.32% 0.34% 0.33% 0.30% 0.29% 445 Since 2004, when the statistics were revised to be more consistent, Australia's impaired loan performance has been an average 0.51% of total assets. The peak of this was 1.04% as a result of the Global Finance Crisis of 2007 and 2008. Recent performance has fallen 0.29% (June 2017), which outperforms most of the western world. Lenders also need to recognise the potential for some loans to default and build this potential into loan pricing. It is equally important for lenders to recognise that loans that default have a corresponding impact on the profitability of the institution and, ultimately, this loss is borne by the shareholder. While bank bad loan performance can be put down to poor individual loan decisions, sometimes the overall performance can be influenced by over reliance on a given sector. This can be seen during mining boom when banks over lent to this sector without taking into account that commodity prices can fall. This risk is called concentration risk and we will address this in other chapters as well. It is the subject of this Industry Insight. INDUSTRY INSIGHT The Australian New Commentary The article highlights how over lending to one sector, in this instance, mining creates a domino effect in other sectors. Over lending to one sector is referred to as concentration risk and this article highlights that it rarely stops at one sector but affects others. In this instance, the growing problem in the mining sector has knock on effects in other sectors as well as geographical locations. Here, the knock on effect has been to contractors and small business, plus the city of Perth (and obviousfy, in the greater scheme, Western Australia). The article highlights the affect the down turn has on both lending facilities of others, here an extension of a facility for one month but the effect on employment as businesses struggle. When banks are over exposed to one sector, a correction can be damaging. The primary issue here is the drop of iron prices from USD 140 a ton ne to just USD 60. As pointed out, when looked at in terms of income, this increased the debt charge from 14 basis points for loans to 73 basis points. The article infers that this problem is partly caused by the business cycle and this is the next topic in this chapter. 446 THE BUSINESS CYCLE A bank's experience with problem Ioans can normally be tracked by examining the business cycle. The new Basle guidelines address this in the new guidelines. It is useful, therefore, to consider the characteristics of the business cycle results in the problem loan experience. This issue is becoming more important. There is a tendency to grant a loan at a point in time and determine that it has become impaired at another point. This distorts management information and portfolio management. Regulators now ask financial institutions to consider loan approvals through the cycle. While we learn later that regulators focus on through the cycle provisions, it is important to realise that rather than provide for a loan at a point of time, loans should be provided through the cycle. Recovery and expansion During this period, confidence in the economy flourishes and new investment increases. Increased consumer confidence leads to increased spending, which normally finds its way into bank deposits. This gives banks an overall increase in money supply. With this increased liquidity, banks look for more opportunities to lend. Unfortunately, this impetus often leads to the relaxation of lending standards. While interest rates also increase at this time, they tend to be relatively low. This exacerbates the situation because it encourages marginal investments by individuals and business, as well as poor lending strategies by lending institutions. Boom This period of the business cycle is exemplified by asset inflation. Much investment goes into real assets such as real estate, with much borrowing against these assets. The other characteristic of this part of the cycle is overconfidence, which may lead to declining credit standards. While interest rates are probably rising, this rise still does not dampen economic activity. Downturn While it is not easy to explain, the confidence in economic activity reaches a peak and then enters a downturn. While a book can be written on the economics of this situation, one characteristic of this section of the cycle is that asset prices decline. The decline of asset prices leads to less spending and, ultimately, declining cashflow for many businesses. Where there have been lax credit standards, loans approved under those standards no longer perform because they were written during more optimistic times. Banks experience their greatest problem loans at this time, because many loans were written against asset valuations that are now worth much less. The marginal investment opportunities taken during more optimistic times become losses for both investors and lending institutions. 447 PROBLEM LOANS, PROVISIONS AND REGULATORY ISSUES When a borrower misses a repayment on a loan, a number of questions are triggered within the lending institution. The first question is whether the missed payment is a temporary situation or one that threatens to be permanent. If the situation is temporary, then the lending institution will manage it differently than a more permanent situation. Internationally, if the situation persists for longer than ninety days, then the loan is defined as an impaired asset or non-performing loan. This is also the situation in Australia. It is called 'impaired' because the lending institution is not receiving full return on the loan and, therefore, the loan is not fully valued on the financial institution's statement of financial position. Even if a loan is partly repaid under these circumstances, then it becomes impaired; that is, a full repayment does not need to be missed for an asset to be declared impaired. Given that the financial institution has funded the loan, any portion of the loan not repaid in a timely fashion will result in the loan being declared impaired, because it has the potential to create negative income. The position of the loan in the institution's statement of financial position and income statement will depend on the likelihood of the loan being repaid and the time for which it has been impaired. For regulatory purposes, the Australian Prudential Regulation Authority considers that if a borrower has multiple facilities and one of those facilities becomes impaired, then all facilities should be classified as impaired. The level of impairment is generally seen to be the face value of the facility less the market value of any security. There are also guidelines on the valuing of security (as discussed in Chapter 12). When a lending institution recognises a problem loan, it needs to raise provisions. Chapter 12 highlights how the provisions are calculated. OTHER CONSIDERATIONS WITH PROBLEM LOANS We have been considering the regulatory and accounting definitions for the classification of problem loans. It is worth recognising that many banks have internal classifications for provisioning purposes, apart from any statistical purposes. One may question why lenders come up with their own schemes as well as the ones required by the profession and regulators. Lenders allocate capital against loans, so apart from the normal capital adequacy guidelines, problem loans are going to affect