Chapter 9 Estimating Default Probabilities:
After completing this reading, you should be able to:
● Compare agencies’ ratings to internal credit rating systems.
● Describe linear discriminant analysis (LDA), define the Altman’s Z-score and its usage, and apply LDA to
classify a sample of firms by credit quality.
● Describe the relationship between borrower rating and probability of default.
● Describe a rating migration matrix and calculate the probability of default, cumulative probability of default,
and marginal probability of default.
● Define the hazard rate and use it to define probability functions for default time as well as to calculate
conditional and unconditional default probabilities.
● Describe recovery rates and their dependencies on default rates.
● Define a credit default swap (CDS) and explain its mechanics including the obligations of both the default
protection buyer and the default protection seller.
● Describe CDS spreads and explain how CDS spreads can be used to estimate hazard rates.
● Define and explain CDS-bond basis.
● Compare default probabilities calculated from historical data with those calculated from credit yield spreads.
● Describe the difference between real-world and risk-neutral default probabilities and determine which one to
use in the analysis of credit risk.
● Using the Merton model, calculate the value of a firm’s debt and equity, the volatility of firm value, and the
volatility of firm equity.
● Using the Merton model, calculate distance to default and default probability.
● Assess the quality of the default probabilities produced by the Merton model, the Moody’s KMV model, and
the Kamakura model.
Agencies’ Ratings vs. Internal Expert-based Rating Systems
Banks’ internal classification methods are somewhat different from agencies’ ratings assignment processes.
Nevertheless, sometimes their underlying processes are analogous; when banks adopt judgmental approaches to
credit quality assessment, the data considered and the analytical processes are similar.
An expert-based approach relying on judgment will require significant experience and repetitions in order for many
judgments to converge. In other words, judgment-based schemes need long-lasting experience and repetitions,
under a constant and consolidated method, to assure the convergence of judgments. As we would expect, therefore,
internal rating systems take time to develop. The failure to attain consistency under the expert-based approach can
be attributed to several factors, some of which are outlined below:
● Organizational patterns are intrinsically dynamic.
● There’s the issue of mergers and acquisitions that blend different credit portfolios, credit approval procedures,
internal credit underwriting powers, and so forth.
● Over time, company culture changes, and so do experts’ skills and analytical frameworks, particularly with
reference to qualitative information.
However, there is no proven inferiority or superiority of expert-based approaches versus formal ones, based on
quantitative analysis such as statistical models.
Earlier on, we looked at the qualities of a good rating system – objectivity and homogeneity, specificity, measurability,
and verifiability. We can compare agencies’ ratings and internal expert-based rating systems along similar lines:
● On objectivity and homogeneity, agencies’ ratings are 73% compliant, while internal expert-based rating
systems are 30% compliant.
● On specificity, agencies’ ratings are close to 100% compliant, while internal expert-based rating systems are
75% compliant.
● On measurability and verifiability, agencies’ ratings are 75% compliant, while internal expert-based rating
systems are 25% compliant.
Linear Discriminant Analysis and the Altman’s Z-score
Linear discriminant analysis
A scoring model is a model in which various variables are weighted in varying ways and result in a score. This score
subsequently forms the basis for a decision. In finance, scoring models combine quantitative and qualitative
empirical data to determine the appropriate parameters for predicting default. Linear discriminant analysis (LDA) is a
popular statistical method of developing scoring models.
The linear discriminant analysis classifies objects into one or more groups based on a set of descriptive features.
Models based on LDA are reduced-form models due to their dependency on exogenous variable selection, default
composition, and the default definition. The variables used in an LDA model are chosen based on their estimated
contribution (i.e., weight) to the likelihood of default. These variables are both qualitative and quantitative. Examples
are the skill and experience of management and the liquidity ratio, respectively. The contributions of each variable
are added to form an overall score, called Altman’s
Altman’s Z-score
Altman Z-score is essentially a bankruptcy prediction tool published by Edward I. Altman in 1968. Mr. Altman worked
with 5 ratios: net working capital to total assets ratio, earnings before interest and taxes to total assets ratio, retained
earnings to total assets ratio, market value of equity to total liabilities ratio, and finally, sales to assets ratio. Below is
the LDA model proposed by Altman:
where:
x1=Working capital/total assets.
x2=Retained earnings/total assets.
x3=EBIT/total assets.
x4=Equity market value/face value of term debt.
x5=Sales/total assets.
