CHAPTER 15
Credit Risk Management
RICK NASON, PhD, CFA
Associate Professor Finance, Dalhousie University Principal, RSD Solutions
The preceding chapter discussed the common elements of credit risk and market risk. Additionally, it covered some of the major principles of managing market risk. This chapter continues the discussion with a focus on credit risk, including an overview of the credit crisis that engulfed international capital markets.
CREDIT RISK ANALYSIS
The rise of credit instruments such as credit derivatives and collateralized debt obligations (CDOs) along with changes in the regulatory capital management rules for financial institutions has generated many new ideas, research, and analytical techniques for the management and trading of credit risk.1 It is important when conducting credit analysis to remember that unlike market risk, credit risk is almost always a downside risk; that is, unexpected credit events are almost always negative events and are only rarely positive surprises. Second, it is imperative to remember that credit events are almost always unexpected. In other words, no one extends credit to a customer, or executes a loan to a counterparty, expecting that it will not be repaid.
Measuring credit risk is not a trivial task. The size of credit risk is composed of three parts: (1) the size of the potential exposure at the time of default, (2) the probability of a default or credit event occurring, and (3) the loss given that a credit event has occurred.
Each of the terms in the above equation has a large amount of uncertainty in their measurement. Additionally, the measurement of each of the above terms, and in particular the probability of default and the exposure amount, are prone to large fluctuations through time.
Several different historical studies have been done of the Loss Given Default, which is also frequently known as one minus the Recovery Rate. The recovery rate is the percentage of the debt owed that the creditor receives when the affairs of the defaulted company are finally settled. Exhibit 15.1 shows the results of one such study by the credit rating agency Moody’s and as reported in Hull.2
Exhibit 15.1 Recovery Rates on Corporate Bonds as a Percentage of Face Value 1982–2003
| Class of Security | Average Recovery Rate |
| Senior Secured | 51.6 |
| Senior Unsecured | 36.1 |
| Senior Subordinated | 32.5 |
| Subordinated | 31.1 |
| Junior Subordinated | 24.5 |
It is important to note that there are wide deviations in the recovery rates based on the defaulting company’s industry, and the nature of the circumstances that led to the company entering into financial distress. A rule of thumb is to assume that recovery will be 50 percent, or for a more conservative estimate to assume recovery of 40 percent, which would make the loss-given default factor 60 percent.
The potential exposure is the size of the credit outstanding at the time of a credit event occurring. For a straightforward instrument, such as a fixed coupon bullet bond, the exposure is simply the face value of the bond. However, if the bond has sinking fund or repayment features then it is obvious that the calculation of the exposure becomes more complicated. If the outstanding balance of the credit instrument can vary then the potential exposure amount is also a function of the credit event’s timing. Calculating the extent of the potential exposure becomes progressively more complicated in a corporate setting where the size of the outstanding credit allowance or receivable is likely to fluctuate depending on the borrower’s buying cycle or working capital cycle and any changes in credit policy of the creditor. Foreign currency fluctuations and changes in market value also add extenuating complications in the calculation of potential exposure.
To simplify the potential exposure calculation, many corporations adopt the policy of assuming that the exposure is the size of the maximum allowable credit limit granted to the counterparty. This conservative assumption is probably quite realistic because a customer in financial trouble is likely to maximize all available sources of financing, and this, of course, would include maximizing their accounts payable to their suppliers.
In the absence of a credit policy and credit limits, it is best to attempt to measure the peak exposure throughout the customer’s buying or working capital cycle. In ideal circumstances this would be measured based on sales projections for each client, but the actual sales to the customer are likely to be correlated with their financial health. As a customer starts to worry about his financial health, he orders fewer goods, which, in turn, leads to stock outs or dated inventory, resulting in less customer satisfaction and sales, which, in turn, leads to worsening financial health, and the “credit death spiral” begins.
Measuring the probability of default at first glance appears to be an objective exercise and there are several well-established methods for measuring the probability of default. Simplistically, these methods can be divided into measures based on fundamental analysis, statistical analysis, and market-based methods. Each of these methods, however, tends to have large changes in value or measurement based on the period of analysis. Additionally, when using these methods, remember that all credit events are unexpected and tend to be caused by sudden and unforeseen changes in circumstances. It is conceptually and practically difficult for either a fundamental or a quantitative model to capture these unforeseen and unique events.
Fundamental Analysis of Credit Default Risk (Probability of Default)
The most basic method to assess the creditworthiness of a company has been the “Five Cs” of credit analysis: (1) capacity, (2) capital, (3) collateral, (4) conditions, and (5) character.
Capacity is the ability of the company to pay its obligations out of the cash flows generated by the business. Capacity is generally assessed by examining various accounting ratios such as the coverage ratio (which will be explained and discussed later in this section). Capital is the amount of cash that the company has on hand, while the collateral is based on the quality of the assets that can be sold in order to repay obligations if a credit event occurred. Conditions refer to the general business conditions that are specific to the company and its industry. Finally, character refers to the willingness of the company to pay its obligations in a timely manner. Basically character comes down to the reputation and integrity of the firm and its management.
