Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Advanced Debt Instruments

4. Credit Analysis Model

Jung-Hyun Ahn

2024
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Roadmap

1 Modeling Credit Risk and the Credit Valuation Adjustment

2 Value, spread of risky bond in an Arbitrage-Free Framework

3 Interpreting Credit Spread Changes

4 Valuing risky FRN

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Modeling Credit Risk

Credit spread, or G-spread, is a common measure of credit risk.

This section, we learn a framework providing credit spread based on the factors of
credit risk.
under more rigorous definition of credit risk.
ignore taxation and liquidity risk

Distinction between credit risk and default risk

Default risk: the likelihood of an event of default.
Credit risk: address both the default probability and how much it is expected to be
lost if default occurs. The term broader than the default risk

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Factors in Credit Risk

Expected exposure, EE (or Exposure at Default, EAD)

Projected amount of money at risk if an event of default occurs, before factoring

in possible recovery.

Recovery Rate (RR), Loss Given Default (LGD)

Recovery Rate (RR): the percentage of the loss recovered

Loss severity: the percentage of the loss = 1 − RR

Loss Given Default (LGD): The amount of loss if a default occurs

LGD = EE × Loss severity = EE × (1 − RR)

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.1. Expected loss, Loss given default

Consider a one-year, 4% annual payment corporate bond priced at par. Given 40%
of recovery rate, compute expected exposure, loss severity, and loss given default.

Sol.

EE = 104
per 100 of par value.

Loss Severity = 1 − RR = 1 − 40% = 60%
.

LGD = EE× Loss severity = 104 × 0.6 = 62.4
per 100 of par value.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Probability of Default (PD)

The probability that a bond issuer will not meet its contractual obligation on sched-
ule in a given period.
Conditional Probability of Default (or Hazard Rate, HRt ): Probability of
default in a given period t assuming no prior default.

(Unconditional) Probability of Default, P Dt : Probability of default in a given

period t (from the perspective of time 0).

Probability of survival, P St : Probability that no default occurs by the end of

period t (from the perspective of time 0)

P Dt = P St−1 × HRt

P D1 = HR1 : Initial PD or PD in first period is same as hazard rate since

there is definitely no prior default.

P St = P St−1 − P Dt
= (1 − HR)t (if hazard rate is constant over all the periods.)
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Measuring Credit Risk

Credit Valuation Adjustment (CVA)

Expected loss, EL = P D × LGD

P
Credit Valuation Adjustment, CV A = P V of expected loss (EL)

Fair value of a risky bond

= Value of otherwise equivalent risk-free bond (VND) - CVA

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.2. Credit Valuation Adjustment

A five-year, $100 par, zero-coupon corporate bond has a hazard rate of 1.25% per
year. Its recovery rate is 40% and the benchmark rate curve is flat at 3%.

1 Calculate the expected exposure, the probability of survival, the probability of

default, the loss given default, in each year, and the credit valuation
adjustment. Assume that default occurs only at year-end on year 1, 2, 3, 4,
and 5.

2 Calculate the credit spread.

3 Calculate the annual internal rate of return (IRR) for the investment under
different default scenarios. Determine the IRR assuming the issuer defaults at
the end of first year (and second, third, fourth and fifth year), as well as the
IRR if the issuer does not default until the maturity of the bond.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol.

1 CVA (See Excel solution file)

(2) EEt : PV of par value at the end of year t discounted by risk-free rate.
(3) Recovery amount: = EEt × RR
(4) LGDt = EEt − Recoveryt = EEt × (1 − RR)
(5) P Dt = P St−1 × HR
(6) P St = P St−1 − P Dt (or (1 − HR)t in (6bis))
(7) ELt = P Dt × LGDt
(8) Discount factor for year t: DFt = 1/(1 + st )t where st is t-year risk-free spot rate.
(9) Present Value of Expected loss at t: P V (EL)t = ELt × DFt
P5 
CV A = t=1 P V (EL)t = 3.1549
per 100 of par value
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

2 Credit spread
Value of the otherwise equivalenet risk-free bond (VND): PV of the bond using
risk-free rate as the discount rate.
100
V ND = = 86.2609
(1.03)5
Fair Value of risky bond:
= V N D − CV A = 86.2609 − 3.1549 = 83.1060
YTM of the bond:  (1/5)
100 100
83.106 = ⇒ Y T M = − 1 = 3.77%
(1 + Y T M )5 83.106
Credit spread = YTM of risky bond - YTM of risk-free bond

= 3.77 − 3.00 = 0.77% or 77 bps


Note: (In CFA L1) The approximation of credit spread commonly used in
practice is computed by P D1 × (1 − RR) = 1.25% × (1 − 0.4) = 0.75%
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