In this model, the higher the Z-score, the more likely it is that a firm will be classified in the group of solvent firms.
Altman worked with a Z-score range from -5.0 to +20.0, although higher scores may occur if a company has a high
equity value and/or low level of liabilities.
A Z-score cutoff, also known as the discriminant threshold, is used to categorize firms into two groups: solvent
firms and insolvent firms. Altman set the Z-score cutoff at Z = 2.675. Firms with a score below 2.675 are categorized
as insolvent while those with a score above 2.675 are categorized as solvent.
The Relationship between Borrower Rating and Probability of
Default
A borrower’s credit rating reflects their probability of default. The higher the rating, the more financially reliable a
borrower is considered to be. This implies that higher-rated issues have a lower probability of default. In fact, the
highest-rated issues almost never default even over a significant period of, say, 10 years. The lowest-rated issues,
on the other hand, often default early and are almost assured of default after a 10-year period.
Rating Migration Matrix
A rating migration matrix gives the probability of a firm ending up in a certain rating category at some point in the
future, given a specific starting point. The matrix, which is basically a table, uses historical data to show exactly how
bonds that begin, say, a 5-year period with an Aa rating, change their rating status from one year to the next. Most
matrices show one-year transition probabilities.
Transition matrices demonstrate that the higher the credit rating, the lower the probability of default.
The table below presents an example of a rating transition matrix according to S&P’s rating categories:
One-year transition matrix
Exam Tips:
● Each row corresponds to an initial rating.
● Each column corresponds to a rating at the end of 1 year. For example, a bond initially rated BB has an
8.84% chance of moving to a B rating by the end of the year.
● You will need to recall the rules of probability from mathematics to come up with n-year transition
probabilities, where n > 1
● The sum of the probabilities of all possible destinations, given an initial rating, is equal to 1 (100%).
● Credit ratings are their most stable over a one-year horizon. Stability decreases with longer horizons.
Probability of Default
This measures the likelihood of default over a single time period of length k. It is simply the fraction of the cohort that
survives to the end of the period:
Where:
PD k = Probability of default.
defaulted t+k , k = Number of issuer names (members of the cohort) that have defaulted
between time tand time t+k
Cumulative Probability of Default
The K-horizon cumulative default rate is defined as the probability of default from the time of cohort formation up to
and including time horizon K
If K is 5 years, for example, the cumulative probability of default in year 5 means the probability of an issuer name
defaulting in either year 1, 2, 3, 4 or 5 (i.e., the sum of defaults in years 1, 2, 3, 4, and 5).
The following table gives the cumulative default rates provided by Moody’s.
Cumulative Ave Default Rates (%) (1970-2009, Moody’s)
According to the table, an issuer name with an initial credit
rating of Ba has a probability of 1.166% of defaulting by the
end of the first year, 3.186% by the end of the second year,
and so on. We can interpret the other default rates in a
similar manner.
Marginal Probability of Default:
The marginal default rate is the probability that an issuer name that has survived in a cohort up to the
beginning of a particular interval k will default by the end of the time interval.
Annualized Default Rate
For discrete intervals, the annualized default rate can be computed as follows:
For continuous intervals,
Hazard Rates
The hazard rate (also called default intensity) is the probability of default for a certain time period conditional on no
earlier default. It is the parameter driving default. It is usually represented by the parameter λ. The probability of
default over the next small time interval, dt, is λdt.
Using t∗ to represent the time of default, the cumulative default time distribution F(t) gives the probability of default
over (0,t).