Ability-to-pay measures are fundamental techniques based on accounting statements to assess creditworthiness. These accounting measures look at the short-term ability to cover exposures, longer term financial flexibility, and finally the safety buffer that the firm has in managing its cash flows.
Two frequently used accounting ratios to measure the short-term ability to pay are the Current Ratio and the Quick Ratio (also called the Acid Test).
These two ratios show the relative amount of assets that the firm could conceivably convert to cash quickly relative to the amount of liabilities that the firm is conceptually expected to pay in the short term. The Quick Ratio is a more conservative measure because it assumes that the value of the firm’s inventory would not be available to turn into cash to pay liabilities in the times of a crisis. Additionally, if a firm is having financial difficulties it is reasonable to assume that the value of its inventory will be significantly impaired.
A third metric of short-term credit stability is the Burn Rate or the related measure Days Cash on Hand. The Burn Rate is the amount of expenses that the firm incurs on an average day and the Days Cash on Hand is the number of days that the firm can continue to pay its expenses without generating any sales.
These two measures are used to measure the credit risk of doing business with a new firm that is still in the product development stage and does not yet have a saleable product. The Days Cash on Hand provides the time period the firm has to either receive cash through developing a product and customer-related inflows or the number of days it has to raise an additional round of financing to continue operations.
The debt ratio is calculated to ascertain the long-term financial stability and flexibility of a firm. There are many different forms of the debt ratio, but the most common are as follows:
or
These two ratios show the level of leverage and financial flexibility in the firm’s capital and operating structure. It is important to realize that each industry will have different average debt ratios that are a function of the level of riskiness in that industry. Generally, companies that are in high capital-intensive industry with stable cash flows (for example, utility companies) tend to have higher debt ratios.
Another long-term measure of the credit risk of a company is the Coverage Ratio.
EBITDA is the earnings before interest and taxes, with depreciation and amortization added back. In essence, EBITDA is a proxy for the cash flow of the firm. The denominator of the above expression is the total contractual payments that a firm must make within the accounting period. Therefore, the Coverage Ratio is the ratio of the cash being generated by the firm divided by the amount of the contractual payments. A large Coverage Ratio implies that the firm has a large buffer of cash being generated to make its payments.
Accounting statements are historical in nature and it is usually several months before they become public. Therefore, when examining the credit risk it is important to pay as much attention to the trend of the ratio as it is to the value of any given ratio at a given point in time. Although the ratios are likely to fluctuate with both operating and business cycles, it is still possible to spot potential troubles early by examining changes in the ratios over time. For example, an increasing Debt Ratio and a declining Coverage Ratio would be signs of decreasing financial health and financial flexibility even though each of the numbers for the latest period by themselves might be within an acceptable range.
When examining accounting statements it is best to examine the ratios of a given company against its industry peers. This way trends in the industry can be separated out from trends in the financial health of the company. It is almost meaningless to judge the credit quality of a company by its accounting ratios on its own. It is only within the context of the industry and the trends through time that the changes in credit quality can be accurately evaluated.
A second and simplistic measure that is often used as an early warning sign of credit risk is to measure any significant and unexplained changes in the time it takes for a counterparty to repay. The stretching of payments by a customer could mean a change in their working capital policy, but more likely it is a sign that there are cash-flow problems that could escalate to a credit crisis.
A final method of fundamental analysis is to use credit ratings as published by the various credit-rating agencies such as Standard & Poor’s, Moody’s, or Fitch. These rating agencies are paid by the company being rated to conduct ongoing analysis of the firm’s creditworthiness. The ratings themselves are widely available while company specific reports are available to subscribers.
Credit ratings by the various agencies are quick and easy to use. A key feature of using credit ratings is the ability to relate the rating to the large databases maintained by each of the rating agencies. These databases give the probability of default for a given period of time and also include transition matrices that show the probability of a credit “migrating” to a different rating. Exhibit 15.2 shows a Transition Matrix from Moody’s as reported in Hull.3
As Exhibit 15.2 shows, a company that begins the year with a Baa rating has an 88.70 percent probability of finishing the year with a Baa rating, a 4.60 percent probability of finishing the year downgraded to a Ba rating, and a 0.19 percent probability of defaulting within a year.