the efficient use of capital, particularly because they are not generating optimal income. Many lenders will therefore seek to redefine provisioning to align with their capital policy. The internal policy will reflect the risk appetite of the lender. If the internal policy results in greater than required provisions, then the lender could be argued to have a conservative risk profile, while the opposite would represent a higher appetite for risk. 448 Chapter 11 highlighted that lending institutions often allocate capital to lending using, apart from capital adequacy, the risk-adjusted return on capital (RAROC) or the CreditMetrics™ method. Considering these methods, we could conclude that: ■ a loan that has become impaired would fail the RAROC hurdle rate ■ CreditMetrics™ should provide a higher capital allocation amount. THROUGH THE MARKET PROVISIONS Reviewing Table 13.1 again, holding impaired assets on the statement of financial position does not generate income. It is true that bad debts occur at the lowest point of the business cycle, which is when financial institutions' profit statements are particularly vulnerable. What this situation shows, however, is that credit will deteriorate over time until some loans default. In other words, lenders need to recognise that: 1. 2. credit risk is not static, but changes over time bad debts should not come as a surprise. While loans are costly to monitor, models examined in earlier chapters, if properly used, should highlight any loans becoming problematic. Financial institutions nevertheless should be looking at methods that smooth the bad debt experience and charge doubtful debts against income over time rather than at the bottom of the economic cycle. The process should lead—rather than lag—the bad debt experience. This approach will minimise income volatility that occurs as a result of problem loans. The process as originally known as dynamic provisioning but will now be part of the Basle 3 guidelines to ensure adequate capital is available for bad debts. The process of estimating through the cycle debt provisions is shown in Chapter 12. DEALING WITH DEFAULTS Having discussed definitions, regulatory requirements and financial institution practice, we are now in a position to examine the processes to undertake when identifying a problem loan. Lending institutions are most reluctant to discuss their experiences and practices in this area, so here we will identify three types of potential default situation and develop principles for dealing with these situations. The three situations are: 1. 2. 3. mild financial distress moderate financial distress severe financial distress. In dealing with these categories, keep in mind that some problem loans can be difficult to categorise and may display characteristics of more than one situation. Two principles always apply, however, in dealing with these loans: 449 1. 2. The primary aim of the bank is to minimise the loss to the bank. In many circumstances, this will mean not liquidating the loan because the collateral will be worth only fire sale value. To manage these problems correctly, the economic worth of the loan is compared with the economic worth of the borrower. We will discuss the three types of financial distress and then complete this section with some comments on the coordination problem loans where multiple lenders and priorities are involved. Mild financial distress Mild financial distress occurs when companies experience temporary cashflow shortages. In most cases, this type of distress never enters the public arena and is not captured by the regulator's definitions. As long as the loan does not stay in arrears for longer than ninety days, this type of situation is opaque to the general community. In many instances, the cashflow shortages are temporary and may be rectified within days. A major receipt might have been delayed, for example. The shortage is sometimes more lasting and more serious remedial action needs to be considered. The overall condition of a company experiencing mild financial distress is that its economic worth is of higher value than the repayment schedule of the loan. In other words, creating a situation of default could rapidly depreciate the value of the firm's assets, causing both the borrower and lender to lose money—a situation to avoid. A number of remedies can be used in this instance. The simplest approach is for the bank to agree to an extension on the repayment, in recognition of the temporary nature of the situation. In most instances, the bank would charge penalties to ensure there are disincentives to prevent the situation arising again. A review should be undertaken to ensure all potential revenues are being exploited and costs are under control. The default of the telephone company One.Tel in 2001 was partly due to the failure of its billing systems to bill its clients properly and identify those customers requiring further action. Some One.Tel customers had accounts that were overdue by more than three hundred days. In other words, expected cashflow did not materialise. Note, however, that banks should take further reviews or actions to ensure that their position is protected. Remembering that cashflow repays the loan, a lender should take steps to protect the cashflow of the borrower. The following suggestions are not exhaustive and critical review skills should be used when examining each situation. The most common cause of cashflow shortages in firms is overly rapid growth of the firm. As firms grow, they need more investment in productive capital, whether computers, plant or land. While this investment is undertaken, revenue normally does not keep pace until the growth stage is complete. The solution may be to delay the investment until the temporary cash shortage has disappeared, because there may be a logical milestone in the expansion where cashflows 450 reach a necessary level. In other words, the maturity of the development is so finite that it is worthwhile accepting the temporary repayment of loan facilities. In some instances, borrowers may hold assets that perform poorly. In many cases, these assets are not core to the borrower and were acquired either through acquisition of another firm or where diversification of firms in a group was seen as a positive step. In any event, selling non-core assets, particularly if they are poor performers that degrade overall performance, can generate valuable cashflow. The final suggested solution is a simple one. Where lending can be a matter of transferring risk to the appropriate party, mild financial distress may be a matter of requiring the shareholder to supply more equity to the business. In other words, it may not be appropriate for the lender to make any concessions. Moderate financial distress As mentioned previously, the difference between the various levels of distress is one of degrees. In the case of moderate financial distress, a temporary cashflow shortage again is evident, but the economic worth of the company is less than the repayment schedule of the loan. If the bank were to wind up the borrower, then it would generate a loss in the process. This loss would depend on the value of any offered collateral. A registered residential first mortgage would have little negative effect on the value of its collateral (the home), while a manufacturing firm in default may find its economic worth rapidly degrades in the absence