The survival distribution is as follows:
Note that the default time distribution and the survival distribution add up to 1 at each point in time.
As t grows very large, the survival probability converges at 0 while the default probability converges at 1. The
intuition is that the probability of default increases as we peer deeper into the future. Even the best-rated bond, say
AAA, will eventually default.
The default time density function is the
derivative of the default time distribution, w.r.t
t. Sometimes, it is called the marginal default
probability.
The default time density function is always
positive because default risk tends to
“accumulate” over time. However, the rate of
increase depends on λ. With a big value, default
risk will increase at a quick pace.
The survival probability, again, decreases over
time.
If we hold the hazard rate at a constant value λ, we will find that
the marginal default probability is positive but declining. The
implication is that although the probability of default increases
the further out we peer into the future, the rate at which this
probability accumulates declines.
Conditional Default Probability:
The conditional default probability gives the probability of
default over some horizon (t,t+τ) given that there has been no
default prior to time t.
For example, the conditional one-year probability of default, assuming survival during the first year, is equal to the
difference between the unconditional two-year PD and the unconditional one-year PD divided by the one-year
survival probability.
The conditional one-year PD is given by:
Example: Computing the Conditional PD
Compute the one-, two-, and three-year cumulative default probabilities and conditional default probabilities,
assuming that the hazard rate is 0.10
Solution :
Notice that the two conditional
probabilities are equal.
Risk-neutral Default Rates
Default probabilities can also be extracted from market prices. The resulting probabilities are risk-neutral. This
implies that the probabilities include compensation for both the loss given default and bearing the risk of default and
uncertainties that come along with it.
The spread over the risk-free rate on a bond that is defaultable with maturity T is denoted by zt, and the constant
risk-neutral hazard rate at time T is λ∗T.
If recovery is zero,
The equation above implies that the hazard rate is equal to the spread.
For small values of x, we can use the approximation so that:
After a bit of algebraic manipulation, it can be shown that the average default intensity over the life of the bond is
approximately,
where s is the spread of the bond’s yield over the risk-free rate, and R is the recovery rate.
Example: With a five-year bond that has a spread of 200 bps and a recovery rate of 40%, for example, the average
default intensity (hazard rate) = 0.02/0.6 = 0.0333.
Recovery Rates and their Dependencies on Default Rates
Recovery rates are a key metric in the assessment of credit risk, representing the proportion of principal and accrued
interest that can be recuperated in the event of a borrower’s default. It is often expressed as a percentage of face
value that creditors recover after a firm’s default.
Historical data show variations in average recovery rates among different bond classes. For example:
● First lien bonds exhibit the highest recovery rates because they have the first claim on assets.
● Senior unsecured bonds, which are not backed by collateral, typically recover less than secured bonds but
more than subordinated options.
● Subordinated bonds, including junior subordinated bonds, recover the least as they are lower in the
repayment hierarchy.
The Interplay Between Default and Recovery Rates
During financial distress, as more firms default, more assets enter the market, pushing down their sale prices.
Consequently, this affects recovery rates negatively:
● In economically strong years with fewer defaults, average recovery rates on bonds tend to be higher.
● Conversely, in years with high default rates, recovery rates often decline due to the oversupply of assets for
sale and poorer economic conditions.
The dependency of recovery rates on default rates has significant implications for lenders and investors. A high
default rate can not only mean more frequent losses but also lower recoveries on investment, leading to a
compounded negative effect on returns.
Studies have noted this negative relationship, pointing out that recovery rates on corporate bonds tend to drop in
years when the default rate is high, and vice versa. The correlation indicates that default and recovery rates are
inversely related, a factor which must be considered when estimating potential losses from credit risks.
Credit Default Swaps: Definition and Mechanics
Defining a Credit Default Swap (CDS)
A credit default swap (CDS) is a financial derivative instrument that works similarly to insurance on credit events. It is
primarily used to manage and trade credit risks associated with various financial instruments, typically bonds. They
grew rapidly in popularity until 2007 due to their ability to ‘insure’ against the default of a particular company—known
as the reference entity. Following a default event, CDS contracts facilitate a transfer of credit risk from one party to
another.
