Incidents of default Vivien BRUNEL
In this article, we are going to focus on single corporate obligor defaults, but the way to model defaults, default probabilities and loss severity are already the subject of several articles in this encyclopedia (see related entries below). Our aim here is rather to provide with a more economical viewpoint through the statistics of defaults and to describe the drivers of the incidents of default without describing the famous bankruptcies of the last few years. This article is organized as follows: first we deal with corporate defaults and recoveries, show the statistics and explain the drivers of corporate defaults. Second, we give some details on the way the rating agencies measure credit quality and/or creditworthiness. We often define an incident of default as a missed or delayed payment of interest and/or principal or bankruptcy. However, in spite of a precise definition, the reality of the incidents of default is less clear, and in particular, it strongly depends on the type of obligation the debtor has and on the commercial link between the lender and the debtor. In particular, when a bank contracts a loan with a client, the bank mustn’t beak the confidentiality requirement. A failure to meet a loan repayment may be a private event between the bank and the client. Default risk has two components: default likelihood and default severity. Default probabilities aim at measuring the likelihood of incidents of default, or from the reverse point of view, the surprise the investor faces when observing a default of a given obligor. After the default, the severity drives the loss amount on the defaulted instrument. The sensitivity of defaults to the global economy is clearly shown by historical default rates series. Moody’s (Moody’s, 2004) and Standard & Poor’s (Standard & Poor’s, 2004) statistics for high yield default rates show the very high correlation of default rates to the economic cycles (Figure 1), meaning that one of the main drivers of default events is the global state of the economy. S&P Moody's GDP
12% 10% 8% 6% 4% 2% 0% 1980
1990
2000
2010
This view is supported by many statistical studies. The link between default rates and Gross Domestic Product (GDP) is perfectly clear, and both variables are highly correlated. An empirical study from data over the last 20 year confirmed that default rates and GDP had a negative correlation of order of –80% (First Union Securities Inc, 2001). The predictive power of future default rates by estimates of GDP is thus very strong. Even if the state of the economy is the main driver at the portfolio level, everyone agrees that the drivers at the firm level are firm specific. There are several firm specific factors that have an impact on the ability of a firm to default. An obvious one is the level of the debt which, combined to the market conditions (interest rates), determine the total cost of the debt. This is the very reason why academics focus a lot on capital structure models. However the indebted firm is not only vulnerable because of its liabilities, but also because of the expected cash flows and their volatility. Actually, when the economy is healthy, leverage ratios increase and cash flow multiples are getting worse. Then, a firm with bad ratios is less likely to face a contraction of the economy and is less able to refinance on the markets at an interesting price. This is exactly what happened in
1987 when leverage ratios reached very high values after years of a sustained growth, and just before a large depression. Of course, this situation was also created by an aggressive lending strategy from the banks. Even if the leverage ratios of 1987 have never been observed since that time, what happened in the 1990s was also an decrease of the quality of leverage ratios and cash flow multiples up to the year 1997. Unfortunately, the end of the 1990s where characterized by a decrease of the GDP and by an irrational behavior of the financial market around the technology values. From an empirical point of view, data show that the loans originated between 1996 and early 1999 suffered significantly more defaults than the loans originated before 1996 or after the second half of 1999. The rise of the defaults between 1997 and 2003 was one of the 3 major credit crisis of the last 80 years. The default rate hit a maximum close to the peaks of 1933 and 1991, but what is very specific to this crisis is its duration. In the two first crisis, the default rate felt of 50% in the year following the peak. In the case of this crisis, 12 months after the peak reached in the early days of 2002, the default rate has decreased of only 2.5%. The other important aspect of default risk is the level of the recoveries. Most of the bond contracts provide bondholders with recoveries in case of default. The recovery rate is often estimated from the market price of the defaulted instrument 30 days after the date of default. Indeed, after a short period of time necessary to reach an equilibrium, the post-default market price is a widely accepted indicator of the amount that an investor holding the bond on the long term would recover. Moody’s empirical studies (Moody’s, 2004) confirm the negative correlation between default rates and annual changes in recovery rates. This clearly shows that the macroeconomic factor is also an important driver of recoveries; when the economy does not perform well, default rates increase whereas recoveries decrease. This double effect of the macroeconomic factors is a key element to take into account for portfolio studies as explained by (Altman et al., 2003). The following table presents recovery rates for 120 defaulted loans and bonds in 2003 (Moody’s, 2004). The table shows the mean value-weighted recovery rates according to seniority.