of a buyer of its assets, particularly if the assets are unique. The lender should simply liquidate the registered residential first mortgage but exercise more care in the latter case. It may be more beneficial for the lender to restructure the loan. As a simple example, Little Company owes Big Bank $300. The owner of Little Company has some special skills to run the firm for $15. These special skills result in the firm being worth $315 with a probability of 0.8, otherwise zero. We will use these values to calculate the expected value of the firm. (Additional information on the calculation of expected values can be obtained from any good business or corporate finance textbook.) The liquidation value of the firm is $200. Before going through the options available, it is important to point out that the Little Company's owner's special skills align with the fact that liquidating the firm would cause the assets of the firm to depreciate in the absence of a buyer. This is the same as saying that the assets are unique. In this example, there are two options. The first option is to liquidate. This is the appropriate strategy for Little Company because its expected value (or net present value) is negative, as follows: 0.8 (315 - 300) + 0.2(0)- 15 = -3 451 What this tells US is that Little Company's pay-off will be $315, with a probability of 0.8 in the good state. The firm needs, however, to make a repayment of $300. It receives nothing in the bad state and spends (exerts) $15 worth of energy to produce. Its expected value is zero, so it would prefer to liquidate and receive nothing. Big Bank would receive $200—a loss of $100—in liquidation, which is the difference between the loan amount and liquidation value. The second option is to restructure the loan to give the owner the incentive to continue. This restructure is determined by calculating the break-even amount of the loan, which we will call x—that is, the circumstances under which Little Company would continue: 0.8 (315 -x) + 0.2(0)- 15 = 0 The above formulation is the same as for the first option, except we are solving for X when the expected pay-out is zero (the minimum that Little Company will accept). We find that X, or the break-even loan amount, is $296.25. A loan of $296.25 would be better than liquidation value of $200 and the owner of Little Company would have the incentive to continue. Note that Little Company would continue the gain as the value of the loan falls. In these circumstances, the bank loses $3.75 rather than $100. Severe financial distress This financial distress is the most obvious of all. It normally finds its way into the provisions for doubtful debts of a bank. Under this scenario, severe financial distress is characterised by a missed debt payment as well as the borrower having an economic worth less than the repayment schedule. The normal course of action is to wind up the firm, but this may not always be the best course of action. A number of issues need to be considered. The first issue is whether the borrower has a sound business. Is the default due to reasons other than the nature of the business? When Fairfax Limited defaulted on its loans in the early 1990s, for example, the business was quite sound. The banks were able to arrange the company as a going concern and refloat it. Given the level of intangibles (being the mastheads, such as the brand name of the Sydney Morning Herald), there would have been little point to winding up the company. Can we demonstrate how this would occur? The following example should help. Big Bank is again having problems with Little Company (some lenders never learn!). Having found its way out of trouble, Little Company finds that it owes senior bond holders $75 and Big Bank $500. It still costs Little Company $15 to run the company and, if it continues, it will be worth $520 with a probability of 0.75, otherwise zero. The liquidation value of the firm is $90 and Little Company wants to default. What should Big Bank do? This is a simple process of computing pay-offs. 1. The bond holders would prefer to liquidate because they would receive $75 with certainty; otherwise, their expected payment is $67.50 ($90 multiplied by 0.75, taking into account that we are looking at the start of the period, not the end). 452 2. 3. Big Bank's expected pay-off is $350 ($500 multiplied by 0.75) if the firm continues; otherwise, it receives only $15 from liquidated funds. The bond holders are seen to be the senior lenders in this example and $15 would be the residual after they are paid. Little Company has little incentive to continue, not wanting to lose $ 15. Big Bank, therefore, needs to restructure the debt to ensure Little Company will want to continue, but not lose in liquidation. The process is quite simple. Big Bank needs to buy out the senior bond holders and find Little Company's break-even point. Big Bank's loan is now $575 (original loan plus bond holder debt of $75), so we need to work out the break-even debt that keeps Little Company interested in continuing. We do the same calculation as before: 0.75 (520-x)- 15 = 0 where X = 500, which represents the loan amount that ensures Little Company will not lose in liquidation. We need to compare the $500 against the liquidation value rather than the debt of $575. The coordination problem In the previous problem, we assumed that the senior bond holders would be happy to be bought out by the junior debt provider. In our example, this would be true because the bond holders would receive full value for their debt. In many restructures, however, not all debt holders receive their full entitlement and they can hold up restructures by demanding to receive their repayments in full. This is known as the coordination problem. Much of the syndicated debt in the 1980s was arranged along these lines, with many banks being involved. The important point of these arrangements was that relatively junior lenders were in a position to affect the restructuring process by stopping the rollover of facilities. That junior lenders were in such a position of power may be construed as faulty contract design. We return to Big Bank and Little Company to see how junior lenders can have an effect on the senior lenders' position with problem loans. In this example, we find that Big Bank is the senior creditor, ahead of two bond holders (one senior and the other junior). Big Bank is owed $300, the senior bond holder is owed $100 and the junior bond holder is owed $50. Given business conditions, Little Company wants to restructure the business and offers two plans: A and B. Plan A offers a pay-off of $400 with a probability of 0.7 and $100 with a probability of 0.3. Plan B offers a pay-off of $500 with a probability of 0.5 and $100 with a probability of 0.5. How would we consider the problem? The first issue is to look at the pay-off of the company under each plan: Plan A = $400 X 0.7 4- $100 X 0.3 = $310 Plan B = $500 X 0.5 + $100 X 0.5 = $300 453 The expected pay-off for Big Bank as the senior lender would be: $300x0.7 + $100x0.3 = $240 What does this tell US? Plan A would leave $70 for the remaining lenders and Plan B would leave $60. This is clearly not enough. There are two possible outcomes to this situation: (1) the junior lenders will need to take a smaller pay-out under the restructure or (2), as occurs in many instances, the junior lenders force the senior lender (Big Bank, in this case) to