Exhibit 15.2 One-Year Transition Matrix—Percentages Moody’s 1970–2006
| Initial Rating | Rating After One Year | ||||||||
| Aaa | Aa | A | Baa | Ba | B | Caa | Ca-C | Default | |
| Aaa | 91.56 | 7.73 | 0.69 | 0.00 | 0.02 | 0.00 | 0.00 | 0.00 | 0.00 |
| Aa | 0.86 | 91.43 | 7.33 | 0.29 | 0.06 | 0.02 | 0.00 | 0.00 | 0.01 |
| A | 0.06 | 2.64 | 91.48 | 5.14 | 0.53 | 0.10 | 0.02 | 0.00 | 0.02 |
| Baa | 0.05 | 0.22 | 5.16 | 88.70 | 4.60 | 0.84 | 0.23 | 0.03 | 0.19 |
| Ba | 0.01 | 0.07 | 0.52 | 6.17 | 83.10 | 8.25 | 0.58 | 0.05 | 1.26 |
| B | 0.01 | 0.05 | 0.19 | 0.41 | 6.27 | 81.65 | 5.17 | 0.75 | 5.50 |
| Caa | 0.00 | 0.04 | 0.04 | 0.25 | 0.79 | 10.49 | 65.47 | 4.44 | 18.47 |
| Ca-C | 0.00 | 0.00 | 0.00 | 0.00 | 0.46 | 2.78 | 11.07 | 47.83 | 37.85 |
| Default | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 100.00 |
A company that has a rating of Baa or better (BBB or better using the Standard & Poor’s or Fitch rating systems) is considered to be an investment-grade credit, while companies with ratings Ba or BB and below are called interchangeably noninvestment grade, or high-yield, or junk bonds.
Although ratings are important in calculating the creditworthiness of a counterparty, it is also imperative for a company to maintain a keen focus on its own credit rating. The pricing of debt issuances (and by association the cost of debt) is closely correlated to the company’s rating. A low rating will most certainly imply a higher cost of debt, and may result in the amount of financing available in either the debt or equity markets being limited.
Although easy to use, credit ratings are not without their drawbacks. Credit rating agencies have come under fire in recent years for not changing their ratings frequently enough to capture the changing dynamics of a given company. Rating agencies counter that they look at the credit risk of a company based on the whole business cycle, as opposed to just a point in time. Additionally, they examine the amount and quality of a company assets in an effort to assess recovery rate as part of their analysis. Thus, the rating should not be taken as strictly a moment-by-moment assessment of the default probability. A further component to the “stickiness” of the ratings is that the rating agencies wish to avoid unnecessary volatility in the bond trading of the companies they are rating. Many institutional investors have strict limits on their holdings of noninvestment grade bonds. A frequent flip-flop of a rating between investment and noninvestment grade would introduce a large amount of trading into and out of a company’s bonds, which is obviously quite undesirable. To allow the market time to adjust in an orderly manner to a likely change of rating, the rating companies will issue rating warnings that signal that the company is under review for a positive or negative rating change.
Market-Based Analysis of Credit Default Probability
There are two main methods of measuring default risk using market-based measures. The first is to examine the yields to maturity on a corporation’s debt issues, while the second method is to examine the price of a credit default swap based on the corporation.
The yield to maturity (or more accurately, the credit spread, which is the amount by which the yield to maturity of a risky corporate bond is above the yield to maturity of a similar risk-free Treasury bond that has an equivalent time to maturity) is the traditional measure of the market’s perception of the probability of default of a company. The wider the credit spread the greater the perceived risk of the corporate bond suffering from financial distress. The yield to maturity is affected by the general level of interest rates, so it is quite important to look at the credit spread rather than simply the yield to maturity.
The yield to maturity and the credit spread will generally be affected by the structural features of the bond, or more specifically by any callable, put-able, convertibility, redeemable, extendable, or other embedded options in the bond. To account for the effects of embedded options on the bond’s yield to maturity, a measure called the Option Adjusted Spread (OAS) is used. The OAS adjusts the yield to account for any embedded options and calculates a credit spread that would exist if the bond did not have any embedded option features. Using OAS allows one to compare spreads between conventional plain vanilla bonds and bonds with embedded option features.
The credit default swap market is a relatively new market. In a credit default swap, a counterparty to a trade, called the protection buyer, will pay a periodic fee (generally semi-annually), which is called the credit default swap spread, to the second party, which is called the credit protection seller. This fee is based on a notional amount. In return, the protection seller makes a payment to the protection buyer if and only if the underlying credit obligation (generally a publicly traded bond or a syndicated loan) suffers a credit event such as a bankruptcy or a failure to pay. The payment is generally based on the notional amount multiplied by one minus the recovery rate on the underlying credit obligation. In simple terms, the protection buyer is buying insurance against a credit risk event.4
There are a large numbers of hedgers, speculators, and market makers in the credit derivative market. The large volume of trading (larger even than the trading of the underlying bonds and loans) produces a dynamic market that provides instantaneous assessments by the market as to the credit quality of the underlying corporations that are traded. Since the payout of a credit default swap is directly related to a credit event, and, for instance, not related to interest rates, the credit default swap price is a direct reflection of the probability of a company’s default. Credit default swaps are also one of the primary methods by which a company can hedge its credit risk exposure to other companies, which is a topic that will be covered later in this chapter.
Another market-based measure of assessing credit risk was developed by Moody’s KMV.5 This proprietary method, based on the Merton Model (an option-pricing model), models the equity of a firm as a call option written by the bond-holders of the firm. To see why this is so, consider the case of a company that goes bankrupt. Although the equity holders lose all of the value that they had when they purchased the shares of the company, they are not responsible to make any more payments, and thus their downside is limited to the amount that they paid for their shares. On the upside, however, the value of the shares could rise if the fortunes of the company increase, and conceptually the profits accruing to the shareholders are unlimited. Thus, the payoff to the shareholder is similar to that of the payoff to the holder of a call option—the loss is limited to the amount of the premium paid, and the upside is conceptually unlimited.