Mechanics of a CDS
● Protection buyer and Protection Seller: In a CDS transaction, there are two roles: the buyer of protection and
the seller of protection. The buyer of protection pays a periodic fee to the seller in exchange for a guarantee
of receiving a payoff if a predefined credit event (such as a default or failure to pay) occurs relating to the
reference entity’s debts. The buyer of the protection essentially obtains the right to sell the bonds issued by
the company for their face value upon the occurrence of a credit event. On the other side, when a credit
event occurs, the seller of protection has the obligation to buy these bonds for their face value. The total face
value associated with the CDS contract is referred to as the notional principal.
● Periodic Payments and Credit Events: The buyer of a CDS is obligated to make periodic payments, typically
done quarterly in arrears, until the maturity of the CDS contract or until a credit event happens.
● Physical Settlement vs. Auction Process: The settlement of a CDS can occur through physical delivery, where
the protection buyer delivers defaulted bonds to the protection seller in exchange for the face value.
Alternatively, an auction process can determine the value of bonds post-default, and the settlement is made
in cash based on this price.
● Buyer’s Payoff in Case of Credit Events: If there is no credit event, the buyer receives no payoff and has only
paid the periodic fees. However, if a credit event does occur, a payout, typically a substantial one, is received
by the buyer. For instance, during the bankruptcy of Lehman Brothers, the recovery rate was quite low,
meaning the payout for buyers of protection was about 92% of the notional principal, revealing the level of
credit risk the sellers of protection assumes.
Key Roles in a CDS Contract
● Default Protection Buyer
○ Pays regular premiums (CDS spreads).
○ Has the right to sell bonds at face value if a credit event occurs.
● Default Protection Seller
○ Collects periodic payments.
○ Is obligated to buy bonds at face value or make cash settlement upon a credit event.
Deriving a Hazard Rates from CDS Spread
Before looking at how we can go about deriving a hazard rate curve from CDS spread, let’s remind ourselves of a
few things about credit default swaps.
Generally, no principal or other cash flows change hands at the initiation of the contract.
However, when CDS trade points upfront, a percent of the principal is paid by the protection buyer. This impacts the
counterparty credit risk of the contract rather than its pricing. In addition, some collateral has to be provided at the
onset.
Each contract has a set of agreed future cash flows.
Quarterly spread payments (called the free leg) have to be made by the protection buyer until the maturity date of the
contract, unless and until the reference entity undergoes an event of default.
The protection seller makes a payment, called the contingent leg, only if there is a default. This payment leg is equal
to the loss given default.
The pricing of the CDS is set so that the expected net present value of the CDS contract is zero.
At the onset, the expected present value of the free leg is equal to that of the contingent leg.
The Derivation Process
Assume that we wish to find the default curve for a company. Let’s further assume that we have only a single CDS
spread, for a term of five years. As such, we will have a single hazard estimate. The protection buyer will pay the
spread in quarterly installments. These will be dates t = 0.25, 0.5,…1.5, 5. The installments will be a function of the
unknown hazard rate λ, which is linked to the probability of survival up to time t, πt, as follows:
We will need to work with the CDS valuation equation which equates the PV of the free leg to the PV of the
contingent leg.
The PV of the free leg is given by:
While the PV of the contingent leg is given by:
where:
pt is the price of a risk-free zer0-coupon bond maturing at time t.
R is the recovery rate = 1 – LGD.
πt is the probability of survival up to time t.
τ= Term of the CDS (5 in this example).
Sτ = Spread of the CDS.
λ = Hazard rate (our unknown).