take out some of their loans in return for allowing the restructure to continue. The latter case demonstrates what is known in academic literature as a 'time inconsistent contract'. When loan contracts and lending policy are put in place, they should not be exposed to any unintended renegotiation. This destroys the incentive of borrowers and lenders to perform their obligations. In the next section, we will discuss covenant breaches. Whereas covenant breaches normally put loans into technical default, there are many instances where borrowers offer a penalty payment for breaches of covenants rather than go into default. To accept this offer, the lender implicitly destroys the incentive of the borrower to exert effort to continue to perform under the terms of the lending contract. Such contracts go from time consistent (no re-negotiation in the event of default) to time inconsistent. Knowing that the lender acceded once, the borrower is under less pressure for the next breach. A Process Alvarez and Marcel have come up with the following process that gives the above some structure: I nrỉlllHIOWhiuc I‘II| IT MMMl. WMW.WK HHtf* INDUSTRY INSIGHT Centro debt-for-equity deal unique in its complexity Nabila Ahmed The $3 billion debt-for-equity swap that brought Centro Properties Group and its satellite Centro Retail under the one umbrella and the ownership of hedge fund lenders last year was the most complicated restructure struck in Australia. 454 It involved several different classes of stakeholders: shareholders, senior lenders, hybrid debt holders and junior lenders including bondholders and put option holders. Refinancing negotiations dragged on for two years. The amalgamated group re-floated in December last year on the Australian Securities Exchange. Among the hedge funds that emerged as its biggest shareholders were Silver Oak Capital (7.3 percent), Burlington Loan Management (5.8 percent) and Varde Investment Partners (5.5 percent). Trio bail out Alinta US private equity group TPG led a $2.1 billion bailout of West Australian utility Alinta Energy in 2010. Oaktree Capital and Anchorage Capital partnered TPG and the trio agreed to a compromise deal after other hedge fund lenders baulked at their first proposal. Lenders who were owed $2.8 billion accepted a 'haircut' with the debt falling to $1.55 billion. They acquired all the generation, retailing and pipeline assets, except for the Redbank power station in NSW. I-Med restructure cvc Asia Pacific-owned radiology group I-Med was restructured last year. The new entity was owned 90 percent by senior lenders owed about $550 million; and 10 percent by mezzanine lenders owed more than $300 million. cvc, Total debt in the group was slashed to about a third of the original $900 million, which had acquired it in 2006 as part of its $2.7 billion takeover of the listed DCA Group, did not receive any equity but a small fee of about $5 million. in Of the restructure, Freehills partner lohn Nestel says: contrast to Centro and Alinta we had to transfer the 1-Med group as might be done in a receivership. For the first time in our market we were able to effect a so-called ■ credit bid' whereby the senior creditors essentially exchanged their debt for ownership of the business. This was achieved outside a receivership and through some very unique intercreditor terms. It would be dangerous to assume that the same is possible in other restructures. For instance in Centro the court rejected the senior lenders' attempt to 1 nullify' certain junior creditor rights, despite an argument they were 'out of the money' because of the specific intercreditor terms in that case. ‘There is always plenty of contingency planning but in none of these restructures did the creditors ultimately appoint receivers. Lawyers and insolvency practitioners may be keen but typically lenders don't want their investment used to test these unchartered waters. There is a price for certainty and in 1-Med it was very cheap given how deeply underwater the other stakeholders were. 455 'Warring stakeholders was terrible for business but once peace was achieved, the business was able to be quickly deleveraged and the new management team have done remarkable job restoring confidence and value.' Source: Ahmed, N 2012, ''Centro debt-for-equity deoi unique in its complexity’, Australian Financial Review, 15 September. Centro is a good example of managing a bad loan. If the lenders had foreclosed on the loans, there would have been massive losses. Part of the problem was the nature of the loans, and ASIC banned the auditor for poor practices. However, there was significant value in the business. There would have been a number of ways to restructure the debt: 1. restructure the loan completely; 2. sell some businesses to pay down debt; 3. do a debt to equity swap. In the article, we note that it is the third option taken by the lenders. However, as the article points out the structure is complex and there are warring stakeholders. What this may mean in the future is interesting. The question we need to ask ourselves is there any possibility for the co-ordination problem. Other breaches In concluding this section, it is worth recognising that not all defaults are generated by missed loan repayments. With many company loans, the approval of the loan is subject to the company agreeing to various conditions, based on financial ratios. These give a lender some comfort that: ■ cashflow will not be unduly withdrawn from the company that would be available to repay loans ■ the overall risk of the company cannot be substantially changed. Such ratios are known as covenants and are normally written in such a way that a loan is repayable if a covenant is breached. Covenants can include: ■ gearing ratios ■ dividend pay-out ratios . ■ interest coverage. The question then becomes: what is the correct procedure to follow when a company breaches its covenants (a technical default) rather than misses a repayment? There is no reason not to follow the principle of assessing the economic worth of the firm. If the breach proves that the firm has become unviable, which is what a breach of a cashflow covenant would probably show, then remedial actions can be taken. Often, however, a breach of covenant has not affected the company's ability to make normal repayments, in which case the covenant should be renegotiated. 