Moody’s KMV calculates an Expected Default Frequency (EDF) by utilizing the Merton Model of the firm. It calculates the asset volatility of the firm (implied by share prices) and the known market value (both equity market value and debt market value) of the firm. The asset volatility can be used to calculate the probability that the market value of the firm falls below the debt level of the firm. When that happens the firm is assumed to be bankrupt. These calculated levels of default are then correlated to empirically observed levels of default to create the EDF of the firm.6
The advantage of EDF measures are that they are dynamic, being based on the current market assessments of firm value and volatility. Additionally they are a forward-based measure as the market inputs are also forward-based.
Statistical-Based Models of Credit Risk
Credit scoring is a statistical-based method of estimating credit default risk. Credit scoring develops a set of factors from readily observed characteristics, each with a specific weighting that when added together provide a score by which to rank a company’s creditworthiness. The most well-known credit scoring model is the Altman Z-Score, which is given by the following formula:
Where:
If the calculated Z-Score is below 1.80, then there is a high probability of the company encountering financial distress. If the Z-Score is above 2.99, then the company is considered to be a safe credit risk. Values of the Z-Score between 1.80 and 2.99 are considered to be in the gray zone.7
For different types of situations and different types of credits, different credit scoring models have been developed. For instance, for consumer credit, credit scores are calculated based on factors such as the length of time that the individual has been in their current employment, length of time in current residence, level of education, income, current debt levels, and past history of late payments.
The major drawback to credit scoring models is that they are dependent on a large amount of historical data. Without a large database, the statistical validity and reliability of a model is in serious question. An additional complication is that a credit scoring model assumes a large portfolio of credits under consideration. With a large portfolio of credits under examination, the statistical properties of a credit scoring model are more likely to be borne out. Since a credit scoring model is a quantitative model, it cannot account for company-specific effects or events, and thus relying on a model for a small portfolio of credit accounts may lead to misleading conclusions. This is why credit scoring is frequently used for securitizations of packages of credit cards or accounts receivables that involve a large number of credits.
Credit Risk Mitigation
As mentioned earlier there are three forms of credit risk: (1) customer credit risk, (2) sovereign risk, and (3) funding risk. Additionally, liquidity risk may be considered a special case of funding risk, as in, for example, the role of liquidity in the credit crisis that began in 2007. Mitigation of each of these risks will be discussed in turn.
Customer credit risk mitigation is based on setting policies such as the size of exposure to be taken with a given counterparty, the terms of repayment—including time to repay and any interest charges incurred, and whether collateral or partial prepayment will be required.
As mentioned earlier, extension of credit to customers is frequently used as part of a company’s marketing package. The more liberal the credit terms offered to clients, the larger the expected sales, but also the larger the expected losses from credit risk. Additionally, there is an adverse selection aspect to extending liberal credit terms to customers because the clients with the most desperate credit situations are likely to take full advantage of credit. Therefore, the credit policy of a company toward clients is a major strategic risk decision that involves both the financial and marketing sides of the organization.
Generally companies will offer some form of credit to clients in order to facilitate sales and the practicalities of doing business. The most usual form is to offer a certain number of days to pay, with perhaps a set discount for early payment. However, if it is deemed upon analysis that the customer has a low likelihood of paying, or the customer has a poor track record of paying in a timely manner, then credit terms may be refused with cash on delivery being demanded, or even payment—full or partial—being required before an order is processed. Refusing a client credit may result in the loss of the client, which could lead to a competitor being stuck with credit losses to that client.
Another reason to reject credit to clients is that it is costly in its own right. A firm that extends credit has to secure additional financing to pay its own expenses that may need to be covered before the customers pay. A company needs to make sure that its extension of credit to customers does not impair its own working capital cycle, and by extension impair its own credit risk. Frequently, companies get into credit trouble on their own by extending favorable credit terms to customers, which in turn increases sales, but also increases the need for working capital. Thus, the firm gets into credit difficulty not through lack of sales, but through letting its credit policy dictate and dominate its working capital cycle and short-term financing flexibility.
There are a couple of techniques to manage credit receivables to clients besides tightening credit terms. If a company has sufficient receivables it can package the receivables into a structured note and sell it off to market investors in a securitization. This is the preferred method of receivables management by large national manufacturers that have a large and diverse group of customers who make credit purchases. Perhaps the best known examples of this are the financing arms of General Motors (General Motors Acceptance Corp., better known as GMAC) or Ford (Ford Motor Credit), which, on an ongoing basis, package auto loans into securitized notes. Note that in most cases a securitization will be without recourse, which means the risk of customer defaults is borne by the buyer of the securitization notes. Thus, the issuer of the note (for example, GMAC) not only receives financing by selling the loans, but also sheds the credit risk associated with the customer loans.