Defaulted instrument Bank loans Sr. Secured Sr. Unsecured Bonds Sr. Secured Sr. Unsecured Sr. Subordinated Subordinated Preferred stocks
Recovery 76% 80% 39,7% 54,1% 44,4% 29,2% 12,0% 1,1%
This table shows a high level of recovery for senior secured loans, coming from their secured nature, seniority in capital structure and strong protective covenants. The senior unsecured recovery comes from a single observation of 80% recovery. The average recovery rate on defaulted bonds is much lower compared to loans (around 40%). The data clearly show that the seniority and the collateral of the transactions are the main driver of the recovery rate. Ratings are a measure of the creditworthiness of an issuer, including both the likelihood of the default of the issuer and the severity of the loss faced by the debt holder. The obligor is evaluated thanks to some qualitative and quantitative information. Of course, the evaluation of an issuer is not the result of a mathematical model, but rather the result of the judgment and experience of the credit analyst. This is why there is room for several rating agencies. However, for banks, nothing can replace an internal rating system, because the bank has the commercial relationship with the client and should know the client better than anyone else. As a consequence, the bank’s credit analyst is the best placed for measuring creditworthiness of the bank’s clients. In practice, banks have several rating models, one based on the financial ratios of the obligor, another one based on answers to an economical form. Combined with the external ratings, the credit analyst has several scores on the obligor and has in hand all the elements to build the bank’s score onto the client. The rating agencies use an ordered scale of ratings in terms of a letter system ranging from AAA for the best credit quality down to D for defaulted firms. This is Standard & Poor’s nomenclature. Between the two extremes, there are intermediate ratings ; the whole list is AAA, AA, A, BBB, BB, B, CCC, D. Moody’s rating scale is slightly different in terms in rating letters and meaning. Actually, Fitch’s and S&P’s rating system correspond to default probabilities whereas Moody’s rating are rather associated to expected losses. The mapping of rating categories onto default probabilities is a calibration of the credit risk model. There are two ways of doing that, either thanks to historical series of defaults, or thanks to market data. However both methods
lead to significantly different results because the second one incorporates into the probabilities of default the risk aversion of the investors; the probabilities inferred from market prices are called risk-neutral probabilities. They are market implied probabilities and useful for pricing credit instruments, but are not relevant for risk purposes. From the rating agencies websites, it is possible to get the statistics of defaults for each rating category. The following figure gives the default probability as a function of the rating in a logarithmic scale.
100,00% 10,00% 1,00% 0,10% 0,01%
C
C C C
C C
C +
B-
C C
B
B+
BB
+
BB
B-
B
BB
BB
B+
BB
A-
BB
A
-
A+
AA
AA
AA
A AA +
0,00%
This curve is accurately (the R2 is larger than 98%) fitted by a regression line, meaning that the default probability grows exponentially with the rating, as shown by many empirical studies.
References Altman E. et al., “The link between default and recovery rates: implications for credit risk models and procyclicality”, forthcoming in Journal of Business (2003). First Union Securities Inc., “The default cycle and implications for CDO valuation” (2001), available on www.fusiresearch.com. Moody’s, “Default and recovery rates of corporate bond issuers” (2004), available on www.moodys.com. Standard & Poor’s, “Corporate defaults in 2003 recede from recent highs” (2004), available on www.standardandpoors.com
Further reading Bluhm, C., L. Overbeck and C. Wagner, An introduction to credit risk modeling: Chapman & Hall/CRC, 2002. Crouhy, M., D. Galai and R. Mark, “Prototype risk rating system”, Journal of Banking & Finance, 25, 41-95 (2001). Crouhy, M., D. Galai and R. Mark, Risk management: McGraw-Hill, 2001. Fons, J. et al., “The evolving meaning of Moody’s bond ratings, Moody’s Special Comment, May (2003). Garman, C., “High yield expanded outlook 2004”, Merrill Lynch High Yield Research, December (2003). Hu, Y. and W. Perraudin, “The dependence of recovery rates and defaults”, Working paper, Birkbeck College, Bank of England and CEPR, February (2002).