456 EXAMPLES FROM THE LAW The previous section was designed to highlight that liquidation is not always the appropriate action when debts are not repaid. In many cases, however, restructure of debts will not be possible and liquidation of the company will be the only course of action. This text is not intended to be a legal text, but the following information should be useful for understanding the winding up of debts. In most cases, a lender appoints a liquidator to wind up a company that is not in a position to repay its debt. If the lender is secured, then actions will be taken to sell the security to repay the lender. The situation is not so easy if the lender is not secured. The liquidator assesses the amount that can be recovered from asset sales. Generally, this amount is less than the total amount owing and lenders normally write off the remaining amount. While this is simplistic, some court decisions on lending and problem loans highlight the care that should be taken when lending. In a celebrated case, the State Bank of Victoria lent money to the Victorian Division of the National Safety Council. The monies were meant to be secured by containers of sophisticated rescue equipment worth $250 000 in total; in reality, however, they were empty and sold for $1,592 each. The problem here was that lenders had not exercised due care in investigating the security and examining the company. The judgement of the case was scathing of the conduct of bank officers. A FINAL WORD The fall out from the Global Financial Crisis not only focused lending institutions on their lending decisions, but also how they structure their recovery operations. Again, Alvarez and Marshal are helpful here in outlining how banks can improve their operation: ■ aligning their businesses with regulatory requirements such as setting up separate dedicated in-house NPL units; ■ identifying, categorising and provisioning NPLs more rigorously; ■ standardising and improving work-out, legal enforcement and underwriting processes; and ■ developing additional restructuring products. SUMMARY 1. Why do loans default? Loans default for a number of different reasons. Most reasons are not 'bad luck'; rather, they are poor lending practices. The reminder is always there that some loans in a portfolio will default. 457 2. What is the extent of problem loans? The experience has been that the level of problem loans does not remain static through time. Quite often, the number of problem loans is influenced by factors such as the business cycle. 3. Why is the business cycle important for problem loans? The ability to repay loans depends on the ability to generate cashflow. A firm's or individual's ability to generate cashflow can be affected by the general state of the economy. During a recession, demand is dampened, as is earning; the reverse occurs during expansion and booms. It is no surprise to find that problem loans increase during a recession. 4. How would you define problem loans, provisions and regulatory issues? The treatment of problem loans tends to be driven by regulatory requirements. A problem loan is generally defined by the lateness of repayment, with the benchmark being ninety days. After this time, a lending institution will make provisions for that loan. The provision will be either general or specific. The regulatory authorities specify statutory provisions. If the loan is deemed to be irrecoverable, it is written off. 5. What are the capital issues of problem loans? Problem loan management also has an impact on the allocation of capital. Capital allocation for loans covers unexpected losses, so it is not surprising that many lending institutions align their provisioning policies with their capital policies. 6. What is structure dynamic provisioning? As noted, problem loans can introduce volatility into a lending institution's earnings. In recognition that problem loans will occur, lending institutions seek to use their historical experience to forecast future problems’and thus smooth the problem loan experience. This method is known as dynamic provisioning. 7. How are problem loans restructured? There a number of ways of dealing with defaults. The approach taken will depend on the extent of financial distress being experienced by the borrower. There are three types of financial distress: mild, moderate and severe. The general approach is to compare the economic value of the firm with the repayment schedule. 8. What do cases from law illustrate? In illustrating the issue of generating problem loans, legal cases often find that blame lies with the conduct of lending officers investigating the security and examining the company. 458 DISCUSSION QUESTIONS , Why are problem loans an issue? Explain the difference between accounting, regulatory and internal provisioning policies. Why are some parts of the business cycle identified with increased numbers of problem loans? 4. Compare and contrast dynamic provisioning and other methods of assessing provisioning. 5. Discuss the advantages and disadvantages of dynamic provisioning. 6. Explain how to distinguish between the various forms of financial distress. 7. Would the timing of the business cycle influence the management of the business cycle? 8. Ifi Corporation has two loans outstanding. One loan is to Certain Bank for $400, while a senior bond holder is owed $150. Ifi Corporation wants to put itself into liquidation and default on its loans. The liquidation value is $160. The management of Ifi Corporation has special qualities that would result in a pay-off of $420 with a probability of 0.8, otherwise zero. For the management to continue, it would have to be paid $10. Carefully outline the options available to Certain Bank. 9. In the case of a syndicated loan, there are often senior and junior debt providers. Where a borrower defaults under this arrangement, the senior debt providers would be assumed to be relatively well protected. Under what circumstances does this not occur? What steps should be taken to protect senior debt providers? 10. What steps would you take if a borrower breached a covenant, leading it to technical default? Your answer should highlight the contract issues. 1. 2. 3. REFERENCES AND FURTHER READING Alvarez and Marshal, 2016, Best Practices for Effectively Managing Non-Performing Loans Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.1, Impaired Asset Definitions, September, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.2, Security Valuation and Pro­ visioning, September, Canberra. Australian Prudential Regulation Authority 2000, Guidance Note AGN 220.3, Prescribed Provisioning, September, Canberra. Bennet, M October 2017, "’’Banks getting nervous about exposure to mining loans”, The Australian Golin,J 2001, The Bank Credit Analysis Handbook, John Wiley & Sons, Singapore. Greenbaum, SJ & Thakor, AV 1995, Contemporary Financial Intermediation, The Dryden United States of America Press, Orlando, Florida. Hogan, W, Avram, K, Brown, c, Ralston, D, Skully, M, Hempel, G & Simonson, D 2001, Management of Financial Institutions, John Wiley & Sons, Brisbane. Mellish, M 2001, 'Majors likely to lift bad loan provisions', Australian Financial Review, 23 April, p. 41. Shanmugham, B, Turton, c & Hempel, G 1992, Bank Management, John Wiley & Sons, New York. QUANTITATIVE FINANCE LEARNING OBJECTIVES By the end of this chapter you should be able to: 1. create a framework for modelling 2. explain and measure concentration risk 3. define expected losses 4. define and measure probability of default 5. define and measure loss given default 6. define and measure prepayment risk 7. identify the problems with quantitative modelling. KEY TERMS asymptotic single risk factor expected losses loss given default prepayment risk burnout granularity macro factors probability of default 519 concentration risk Herfindahl-Hirschmann index micro factors transformation regressions 520 INTRODUCTION As mentioned in the chapter on credit scoring, modelling and the explosion of technology has allowed the development of a plethora of financial models. This is the case also for credit-related models. There are many ways of looking at these models, but we will take an approach that is initially regulatory driven, and then concentrate on areas that are of primary concern to lenders. However, this chapter needs to be understood by those who are not necessarily quantitative driven, and so those who require more advanced treatments should consult more