3 IRR: Compute IRR given the price $83.1060 and Cash-flow upon whether and
and default occurs.
Annual IRR if default occurs at the end of year 1:
35.5395 
83.1060 = ⇒ IRR = −0.5724, or −57.24%
1 + IRR 

Annual IRR if default occurs at the end of year 2:
36.6057 /2
 1
0 36.6057
83.1060 = + 2
⇒ IRR = − 1 = −0.3363
1 + IRR (1 + IRR) 83.1060

or −33.63%



In similar way, the annual IRRs if default
 occurs at the end of year 3, 4 and 5
are −23.16% , −17.32% , and −13.61% , respectively. (See Excel




solution file.)
Annual IRR, if no default occurs till the maturity of the bond:
0 0 100
83.1060 = + + ··· +
1 + IRR (1 + IRR)2 (1 + IRR)5

100
1/5 
⇒ IRR = − 1 = 0.0377 or 3.77%

83.1060
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Task in practice

The above example is a very simple example of credit risk model. In practice,

hazard rate and recovery rate may vary period by period.

risk-free yield curve may not be flat.

we use the risk neutral probabilities of default rather than actual (or
historical) default probabilities.

we may have to analyze coupon bond in which case you consider coupon at
each period as well as present value of the cash flow occuring in the
subsequent periods.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Risk-neutral probability of default

The probability of default implied in the current market price assuming that investors
are risk-neutral.

Consider a one-year, $100 par value, zero-coouon bond trading at $95. The bench-
mark 1-year rate is 3% and the recovery rate is assumed to be 60%. In one year, if
the bond dose not default, it pays promised par value $100, while cash flow will be
only $60 in case of default ($100 × 60%).

The risk-neutral probability of default, p can be estimated by

p × 60 + (1 − p) × 100
95 = ⇒ p = 5.38%
1.03

Basically, risk-neutral default probability is higher than actual (historical) default probability because the
former does not include
the risk premium associated to the uncertainty related to default loss (average investors are
risk-averse);
liquidity risk and tax consideration.
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Relative Credit Risk Analysis

Ex.4.3. Relative Credit Risk Analysis

A fixed-income analyst is considering the credit risk over the next year for three cor-
porate bonds currently held in her bond portfolio. Her assessment for the exposure,
probability of default, and recovery is summarized in this table:

Although all three bonds have very similar yields to maturity, the differences in the
exposures arise because of differences in their coupon rates.

Based on these assumptions, how would she rank the three bonds, from highest to
lowest, in terms of credit risk over the next year?

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol.

Analysis should be based on the EL next year computed by LGD × P D. LGD

can be computed by EE− Recovery amount.


B>C>A

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Valuing risky bonds in an Arbitrage-Free Framework

Input:
Interest rates elements: Par curve of benchmark bonds, interest rate volatility;
Credit risk factors: recovery rate, hazard rate (conditional probability of default,
initial probability of default, annual probability of default)

Process:
1 Compute spot, forward rate rates and discount factor for each year.
2 Build risk-free interest rate tree (if volatility >0)
3 Compute the value of the bond at each node assuming no default and then VND
(value at t = 0)
4 Compute EE, LGD, PD, EL and CVA for each year(present value of EL)
P
5 CVA = CV At
6 Fair value of risky bond = VND - CVA
7 Credit spread = YTM of risky bond -YTM of risk-free (benchmark) bond

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.4. Valuing risky bonds: Case with non-flat yield curve and volatility

Consider a five-year, 3.50% annual payment corporate bond. A fixed-income analyst assigns an
annual default probability of 1.25% and a recovery rate of 40% to this bond and assumes 10%
volatility in benchmark interest rates. The following tables presents spot rates, discount factors,
and forward rates extracted from the par curve for annual payment benchmark government bond
and interest rate tree built upon this information.

Estimate the fair value of this bond and the credit spread.
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol.(1)

1 Value of the 3.5% annual payment corporate bond assuming no default:

(See Excel solution file for computation in detail.)


VND = 103.5450

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol. (2)

2 Expected exposure (EE) at each time t:

P
Pr. of being at each node at t × Value at each node at t
Probability of being at each node

For example, at t = 2, PrU = 0.5 × 0.5 = 0.25,

PrM = 0.5 × 0.5 + 0.5 × 0.5 = 0.5, PrL = 0.5 × 0.5 = 0.25
Then, for example EE at t = 3:

94.6788 × 0.125 + 96.7465 × 0.375 + 98.4909 × 0.375 + 99.9551 × 0.1250 =
101.0433

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol. (3)

3 Credit Valuation Adjustment, CVA:

(See Excel solution file for the computation in detail)

LGDt = EEt × (1 − RR)

P Dt = P St−1 × HR
P St = P St−1 − P Dt (or (1 − HR)t )
ELt = P Dt × LGDt
Discount factor for year t: DFt = 1/(1 + st )t where st is t-year risk-free spot
rate.
CVA at t: P V (EL)t = ELt × DFt

CV A = 5t=1 CV At = 3.5394
per 100 of par value
P
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol. (4)

4 Fair Value of the bond = V N D − CV A = 103.5450 − 3.5394

 
= 100.0056 per 100 of par value .
 
5 YTM = RATE(5,3.5,-100.0056,100)=3.4988%
6 Credit spread = 3.4988 - 2.75 (from YTM of 5-yr Bencbmark bond)
 
= 0.7488% (74.88 bps)
 

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.5. Using Credit Analysis in Decision Making

Lori Boller is a fixed-income money manager specializing in taking long positions

on high-yield corporate bonds that she deems to be undervalued. In particular,
she looks for bonds for which the credit spread over government securities appears
to indicate too high a probability of default or too low a recovery rate if default
were to occur. Currently, she is looking at a three year, 4.00% annual payment
bond that is priced at 104 (per 100 of par value). In her opinion, this bond should
be priced to reflect an annual default probability of 2.25% given a recovery rate of
40%. Ms. Boller is comfortable with an assumption of 10% volatility in government
bond yields over the next few years. Should she consider buying this bond for her
portfolio? Use the government par curve and the binomial interest rate tree in
Exercise 4.4.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Interpreting a change in a credit spread

Credit spreads include compensation for

credit risk
liquidity risk, taxation disadvantage relative to the benchmark

In our analysis, we focus only on the credit risk component (i.e., CVA) as it is the
most important and most-commonly-used in practice.

Credit speads change as investor perceptions about the credit risk changes
Credit spreads increase as the probability of default increases, and as the recovery
rate decreases.

These perceptions depend on expectations about the state of the economy.

Expectations of impending recessions lead to expectations of higher defaults and
lower recovery rates, therefore lead to higher credit spreads.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.6. Impact of credit risk on spread

Consider the same bond and interest rates information in Exercise 4.4. The analyst
expects that there will be a severe recession in coming years and that the annual
probability of default will be doubled to 2.5% for two coming years, then if will
return to 1.25% for the remaining period if this bond survives. Estimate the credit
spread based on this new scenario.
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.7. Impact of credit risk on spread (2)

Joan Da Sliva, an analyst, is estimating the fair price of a 3-year subordinated bond
that XYZ Corporation is considering issuing. It pays 3% coupon rate annually. Da
Silva could not find any bond with the same seniority recently issued by XYZ or
by other entities with similar risk. However, Da Silva found a senior bond with the
same maturity and coupon rates issued recently by XYZ, which is traded at 102.00
per 100 of par value. Currently, the recovery rate of the bond issued by the firms
with the same credit rating as XYZ is estimated as 70% for senior bonds and 40%
for subordinated bonds, respectively.
The benchmark yield curve is flat at 1.75%, and the volatility is assumed to be 0.
Estimate the price and the credit spread of the subordinated bond.

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Ex.4.8. Valuing Risky Floating Rate Note (FRN, Floater)

An analyst considers a five-year floater that pays annually the one-year benchmark
rate plus 0.50%. He expects that for the first three years, the annual default
probability is 0.50% and the recovery rate is 20%. However, it is expected that the
credit risk of the issuer then worsens: For the final two years, the annual probability
of default goes up to 0.75% and the recovery rate goes down to 10%. Compute
the fair value of this bond and its discount margin. Use the government par curve
and the binomial interest rate tree in Exercises 4.4.

Notes: FRN

Coupon rate = Benchmark rate + spread (“quoted margin”)

Coupon payment is “in arrears,” meaning that rate is set at the beginning of
the period and paid at the end of the period.
Discount margin (DM): constant spread added to each discount rate, which
make the present value equal to the price. (equivalent to the credit spread in
fixed-rate bonds)

Advanced Debt Instruments, J. Ahn

Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol. (1)

(See Excel solution file for the computation in detail.)

 
VND = 102.3633 per 100 of par value
 
Advanced Debt Instruments, J. Ahn
Modeling Credit Risk & CVA Valuing risky bond Credit Spread Valuing risky FRN

Sol. (2)

(See Excel solution file for the computation in detail.)

D − CV A
Fair value of FRN = V N 
= 102.3633 − 2.4586 = 99.9047 per 100 of par value
 
Discount margin: Find the value, when added in the discount rate at each
node, making the price equal to the estimated fair value 
(99.9047) of
 the
FRN using goal seek function of Excel. −→ 0.5205% or 52.05 bps
 

Advanced Debt Instruments, J. Ahn