Provided all the variables are known, we can substitute them in the equation and get the value of λ
Example: Single Hazard Rate
We can work out a value for λas follows:
Solving this gives λ=0.0741688
CDS-Bond Basis Definition and Explanation
Definition of CDS-Bond Basis
The CDS-bond basis is defined as the difference between the CDS spread and the bond yield spread for a specific
company.
Under ideal conditions, the basis should be near zero. This is based on arbitrage arguments, which suggest that the
cost of credit protection acquired through a CDS should, in theory, align with the extra yield demanded by the market
for holding a risky bond over a risk-free bond.
Factors Causing Deviation
In practice, the CDS-bond basis can deviate from zero due to various reasons:
● Bond Prices Relative to Par: Bond prices that deviate significantly from par value can result in a positive or
negative basis. Prices above par tend to move the basis in a negative direction, while prices below par tend
to give rise to a positive basis.
● Counterparty Default Risk in CDS: The existence of counterparty default risk in a CDS contract generally
pushes the basis towards negative territory.
● Cheapest-to-Deliver Option in CDS: This element gives the protection seller an option to determine the
most economical bond to deliver in the event of a default, which can nudge the basis positively.
● Settlement of Payments Not Including Accrued Interest: If the CDS payout does not cover accrued
interest on delivered bonds, this can depress the CDS-bond basis.
● The Restructuring Clause: In certain scenarios, the CDS contract may stipulate different conditions for
payouts, unrelated to a traditional default, which can direct the basis positively.
● Discrepancy in Benchmarks for Yield Spreads: The calculated bond yield spread might be relative to a
“risk-free” rate that differs from the one commonly used by the market, affecting the basis’ magnitude and
direction.
Historical Performance
The historical performance of the CDS-bond basis has varied:
● Pre-2007, the basis tended to be positive, with estimates hovering around 16 basis points.
● During the Global Financial Crisis, the basis became profoundly negative. Liquidity shortages and other
challenges impeded financial institutions’ ability to arbitrage between bonds and CDSs effectively.
● Post-crisis, the basis has mostly been observed to be small and negative.
Comparison of Default Probabilities: Historical Data vs. Credit
Yield Spreads
Default Probabilities from Historical Data
Default probabilities calculated from historical data represent “real-world” or physical probabilities of default. These
probabilities are inferred from past occurrences of defaults within certain classifications or ratings assigned to
companies. Credit rating agencies like Standard & Poor’s publish average cumulative default probabilities based on
extensive historical data. For example, they may calculate the seven-year average cumulative default probability by
analyzing the historical performance of companies within each credit rating category.
Default Probabilities from Credit Spreads
On the other hand, default probabilities calculated from credit spreads are referred to as “risk-neutral” probabilities.
Credit spreads are the risk premiums over the “risk-free” rate that investors demand for taking on the credit risk of a
corporate bond, reflecting the market’s perception of credit risk. These spreads can be used to calculate an implied
hazard rate, which can then be translated into a risk-neutral default probability.
Comparison and Analysis
Hazard rates implied by credit spreads tend to be higher than those calculated from historical default rates. This
suggests that the market demands a premium for bearing credit risk, which goes beyond an actuarial estimate of the
cost of defaults.
Moreover, this risk premium or excess return increases as the credit quality declines, indicating that investors are
typically compensated more generously for assuming higher credit risks. The discrepancy between the two estimates
becomes more pronounced during periods of financial stress, such as the Global Financial Crisis when credit
spreads soared significantly.
The choice between using real-world and risk-neutral default probabilities depends on the application. For valuation
purposes, risk-neutral default probabilities, derived from credit spreads, are typically used, consistent with the
principle of using market-based valuations. For scenario analysis, where the goal is to assess different potential
future states, it is more appropriate to use real-world probabilities that consider historical default rates.
Real-World vs. Risk-Neutral Default Probabilities
Real-World Default Probabilities
Real-world, or physical, default probabilities are derived from historical data. These are probabilities published by
rating agencies, based on historical default rates of entities within various credit rating categories. They provide an
empirical measure of the frequency of default observed in the past and are seen as an estimate of the actual
likelihood that a company will default on its debt obligations in the future.