advanced texts. In this introduction, we need to address two issues. Firstly, what it is that we are trying to measure, and, secondly, the assumptions that drive some of our models. After defining these models, we will look at a number of other modelling issues. For the most part, most of the modelling we have looked at has focussed on the issue of measuring the probability of default (PD). While this is vitally important, it is only half of the task at hand. The other task is to measure the amount of capital that should be provided for PD. We use the PD nomenclature because it is prevalent in regulatory literature, but is generally the same as credit risk. There are two types of capital: regulatory and economic. In terms of this discussion, we will focus more on economic capital since lenders focus more on this type of capital and seek to minimise it. More importantly, lenders will seek to provide capital on a portfolio basis, and this is where we need to understand the assumptions behind Basel guidelines. There are two fundamental assumptions behind the Basel guidelines. The first assumption is known as granularity, while the second one is the asymptotic single risk factor (ASRF) model. The fundamental understanding of granularity is that the nature of a credit portfolio is such that an additional loan will not change the risk of the portfolio. In other words, the portfolio is near or fully diversified. The second assumption is that credit portfolios are subject to or affected by the ASFR model which is denoted by: v.-pM. + ^i-pZ, where: V is the value of the assets of lender i at time t M is the systematic risk z is the unsystematic risk p is the risk attached to the systematic risk. The importance of this model is that the value of the lenders portfolio is affected by a single variable. 521 The import of these assumptions cannot be understated. If they do not hold for a portfolio, then the assumptions around Basel II, and in particular credit portfolios, are violated. Therefore, we will start with concentration risk, and then move onto the following modelling issues: ■ expected losses ■ probability of default • loss given default ■ prepayment models. CONCENTRATION RISK The assumptions mentioned above for Basel II are highly restrictive and, if breached, would need extra capital to cover this. There are two types of concentration risk: name and sectoral. If either of these are breached, then capital calculated by prudential guidelines may be understated. Name risk is lending excessively to one counterparty. To highlight the issue clearer, name risk breaches the granularity assumption of Basel by adding risk to the portfolio by adding a new loan. In other words, the portfolio is not perfectly diversified. The second assumption of the ASRF model can be breached by sector concentration. Sector concentration is broader than name risk. It can be across industries, jurisdictions or geography. If any of these are existent, then the ASRF model will be breached as it assumes that a single risk factor can affect the value of the portfolio. Clearly, one risk factor will not affect all industries equally, let alone jurisdictions or geographical regions. One other issue that affects this is that there will also be interactions between industries as well. Given that it is unlikely that credit portfolios will be compliant with the assumptions, the task is to measure concentration and then assign capital to it. The normal starting point is the Hefindahl-Hirschman index (HHI), which is defined as: i= 1 where: H is the HHI s is the proportion of each firm's Ioans to the overall portfolio n is the number of loans. The HHI in reality measures the amount of concentration without reference to credit risk; in other words, the credit risk is homogeneous. Regardless of this assumption, while it does measure concentration risk, there is no reference to the amount of capital that needs to be put aside. 522 There are a number of ways to refine this measure to make it more useful, albeit it will not be perfect. What is required is a model that indicates how far away the portfolio is from granularity, in other words, the level of diversification. Semper and Beltran provide an approach for this (Semper & Beltran 2011). There are two steps: First, reformulate the HHI into a vector: HHI= SHT.I.SH where SH are the individual exposures of each sector and I is the identity matrix. Then, as the identity matrix assumes equal weighting, we replace it with a variance covariance matrix of the exposure between sectors (VCM): CI=SHT.VCM.SH where CI is the concentration index. For complex reasons, the minimum covariance will be 0, so negative covariances will be counted as zero. There are a number of attractions to this approach: 1. The maximum concentration will be in the sector of the highest variance. Thus, if all loans were in this sector, there would be maximum concentration. 2. If there is maximum diversification, then the CI = 0. 3. If loans are provided to a new industry, then the concentration will only be reduced when variance of the new loans is lower than the variance of the overall portfolio. While there is no specific charge for capital, this approach gives a more refined view of concentration risk. Most other approaches use value at risk methodologies. Gurtler, Hibbeln and Vohner (2010) give a good summary of such approaches. However, they also point out that a number of statistical properties are breached in this approach. EXPECTED LOSSES All credit portfolios make losses; it is a fact of life. However, there are a number of reasons why forecasting expected losses is important. There are two major reasons, which can be divided neatly into a now issue and a future issue. We have dealt with the future issue in the problem loans section, that is, the need to put capital side. Expected losses are an issue for lenders now because they need manage this risk. For the present, lenders tend to estimate expected losses and price them in the interest rate on borrowings. In simple terms, banks can estimate these from past experience, and many banks have become quite adept in estimating such losses. However, more and more lenders are estimating the components that make up expected losses. 523 Expected loss is normally defined as: EL = PD X LGD X EAD where EL is expected losses PD is probability of default LGD is loss given default EAD is exposure at default. It is the first two that banks focus on, so we will concentrate on them. PROBABILITY OF DEFAULT . Probability of default is a dừect result of a lender's culture, policies, systems and models. Therefore, it can be said that there is no one approach to modelling default probabilities. For example, the probability of default for a lender who predominantly provides home loans will be quite different to one who provides commercial loans. For this reason, the approach we take here, again, will be a general approach. However, the advantage of a well-specified model is the ability to stress test the credit portfolio. Stress testing allows for variables to be changed and the results to be observed. Regulators are increasingly using stress testing to ensure financial system stability. The probability of default can, generically, be defined as: P(d) = f(lf, ne) where: P(d) is the probability of default If is liquidity failure ne is negative equity The functional form above is intuitively satisfying, as the two states that affect default is insufficient cashflow to repay loans in the case of If, and the disincentive for borrowers to repay when the value of their assets falls below the value of their borrowings. The ne variable is often resolved by using option pricing techniques such as KMV. However, the simplicity of the functional form does not represent reality. In reality, If and ne are both microeconomic variables which are affected by macro economic variables and therefore will not fully explain the probability of default. The following simple example may assist. Imagine a home loan borrower who loses their job and is unable to repay their loan. The resultant unemployment may be the result of such factors as the level of the 524 Australian dollar, inflation or interest rates, all of which the individual borrower has no control over. Such factors would also affect the value of the property. For a model to correctly specify probability of default, it needs to include both micro and macro economic variables. We can then specify the probability of default as follows: P(d) = f(micro - factors, macro - factors) In general, the data for estimation would be defaulted and non-defaulted entities, which lends itself to a probit regression where the probability of default would be 0 for solvent entities and 1 for defaulted entities. So using an expanded form, it would be common to express the probability of default as: V. Pl micro - factors + i 2 p. macro - factors + i Where p is the probability of default. In building this model, the major task will be the selection of variables for both micro and macro factors. To some extent the micro factors are easily specified, given the long period of research into this area. Less so are the macrofactors, although macroprudential regulation, the stress testing of the financial system, has brought this issue to the fore. The micro-factors can be found in the credit scoring chapter. Miu and Ozdemir (2009) provide some useful insights to the macro-factors. Before indicating them, it is important to understand that the probability of default will be heavily dependent on the sector. So, if we use Miu and Ozedemir's example of home loan lending in Canada, the models have the following variables. Table 16.1 Home loan lending in Canada ■ ■ ■ • ■ ■ ■ • ■ • ■ Canada general model (a broad-based model) Real GDP Industrial production Unemployment Corporate profit Slope of yield curve High-yield spread Equity indices Composite index of leading indicators for Canada Consumer credit Delinquency rate Short-term interest rate Real estate model (a sector-based model) General health of the economy - GDP ■ Unemployment ■ Interest rate ■ Retain sales ■ Composite index of leading indicators • Consumer sentiment index ■ Durable goods ■ s&p 500, TSE Composite index Specific to the sector: • S&P/TSX Capped real estate index ■ New building permits 525 • New housing starts ■ Residential real estate (US) ■ House price appreciation ■ Commercial real estate (US) o Vacancy rates o Rents per square foot Source: Mỉu & Ozedemir 2009. They also provide some helpful insights on picking the explanatory variables: ■ Explanatory power. Naturally, we would like to use variables with high explanatory power in the modelling of the systematic credit risk. ■ Forecastability. We need to make sure the variables selected can be forecasted by the bank (typically its economic department) under the specified stress events in a consistent fashion. ■ Stress testing versus forecasting models. If we would like to use these models also for forecasting of PD using the current values of the explanatory variables (rather than during the stressed scenario), we need to consider the use of various leading indicators as explanatory variables (e.g. the national composite leading indicators). ■ Model coverage. For broad-based models, we need to use more generic macroeconomic factors (e.g. interest rate and GDP); whereas, for sector-specific models, we should also include industry-specific factors. ■ Data availability. Data availability is essential for modelling and forecasting. We need to be careful that, even though the historical time series of some of the variables are available, the calculation methodology of the variables has been changed, creating consistency problems. ■ Correlations among variables. We should not include highly correlated variables in the same model to prevent multicollinearity. ■ Representation and coverage. We should try to include variables explaining the credit environment (e.g. ratio of downgrade to total rating actions, S&p's outlook distribution, etc.), as well as general economic and financial indicators. LOSS GIVEN DEFAULT Modelling loss given default (LGD) has proven to be difficult when compared to the modelling of probability of default. It is interesting to note that most of the investigations into LGD have focussed on the variables rather than the modelling technique. The variables have tended to focus on the nature of loan, contract characteristics and economic conditions. To some extent, this looks almost like the probability of default. Qi and Zhao (2011) rightly point out that taking the distribution into account is important, and their survey of methodologies is a helpful framework for this section. Whereas the probability of default often uses regression methodologies, such as probit regression, this is more difficult with LGD. This is because most LGD distributions are not 526 normal. Therefore we will split two sections into two: variable selection and modelling. The variables for loss given default are relatively standard: ■ type of loan, ■ seniority, ■ collateral, ■ term, • seniority of mortgage, ■ characteristic of company's liquidity, and ■ level of interest rates. Depending on the nature of the lending institution, there will be other important variables. Again, like probability of default, they may be industry specific variables. As mentioned, most modelling is based on ordinary least squares. The problem with this approach is that LGD should be bounded by 0 and 1; however, most regression methodologies are bounded by infinity. So the discussion below summarises Qi and Zhao's survey. The first model that they discuss is fractional response regression. The functional form of this model is: £(LGD|x) = G(x0) where X is a series of explanatory factors for LGD and G(.) is a transformation function usually a logistic function as follows: G(xJJ) = 1 1 + exp (-X0) or a log function as follows: G(x0) = exp (- exp (-xp)) To estimate the coefficeints, then, the following function is maximised: E 1, /3 - ỵ ÍLGD( X log (G(x,3)] + (1 - LGDj) X log [1 - G(x,|3)] Ị TRANSFORMATION REGRESSIONS Transformation regressions use the inverse Gaussian function (as below) by taking LGD (0,1) and transforming it to LGD(-O°,O°) to estimate the factors using OLS and then transform them back to LGD(0,l). 527 If you use this, be aware that LGD (0,1) will not be defined, and therefore some small adjustments will be required. DECISION TREES AND NEURAL NETWORKS Decision trees and neural network, while being developed differently, have a similar concept. Decision trees use rule-based trees which split as decisions are made about the factor. Neural networks use rules based on the way humans think. Again, the results will be trees of decision matrices. Figure 16.1 is an example of a decision tree. Figure 16.1 Example of a decision tree As a final comment regarding the accuracy of the above methods, assuming that the right variables are selected, each of the models provide relatively accurate predictions. PREPAYMENT RISK The last model that we will look at is that of prepayment risk. Given that a whole set of fixed rate lending is prepaid, lenders are constantly refining their models to estimate prepayment risk. This is because lenders cannot re-invest in new loans at the same rate, thus making a loss if borrowers refinance at lower rates. Many prepayment risk models are based on options. This is because there tends to be a level of interest rates. However, these models tend to perform suboptimally because borrowers do not exercise the option optimally. 528 The general formula that is used is: Monthly prepayment rate = (Refinance incentive) X (Season multiplier) X (Mont multiplier) X (Burnout) The interpretation of this formula is: 1. 2. 3. 4. Refinance incentive is the current level of interest rates with respect to interest rates in the portfolio. If current interest rates are higher, then there will be little incentive to refinance or prepay loans. Season multipliers recognise that there are times when there are higher than normal prepayments. The month multiplier is similar to the season multiplier Burnout recognise that the longer loans exist, the more likely they are to be prepaid. One problem with the above modelling is that it is heavily dependent on interest rates as the dominant singular factor in loan prepayments. However, the problem is that most of the factors use historical data to estimate future prepayments. The historical data does not fit comfortably with interest rates being the dominant factor. What this should highlight is that there are other factors that are taken up, but not exclusively, in the burnout factor. A simple example helps here. Home loans that are written for terms of 25 years usually refinanced after about 7 years. This is because people normally refinance to trade up at this time, rather than for interest rate purposes. Therefore, to successfully model prepayment risk, the other factors must be recognised. CONCLUDING REMARKS To some extent, writing a chapter on quantitative credit modelling is a challenge, as this type of modelling is relatively young and in a constant state of flux. Improvements are likely come along as technology and the understanding of credit risk improves. Much of the development is driven by regulatory change and as we go through periods of regulatory change, modelling and the approach to modelling will continue to change. However, there is a much deeper issue. Modelling cannot be carried out without data, and most lending systems were established well before models were required. So, when it comes to data, it is either not available or it is available in a form that is not particularly useful. Carrying out good modelling will, therefore, be a challenge for analysts into the future. 529 SUMMARY 1. What is the framework for modelling? Given the importance of prudential regulation, these regulations and their assumptions drive modelling. The main assumptions around these regulations are granularity and ASRF. Any breach will mean higher capital charges. 2. Why is concentration risk important and how do we measure it? The granularity assumption for prudential regulation means that credit portfolios are well diversified. The ASRF assumption means that credit portfolios can be measured by a single factor. Breaches to either of these policies will result in concentration risk. The current measure of concentration is the Herfindahl-Hirschmann index. However, as this does measure the amount of capital used, we must transform it. 3. What are expected losses? Expected loss is the losses that normally occur when taking risks. Therefore, losses on loans are expected. Currently, expected losses are priced into loans, but in the future they will be the basis of loan provisions. Expected losses are the product of probability of default, loss given default and exposure given default. It is the former two that command the most modelling. 4. How do we define and measure the probability of default? Probability of default measures the probability that a repayment will not be made for a loan. Historically, it has been measured by micro factors. However, modelling now include macro-factors as it is recognised that these affect the ability to repay. The task is to select the correct variables. 5. How do we define and measure loss given default? Loss given default is a prediction the loss generated in default. Like probability of default, much of the research is given to selecting the correct variables. However, it is more important to note that ordinary least squares, the method normally used, does not provide accurate results. 6. How do we define and measure prepayment risk? Prepayment risk is the risk that fixed rate loans will be repaid early and cannot be reinvested in loans at a similar rate. Prepayment rate models predominately use option model technology around interest rates. However, non-interest rate variables are required and these are captured in the burnout variable. 7. What are the problems with quantitative modelling? Quantitative modelling is still in its infancy. It is affected by constant regulatory change. However, the biggest issue is the access to the appropriate data. 530 DISCUSSION QUESTIONS Why are the assumptions of Basel II important when discussing concentration risk? 2. What is the major problem with the Herfindahl-Hirschmann index? 3. What are expected losses used for? 4. What have been the developments in the development of probability of default models? 5. When developing probability of default models, what would you use as the major variables if you are addressing a portfolio of mortgages? 6. Do the same task as Question 5 above for a portfolio of construction loans. 7. What is the problem with using ordinary least squares when measuring loss given default? 8. What are the alternatives to ordinary least squares when estimating loss given default? 9. What is the problem in using interest rates as the major variable when estimating prepayment risk? 10. What are the overall issues in developing credit models? 1. REFERENCES AND FURTHER READING Bessis, J 2010, Risk Management in Banking, John Wiley and Sons, West Sussex. Bolocan, DM and Litan, CM 2011, 'Estimating the Probability of Default with Applications in Provision­ ing the Portfolio of Clients of a Credit Institution' in Transition Finance and Banking Research. Choudhry, M 2007, Bank Asset and Liability Management, John Wiley and Sons, Singapore. Gurtler, M, Hibbeln, M & Vohringher, c 2010, 'Measuring Concentration Risk for regulatory purposes' in The Journal of Risk, Volume 12/Number 3, Spring 2010. Min, Q and Zhao, X 2011, 'Comparison of modelling methods for Loss Given Default' in Journal of Banking and Finance, 35. Miu, P and Ozdemir, B 2009/10 'Stress Testing the Probability of Default and migration rate with respect to Basel II requirement' in The Journal of Risk Validation, Volume 3/Number 4, Winter 2009/10. Saita, F 2007, Value at Risk and Bank Capital Management, Academic Press, Amsterdam. Sy, w 2007, A Causal Framework for Credit Default Theory, APRA Working Paper, Sydney. Semper, JDC and Beltran, JMT 2011, 'Sector Concentration Risk: A Model for estimating capital require­ ments' in Mathematical and Computer Modelling, Elseviei'.