A second method is for the company to sell their receivables to a special purpose company called a factor company. A factor company essentially acts as an investor to buy the receivables of a company. A factoring transaction is much like a small-scale securitization that is bought by one investor, the factoring company. A company can sell its receivables with or without recourse. If the company sells its receivables with recourse, which means that the factoring company can come back to the company to recover any credit losses from defaulting customers, it will receive financing but will obviously not shed the credit risk of customers who default. Factoring, especially without recourse, is generally considered to be an expensive form of financing.
A second aspect of customer credit risk management is to decide in advance what actions will be taken if a customer fails to make timely payments. Will the company continually phone the customer, hire a collection agency, or start legal actions? What will be the sequence of actions, and what will be the level of outstanding credit that will trigger each set of responses? Setting a policy of when and how each of these actions will be taken in advance can save a lot of stress and management time when the inevitable event actually occurs—especially when a large number of customers and transactions are involved.
Another way to manage customer credit risk is through the use of credit derivatives. As stated earlier, in a credit derivative transaction the protection buyer pays a periodic fee called the credit default swap spread to a counterparty, which is generally a well-known financial institution. In return the financial institution makes a payment based on the recovery rate to the protection buyer if the underlying credit incurs a credit event, such as a default or bankruptcy. Credit derivatives are available on most well-known public companies that have outstanding bonds or syndicated loans. However, credit derivatives are generally not available on smaller or private companies.
Credit derivatives are conceptually simple products used to mitigate credit risk. However, on closer inspection, there are many difficult practicalities to using credit derivatives effectively. The first is the bulky nature of credit derivatives. Credit derivatives are generally sold in $10MM notional amounts, which means that their size is out of reach for all but the largest of customer credit accounts. The second aspect is that credit derivatives usually have a five-year maturity, which may be longer than a corporation wants to forecast its credit exposures going forward. An additional issue is that the outstanding credit risk to a customer is likely to fluctuate throughout the business cycle. This means that during parts of the credit cycle the hedging company is likely to be over-hedged, and at other times is likely to be under-hedged as the size of the credit exposure keeps changing. The final concern with using credit derivatives to hedge is that of structuring the credit derivative contract so that it matches the nature of the exposure in the event of a default. Most credit derivative contracts are based on the recovery rate of an outstanding bond. The recovery rate on a bond might be different than the recovery rate experienced by a corporation trying to recover monies owed on a receivable.
Assuming that the exposure is large enough, credit derivatives are useful tools to hedge the exposure to a sovereign. Credit derivatives are available on most sovereigns that corporations may be concerned about. Credit derivatives on sovereigns work the same as credit derivatives on corporations with the exception that additional events of credit risk, such as moratoriums and repudiation, which are more specific to sovereigns, are included in the clauses of default that trigger the payment on a transaction.
The technique of financing a large capital investment in a risky sovereign nation by financing the capital investment with monies from foreign nationals of the country was discussed earlier. This is a natural way to hedge an exposure, assuming that the corporation itself has enough name recognition in the foreign country to attract investors.
When dealing with the credit risk of foreign customers, letters of credit from a governmental agency such as the Export-Import Bank in the United States or the Economic Development Canada agency are used. These governmental agencies are set up to promote international trade by taking on many of the credit risks that are difficult for traditional financial institutions to assess and manage. These agencies provide letters of credit or credit insurance against the default of a foreign customer.
With all of the customer and sovereign credit risk mitigation techniques discussed so far there has been an explicit or an implicit cost involved. As with almost all types of risk management, there is a cost involved in reducing risk. By tightening credit terms, the firm reduces credit losses but also likely experiences a reduction in sales. By factoring receivables without recourse, the company will have to pay an implicit financing charge as well as a credit risk charge. By entering into credit derivatives companies will have to pay an explicit fee as well as carry significant risk in the transaction.
Hedging credit risk is a difficult management task. Effective credit risk management involves a host of trade-offs. The fact that credit risk is event-based makes it difficult to accurately track and efficiently protect oneself.
The final aspect of credit risk management is to ensure that one does not get into credit difficulties and that external financing is readily available at reasonable and competitive costs. This is what may be called the funding risk of a corporation.
The funding risk is a complex mix of the industry the company is in, the state of the capital markets in general, the name recognition of the company in both domestic and foreign markets, and both the perceived and actual financial health of the firm, which in turn is a function of the profitability, operating policies and capital structure of the firm.
The best way to mitigate funding risk is to operate as an efficient and profitable firm with large cash flows. In a competitive industry, however, that is obviously much easier said than done. A firm can mitigate its funding risk mainly by maintaining a conservative capital structure, maintaining a strong relationship with a portfolio of financial institutions, and by diversifying its funding both geographically and in the types of funding it seeks. Each of these elements, however, entails a trade-off in terms of time and or convenience. For example, a more conservative, less leveraged capital structure tends to be more expensive on an after-tax basis due to the tax efficiency of funding a corporation with debt. Partnerships with more banks lead to more complex relationships, and funding in different countries increases regulatory reporting requirements and costs.