Risk-Neutral Default Probabilities
Risk-neutral default probabilities, in contrast to real-world probabilities, are derived from financial markets,
particularly from credit spreads. These are higher than real-world default probabilities because they include not only
the actual historical likelihood of default but also market participants’ risk aversions and the premium that investors
demand for bearing credit risk. Essentially, these probabilities are a measure of the market’s current expectations
regarding the risk of default.
Comparison and Usage
Risk-neutral default probabilities are greater than real-world default probabilities because they reflect the market’s
demand for risk premiums on top of the historical likelihood of default. The difference accounts for the extra
compensation investors need to hold risky debt over risk-free assets.
The distinction between the two comes down to the expected asset growth rate:
● In the Real World: The rate at which a company’s assets are expected to grow is usually higher than the
risk-free rate, reflecting the extra return investors seek for bearing risk over time.
● In the Risk-Neutral World: The expected asset growth rate equals the risk-free rate, assuming individuals
do not require additional returns for bearing risk.
Which to Use?
● For Valuation: Risk managers and financial analysts should use risk-neutral probabilities when valuing
financial instruments, such as bonds or derivatives. This conforms to the underlying principle of risk-neutral
valuation, where all risk premiums are reflected in discounted future cash flows.
● For Scenario Analysis: Real-world probabilities are more appropriate for assessing various potential future
outcomes. They incorporate actual historical experience and provide a basis for estimating frequencies of
future default under different economic scenarios.
Using the Merton Model for Firm Valuation
Calculating the Value of Equity
The Merton model is used to assess a company’s credit risk by modeling the company’s equity as a call option on its
assets. It is built upon the Black-Scholes pricing model and seeks to establish a link between default and a firm’s
capital structure.
In its simplest form, the Merton model makes a set of assumptions:
● The firm pays no dividends.
● There’s only one liability claim.
● Markets are perfect, meaning that there are no taxes, transaction costs, or any other cost in the process of
taking up a contract.
These assumptions have certain implications. If the markets are perfect, for example, then the value of the firm
equals the value of debt plus the value of equity.
In line with these assumptions, let’s assume that a firm has a single outstanding zero-coupon debt with a face value
(principal amount) F, payable at time T. Now, there are two possible scenarios. First, the value of the firm at time T,
VT could be large enough to pay the principal amount, in which case, shareholders have a claim over the balance,
i.e., VT-F. Secondly, the value of the firm at time T could be insufficient such that the firm is unable to settle the
principal amount, in which case, equity holders receive nothing.
Looking closely, the two possible scenarios and their payoffs pretty much resemble a call option, with the underlying
instrument as firm value and the principal amount as the exercise price. The value of equity at time T can therefore
be represented as:
A simplified example will help further drive the concept home.
Example: Computing the Value of Equity
Assume that a firm has issued zero-coupon debt that requires it to pay $160m to debt holders at maturity. Further,
assume that the firm has no other creditors. If the total value of the firm at maturity is $180m, what’s the firm’s value
of equity? What is the value of equity given a firm value of $140m at maturity?
Solution
The following figure illustrates the fact that the payoff for equity
is equal to the payoff for a long call option.
Calculating the Value of Debt
How about the payoff for debt? If we assume that a debt is risk-free and payment of the principal amount F is
guaranteed, the payoff would be F, regardless of firm value at maturity. In the presence of risk, however, the debt
holder receives an amount lower than the anticipated payment if the firm value, VT is less than F. In other words, we
are saying that if VT<F at maturity, then the amount received by the debt holder will be reduced by F−VT.
The payoff for debt is pretty much like that of a combination of a long position in a T-bill with a face value of F and a
short position on the firm value with an exercise price of F. In other words, holders of risky debt effectively buy
risk-free debt and write a put option on the value of the firm with an exercise price equal to the face value of the debt.
We can therefore represent the value of debt, DT, as follows:
