Exercise Session 4. Solutions.

Fixed Income and Credit Risk

Question 1. Consider a stochastic process Yt = eXt , where Xt is an Arithmetic Brownian Motion

dXt = µ dt + σ dWt , (1)

and both µ and σ are real-valued constants.

1. Integrate (1) to determine the distribution of XT given X0 = 0. Deduce the distribution of YT .

To integrate the differential equation (1) means to evaluate
Z T Z T Z T
dXu = µdu + σdWu .
0 0 0

Using the rules of integration, you find

XT − X0 = µT + σ(WT − W0 ),

which under the assumptions that W0 = 0 and X0 = 0 simplifies to

XT = µT + σWT .

Since above WT ∼ N (0, T ) and the rest is deterministic,

XT ∼ N (µT, σ 2 T ).

2. Compute the expected value and the variance of YT .

  1 2 2

Hint: to facilitate computations, use the fact that E eαZ = eαm+ 2 α s for Z ∼ N m, s2 .
Apply the hint to Yt = eXt to find

1 2
E[YT ] = E eXT = eµT + 2 σ T ,
 
2
E[YT2 ] = E e2XT = e2µT +2σ T ,
 

2
 2

2
V[YT ] = E[YT2 ] − (E[YT ]) = e2µT +σ T
eσ T
−1 .

3. Use Ito’s lemma to write down the dynamic of Yt .

By Ito’s lemma,

1 00 1 00
df (t, Xt ) = ft0 dt + fX
0
dXt + fXX d hW it = ft0 dt + fX0
(µdt + σdWt ) + fXX σ 2 dt =
 2  2
0 0 1 2 00 0
= ft + µfX + σ fXX dt + (σfX ) dWt ,
2

1
 where d hW it = dt is a quadratic variation of a Brownian Motion. For f (t, Xt ) = eXt = Yt , we find:

σ2
 
dYt = µ+ Yt dt + σYt dWt ,
2

or, equivalently,
σ2
 
dYt
= µ+ dt + σdWt .
Yt 2
We find that an exponential of an Arithmetic Brownian Motion is a Geometric Brownian Motion.

4. Integrate the differential equation that determines the dynamic of Yt (your result in sub-question 3 above),
and deduce the distribution of YT again. You must get the same result as in sub-question 1 above.
To integrate the differential equation

σ2
 
dYt
= µ+ dt + σdWt ,
Yt 2

you need to consider first the Ito differential of Zt = ln Yt = g(t, Yt ). Applying Ito’s lemma yields

σ2
  
1 1
d (ln Yt ) = gt0 dt + gY0 dYt + gY00 Y d hY it = gt0 dt + gY0 µ+ Yt dt + σYt dWt + gY00 Y σ 2 Yt2 dt =
2 2 2
σ2
  
1 1 1 2 2
= 0 · dt + µ+ Yt dt + σYt dWt − σ Yt dt =
Yt 2 2 Yt2
= µdt + σdWt = dXt .

Question 2. The Merton (74) model defines the ‘distance-to-default’ DD as

2
 
VA

σA
ln D + r− 2 T
DD = √ ,
σA T

where VA is the value of firm’s assets that is governed by a Geometric Brownian Motion with constant volatility
σA , r is the risk-free rate, and D is the firm’s debt with maturity T . The model demonstrates that the probability
of default, under these assumptions, is P [VA 6 D] = Φ(−DD), where Φ(·) is a c.d.f. of a Standard Normal
distribution. Assume that r = 0.01, T = 1 year, and σA = 0.15.

1. Compute the probability of default for VA = 2 bln CHF and D = 1.6 bln CHF.
See an accompanying file merton.R for a numerical solution. The probability of default equals 6.95%.
D
2. Define the leverage ratio l = VA . What is the highest leverage ratio this firm can achieve keeping the
probability of default below 5%.
See an accompanying file merton.R for a numerical solution. The leverage ratio corresponding to a probability
of default of 5% is 0.78 (78%). For the assets of 2 bln CHF, it implies debt capacity of 1.56 bln CHF.

Question 3. Consider a log-normal random variable X ∼ ln N m, s2 (here m and s2 are the mean and the
variance of a corresponding Normal distribution).

1. Demonstrate that
m + s2 − ln k
 
2
m+ s2
E [X · 1X>k ] = e Φ ,
s

2
 where 1A is an indicator function that takes a value of 1 in A and 0 otherwise, and Φ(·) is a c.d.f. of a
Standard Normal distribution.


Recall first that X ∼ ln N m, s2 means that the distribution of X coincides with the distribution of em+s ,
where  ∼ N (0, 1). Then X > k is equivalent to m + s > ln k and

ln k − m
> ≡ ∗ .
s

Hence, Z +∞
E [X · 1X>k ] = E em+s · 1>∗ = em+su · 1u>∗ f (u)du,
 

−∞

where f (·) is the p.d.f. of the Standard Normal distribution. Now, split the integral into two, and then take
into account that the indicator function equals 0 unless u > ∗ :
Z +∞ Z ∗ Z +∞
m+su m+su
em+su · 1u>∗ f (u)du =




e ·1 u>∗ f (u)du = e ·1 u>∗ f (u)du +
−∞ −∞ ∗
Z +∞
= em+su f (u)du. (2)
∗

The next step is to write down f (u) explicitly and complete the square:

1 1 2 1 2 1 2 2 2
em+su f (u) = em+su √ e− 2 u = em √ e− 2 (u −2su) = em √ e− 2 (u −2su+s −s ) =
1 1

2π 2π 2π
2
m+ s2 1 − 1 (u−s)2
=e √ e 2 .
2π

Plug this result back into the integral (2):

Z +∞ Z +∞
m+ s2
2 1 1 2
e m+su
f (u)du = e √ e− 2 (u−s) du,
∗ ∗ 2π

and perform the change of variable such that:

x = u − s,
dx = du,
ln k − m − s2
ν ∗ = ∗ − s = .
s

Then,
Z +∞ Z +∞
m+ s2
2 1 1 2 s2 1 1 2 s2
e √ e− 2 (u−s) du = em+ 2 √ e− 2 x dx = em+ 2 (1 − Φ (ν ∗ )) =
∗ 2π ν∗ 2π
m + s2 − ln k
 
s2 2
m+ 2 ∗ m+ s2
=e Φ (−ν ) = e Φ ,
s

which completes the derivation.

2. Deduce the value of E [ST · 1ST >K ], where St is a Geometric Brownian Motion dSt = µSt dt + σSt dWt , with
both µ and σ being real-valued constants.
First, you need to determine the distribution of ST . For that, you need to consider the Ito differential of

3
Zt = ln St = g(t, St ). Applying Ito’s lemma yields

1 00 1 00 2 2
d (ln St ) = gt0 dt + gS0 dSt + gSS d hSit = gt0 dt + gS0 (µSt dt + σSt dWt ) + gSS σ St dt =
2 2
1 1 1 2 2
= 0 · dt + (µXt dt + σSt dWt ) − σ St dt =
St 2 St2
σ2
 
= µ− dt + σdWt . (3)
2

Integrate (3) from 0 to T to find:

σ2
 
ln ST = ln S0 + µ − T + σWT ⇐⇒
2
2
 
ln S0 + µ− σ2 T +σWT
ST = e ,

Which implies, that

σ2
   
ST ∼ ln N ln S0 + µ − T, σ 2 T .
2
Set

σ2
 
m = ln S0 + µ − T,
2
s2 = σ 2 T

and re-write the result of a previous sub-question accordingly to find:

 
σ2
 
 2
 2
ln S0 + µ− σ2 T + σ 2T
ln S0 + µ − 2 T + σ 2 T − ln K
E [ST · 1ST >K ] = e Φ √ =
σ T
 
σ2
 
S0


ln K + µ+ 2 T
µT
= S0 e Φ √ .
σ T