Ultimately, successful funding risk management can lead to increased opportunities in both marketing opportunities and strategic opportunities—especially in times of economic downturns when funding is harder to find. Companies with financial flexibility can offer more generous credit terms than their competitors, can engage in more aggressive pricing policies, and can undertake more developmental opportunities for strategic gain if they have a funding advantage over their rivals. Indeed, funding imbalances among competitors during times of economic downturns often lead to acquisition activities with companies with the competitive funding advantages taking over their funding constrained competitors.
AN ANALYSIS OF THE CREDIT CRISIS
There are many proposed theories of the causes of the recent credit crunch, and several of the features leading to the collapse of the U.S. subprime lending market along with the market value of collateralized debt obligations (CDOs) hold many interesting and valuable lessons for students of risk management.
To understand the situation it is best to start with examining the U.S. credit, housing, and investment markets at the turn of the millennium. The U.S. housing market was by all anecdotal accounts a stable and growing market. House ownership was seen as a cornerstone of a prudent personal investment strategy. Demand for housing was strong and prices were steadily rising. Additionally, interest and mortgage rates in general were stable, consistently low, and falling as the Federal Reserve kept interest rates at historically low levels. A recurring strategy for individuals was to refinance their homes at each major lowering of interest rates. Furthermore, the lower interest rates, combined with rising housing prices, encouraged families to increase personal leverage and buy larger homes. The refinancing of homes at higher levels of leverage created liquidity because at each refinancing most homeowners would monetize the increase in their house values, as well as monetize their increasing level of leverage. This created a liquidity glut within the United States, which when combined with a matching global increase in liquidity, only reinforced the factors keeping interest rates low.
The low levels of interest rates were not a boon, however, for all, and in particular institutional investors such as pension funds, insurance companies, and endowment funds. These investors relied on relatively high interest rates on their fixed-income investments to be able to meet their future financial obligations. This created a demand for more highly structured instruments that held the promise of higher yields.
A final component to the context of this complex puzzle was the changing regulatory and competitive environment in the financial services sector. The international regulatory capital agreement known as Basel II was proposing changes in how much regulatory capital banks had to set aside as reserves in order to ensure their solvency. The essence of the proposed changes in the rules was that banks had a strong incentive to shed their credit risk by selling off credit risk in whatever form they felt best. An additional incentive of the regulatory reform was for the banks to learn how to better model credit and market risk. This in turn led to a flurry of research and development activity in mathematical models.
The environment was set for the introduction of CDOs. The structure of a basic CDO is shown in the following diagram. A CDO consists of an asset provider, generally a bank that has loan assets it wishes to shed. The bank sets up a special purpose vehicle (SPV). The sole job of the SPV is to act as a legal gateway between the bank and investors. The SPV either buys outright the package of loans from the bank, or more generally purchases the credit risk from the bank by utilizing a series of credit default swaps (CDS), or total return swaps (TRS) linked to each of the underlying credits in the portfolio of credits. The SPV in turn issues a series of structured notes to investors. The series of notes are issued so they create what is known as a “waterfall” structure, such that each tranche carries a different level of risk and a correspondingly different yield. The proceeds from the sale of the notes are generally invested by the SPV in risk-free Treasury Securities. See Exhibit 15.3.
At each period the bank collects the principal and interest payments from the loans in the portfolio, and pays the relevant CDS spread to the SPV. The SPV in turn makes any required payments on default to the bank. With the remaining CDS fees, and proceeds from interest on the Treasuries, the SPV makes payments to the note holders. Holders of the senior tranche (frequently called the Super-Senior as it was considered to be ultralow risk) have their promised yield paid first. Remaining monies are then used to pay the next tranche (generally an investment-grade tranche), and so on until any remaining proceeds are paid to the lowest tranche (which is considered to be equity because it was to receive any residual payments or value only after all other tranche holders received their full payments).
In the early days of these structures, there was strong demand for the high-grade tranches because they carried attractive yields and were considered to be low risk. The lower tranches, and in particular the equity tranches, were difficult to sell. In fact, the nickname for these tranches was “toxic waste” as they would be the first to suffer losses if any of the credits in the portfolio defaulted and the SPV had to make CDS default payments back to the bank. The banks could not readily sell the equity tranches, so they often had to keep them. In other words, the banks, in trying to reduce their risk, sold off that risk through the CDO, but they had to keep the worst of the credits (that is the first to default) for themselves. One positive effect of the banks keeping the equity tranche is that it gave investors confidence of the stability and safety of the more senior tranches. In other words, if the bank created a poor portfolio, they (the banks) would be the first to suffer losses due to their ownership of the equity piece.
Exhibit 15.3 Structure of a CDO
The ironic history is that the default experience of companies in the early days of CDOs was incredibly benign. There was a lot of liquidity, and companies could easily borrow extra money and thus stay afloat. Indeed, the rise of CDOs made banks more willing to grant loans at more favorable terms and to weaker credits because they knew they were at least partially hedged through the increase in their credit risk management knowledge and in the use of instruments such as CDSs and CDOs. Since corporate defaults were low, this implied that the returns of the equity tranches of CDOs were much higher than expected.
The situation for the banks was that by creating CDOs, they were lowering their regulatory capital requirements by shedding credit risk, making large returns off their ownership of the equity tranches of the CDOs they created, satisfying their customers by creating higher yielding securities, and generating large issuance and service fees by creating these structures. This was the backdrop and context that created the credit crisis via interwoven systemic credit risks.
As the market for CDOs developed, the demand by institutional investors for higher yielding investments increased. To satisfy the demand for higher yielding assets, the banks sold lower rated tranches, including the equity tranches of the CDOs and also increased the risk characteristics of the underlying pool of assets. This introduced subprime mortgages into the pool of assets underlying the CDOs. Additionally, banks actively started taking on new sources of credit risk by selling protection via credit default swaps in order to increase their inventory of credit assets that they could repackage into CDOs.
Initially the subprime mortgages performed reasonably well, and this led to more aggressive assumptions when modeling the default history of these relatively new assets. Due to the strong investment performance of CDOs, and the increase in sophistication of the mathematical models underlying their valuation, investors became more comfortable with CDOs as assets, even though the valuation and structuring of them was complex and understood by only a small proportion of investors.
The start of the crisis began slowly and almost imperceptibly with an increase in the default rate of subprime mortgages. The housing market started to slow down in general, along with the general economy, which led to an increase in the number of defaults on conventional mortgages as well. The effect of this was to make the payment rates from the SPVs to the sponsoring banks increase in a number of the CDO structures. By itself this was not a problem because it only directly affected the lower tranches of these CDOs. However, even the investors in the higher tranches of CDOs began to question the valuations of their investments.
A catalyst for the crisis becoming full-blown was when two financial institutions had a dispute about the value of a specific CDO that was posted by one of them as collateral to the other bank. The bank holding the collateral demanded more collateral from its counterparty who countered that the value of the CDOs held was more than sufficient to satisfy the terms of their collateral agreement. The bank holding the collateral disagreed and threatened to sell the underlying CDOs to prove its point. This dispute, which was played out in the media, made several major institutional investors question the value of the CDOs that they held in their own portfolios. This was probably the first time that many of these investors had seriously tried to objectively price CDOs, and thus the first time that the complexity and sensitivity to assumptions in pricing were fully appreciated.
These exercises in valuation led to most of these investors deciding that it was prudent to significantly reduce their holdings of CDOs until they were better understood. The financial institutions were now holding significant portfolios of CDOs themselves, and significant holdings of credit risk in their inventories that they were planning to structure into new CDOs for the market. The dynamics of the market, and an effort to make their clients happy and defend the value of their credit inventories, forced many of the banks to create a market in the CDOs by agreeing to buy back the CDOs from their clients. These efforts to create liquidity and bolster the market did not work, and only led to the banks holding even larger amounts of credit assets that were falling rapidly in value and that proved to be nearly impossible to sell off at anything resembling a reasonable price. Ultimately, confidence was lost in the models for valuing and structuring CDOs, confidence was lost in the underlying assumptions, and the critical blow was that confidence was lost in the liquidity of these complex instruments.
There are many relevant lessons for the management of market and credit risk that can be gleaned from the credit debacle. The first and primary lesson is that of model risk. The world does not work by models. A model is at best a map, and just as a map is not the same as an actual highway, a model is not the same as the actual actions of traders. Models gain acceptance because they seem to work. Indeed the models for credit risk worked well as long as the credit risk in the markets were benign, which they were in the lead-up to the credit crisis. Although there were a couple of major credit incidents (Enron, WorldCom, Parmalat, and Delphi) in the years before the credit crisis, they were thought to be isolated and specific events, and thus did not interrupt the broader credit markets. Additionally, the level of personal bankruptcies was also relatively low as the U.S. economy prospered. However, when cracks in the default rate rose, and in particular mortgages, it showed that assumptions about default rates and recovery rates (the prices at which the foreclosed homes with defaulted mortgages would be sold) were seriously underestimated. Ultimately the lesson learned is that it is critical to understand how a model reacts in stress situations. A model may work fine when things are normal, but then again almost any model will work fine when the economic situation is positive and the volatility low. It is how the model works in times of uncertainty and stress that determines the success of a model.
The second lesson to learn from the credit crisis is that conditions change. Trends do not last forever, and eventually economic prices and rates may revert. Investors gained false confidence in falling interest rates, falling default rates, and high recovery rates on defaulted assets. Also, the uncertainty in the inputs to the model were not fully appreciated and understood, especially in the case of the subprime mortgages because it was such a new market.
A third lesson to learn is the need to understand the underlying dynamics of a model. The mathematics behind CDO and credit risk is complex and questionable at best. However, many investors took the mathematicians at their word even though the investors did not understand the components of the model or how the model components were put together. Investors abandoned their intuition about the markets for the mathematical rigor of the models. Conversely, the modelers who constructed the models for the most part did not fully incorporate real-life market dynamics and intuition into the model. Most of the modelers were mathematicians who had little to no trading experience. The investors who had the trading experience had little to no understanding of the mathematics.
A subtle but important effect that the credit risk and market models did not correctly incorporate was the correlation of default. A mathematical technique from the insurance industry called Copulas was used to model the correlation of default, which is a key component of valuing and trading CDOs. The Copula model, which is based on actuarial science, works well when a large number of investments are being made (such as a large number of life insurance policies being underwritten). However, in the case of CDOs, although the underlying package of credits was large, generally only a small number of CDOs were being purchased by any one investor. Second, it is not clear that the subtleties of default correlation are fully understood by either traders or mathematicians. For instance, if General Motors suffers a credit event, what does that imply about the change in the probability of default for Ford? One could argue that it implies an increase in the probability of default since the default of GM is obviously a sign that the car industry is in distress. Conversely, one could also argue that it implies that Ford’s probability of default has gone down since one of its major competitors is hobbled by a default.
A related aspect to correlation risk is the presence of feedback loops. In the credit cycle, consumer confidence in housing prices increased, which led to the buying of more and larger homes. This increased prices, which in turn increased investors’ confidence in lending money to home buyers. This created more credit, increased the incentives and availability of credit to buy larger homes, which in turn fed back into higher prices and the cycle continued upward. On the downside, as the economy started to turn, it become harder to refinance houses, and home buyers could no longer finance their increased leverage levels. This led to an increase in defaults, which led to investors pulling back from supplying credit to the housing market, which in turn led to lower prices, which led to more defaults until the cycle accelerated on the downside.
Although the market for CDOs was strong, liquidity in the market was high. This increased the confidence of even skeptical investors to enter the market as assurances of market liquidity provided an escape route if things went sour. However, at the exact moment that the need for liquidity was the greatest, the liquidity dried up as the market became a one-way market of sellers. Although several investment banks attempted to prop up the market through acting as market makers, this only increased their own risk and liquidity issues. As with models, it is likely that liquidity will always be there in normally functioning markets. However, when markets get distressed both liquidity and models tend to disintegrate, and indeed may negatively feed off each other. The practical lesson is that liquidity assumptions need to be aggressively stress-tested. The more complex or inexperienced the market is with a product, the more aggressive the stress-testing should be.
Ultimately, what crippled the market and created the crisis was the crisis of confidence. As investors became concerned about the valuation models and assumptions underlying their investments, they sold indiscriminately, which created a mini-panic in the market. The reliance on models and market liquidity may have gotten ahead of itself before the credit crisis, but an argument can be made that the pendulum has swung equally far in the opposite direction with an unwarranted extreme lack of confidence in risk management models becoming the norm.
CONCLUSION
Credit and market risk management must be an integral part of a firm’s enterprise risk management strategy. Not only are credit and market risk important variables in the profitability of the firm, but they are also the risks for which well-developed methods of analysis and management exist. Despite the plurality of methods for managing these risks, it takes a combination of a clear strategy, a knowledge of the analytical tools, an understanding of the risk management instruments, responsible oversight and direction from senior management and the board, and perhaps most importantly, the ability to think clearly, creatively, and intuitively in order to balance the art and the science necessary for success in this branch of risk management.
NOTES
REFERENCES
Altman, E. 1968. Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. Journal of Finance (September) 189–209.
Caouette, J.B., Altman, E.I., Narayanan, P., and Nimmo, R. 2008. Managing credit risk: The great challenge for global financial markets, 2nd ed. Hoboken, NJ: John Wiley & Sons.
De Serviguy, A., and Renault, O. 2004. The Standard & Poor’s guide to measuring and managing credit risk. New York: McGraw-Hill.
Hull, J.C. 2008. Options, Futures and Other Derivatives, 7th ed. Upper Saddle River, NJ: Prentice Hall.
Meissner, G. 2005. Credit derivatives: Application, pricing, and risk management. Hoboken, NJ: John Wiley & Sons.
ABOUT THE AUTHOR
Rick Nason, PhD, CFA, has an extensive background in the capital markets and derivatives industry having worked in equity derivatives and exotics, credit derivatives and capital markets training in a senior capacity at several different global financial institutions. Rick is a founding partner of RSD Solutions, a risk management consultancy that specializes in financial risk management consulting and training for corporations, investment funds, and banks.
Dr. Nason is also an Associate Professor of Finance at Dalhousie University in Halifax, Nova Scotia, where he teaches graduate classes in corporate finance, investments, enterprise risk management, and derivatives. He has been awarded several different teaching awards as well as being selected MBA Professor of the Year several times. His research interests are in financial risk management, enterprise risk management, and complexity.
Rick has an MSc in Physics from the University of Pittsburgh and an MBA and a PhD in Finance from the Richard Ivey Business School at the University of Western Ontario. He is a Chartered Financial Analyst charterholder. In his spare time he enjoys practicing risk management principles as he plays with his collection of pinball machines.