Basket Cds Pres

Structural Models of Credit
Spectral Element Methods
Structural Models of Credit: A spectral Element

Approach.
Pierre Garreau
Financial Mathematics Seminar

Florida State University
November 17, 2011
10
Baa
Aaa
20Y TBill
9
2
01/01/05
01/01/06
01/01/07
01/01/08
01/01/09
01/01/10
01/01/11
Figure: 20-Year Aaa and Baa bond spreads.
01/01/12
Definition
A credit spread is the premium paid to insure the risk of default of the
underlying entity, i.e. it is the difference between the risk free rate, rt ,
and the yield on the corporate bond, yt .
Bt
e rt t
Dt
st
Le yt t
ln Dt /LBt
6
MF Global
0
08/13
08/27
09/10
09/24
10/08
10/22
Figure: MF Global stock price.
11/05
11/19
Asset
Vt
Liabilities
St
Dt
Table: Modigliani-Miller theorem
(1 R)
t0 = 0
tN = T
...
st st st
st
i
h
R
DL = E (1 R)1 <T e 0 rs ds
(1 R)
t0 = 0
tN = T
...
st st st
FL = E
st
"
N
X
i=1
st 1 >ti e
R ti
0
rs ds
DL = (1 R)
B(0, s)dP( s)
0
(1 R)
t0 = 0
tN = T
...
st st st
FL =
st
N
X
i=1
st P( > ti )B(0, ti )
DL = (1 R) 1 e
rT
P( > T ) r
P( > s)e
rs
ds
(1 R)
t0 = 0
tN = T
...
st st st
FL =
st
N
X
i=1
st P( > ti )e rti

RT
(1 R) 1 e rT P( > T ) r 0 P( > s)e rs ds
s=
PN
rti
i=1 t P( > ti )e
Default time = inf{t : min0st Vs D}

P( > t) = P(min0st Vs > D) = E 1min0st Vs >D
DT = RT .1{ T } + D.1{ >T }
16
14
12
10
4
0
0.05
0.1
0.15
0.2
0.25
0.3
Figure: Vt = V0 e t+Wt +
0.35
PNt
i=0
0.4
Yi
0.45
0.5
Default time = T .1{VT D} + 1{VT >D}

P( > T ) = P(VT > D)
DT = VT .1{VT D} + D.1{VT >D}
16
14
12
10
4
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
PNt
i=0
0.4
Yi
0.45
0.5
Default time = T .1{VT D} + 1{VT >D}

P( > T ) = P(VT > D)
DT = D (D VT )+ Dt = De r (T t) Pt (VT , D)
16
14
12
10
4
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
PNt
i=0
0.4
Yi
0.45
0.5
(1 R 1 )
(1 R k )
t0 = 0
tN = T
(1)
(2)
...
st st st
st
DL = E (1 R)1 (k) <T e
R (k)
0
(1 R 1 )
rs ds
(1 R k )
t0 = 0
tN = T
(1)
(2)
...
st st st
FL = E
st
"
N
X
i=1
st 1 (k) >ti e
R ti
0
rs ds
i
h
R min(1 ,2 )
rs ds
DL = E (1 R)1min(1 ,2 )<T e 0
(1 R 1 )
(1 R k )
t0 = 0
tN = T
(1)
(2)
...
st st st
FL = E
"
N
X
i=1
st
st 1min(1 ,2 )>ti e
R ti
0
rs ds
First to default time = inf{t : min0st Vsi D i , i = 1, 2}
P( > t) = E 1{min0st Vs1 >D 1 ,min0st Vs2 >D 2 }

DT = Vi (1) .1{ (1) T } + D i .1{ (1) >T }
14
13
12
11
10
9
8
7
6
5
4
3
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
Figure: Vti = e t+ Wt +
PNti
k=0
0.4
Yki
0.45
0.5
First to default time = T .1min(V 1 ,V 2 )D} + 1{min(V 1 ,V 2 )>D}

T
P( > T ) = P(min(VT1 , VT2 ) > D)

DT = min(VT1 , VT2 ).1{min(V 1 ,V 2 )D} + D.1{min(V 1 ,V 2 )>D}
T
14
13
12
11
10
5
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
PNti
k=0
0.4
Yki
0.45
0.5
First to default time = T .1min(V 1 ,V 2 )D} + 1{min(V 1 ,V 2 )>D}

T
P( > T ) = P(min(VT1 , VT2 ) > D)

DT = D (D min(VT1 , VT2 ))+ = De r (T t) Ptmin (VT1 , VT2 , D)
14
13
12
11
10
5
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
PNti
k=0
0.4
Yki
0.45
0.5
Summary
One period models, default only happens at T .
Default is triggered by crossing a level, recovery happens at maturity.
In 2 dimensions, we are assuming rescaling.
Pricing a 0-coupon bond / computing survival probabilities
One dimension
Two dimensions

E 1min0st Vs >D

E 1{VT >D}
Pt (VT , D)

E h1{min0st Vs1 >D 1 ,min
i 0st Vs2 >D 2 }
Black Scholes framework !!
E 1{min(VT1 ,VT2 )>D}

Ptmin (VT1 , VT2 , D)
Pricing problem
Theorem (Feyman-Kac)
i
h
F (t, x) = E e r (T t) f (STt,x )|Ft
(u + Au r ) F (u, x) = 0 t < u < T , x (0, )n

F (T , x) = f (x)
x (0, )n
Pricing problem
Theorem (Feyman-Kac)
h
i
F (t, x) = E e r (T t) f (STt,x )|Ft
(u + Au r ) F (u, x) = 0 t < u < T , x (0, )n

F (T , x) = f (x)
x (0, )n
Proof.
t
V
=
=
e rt F (t, St ) = F (t, St )
Z t
Z t
(u + Au r )Fdu
u d Su +
F (0, S0 ) +
0
One dimension
Problem (1d Black-Scholes PDE)
t = x ((x)x ) + u(x)x + (x)

(0, t) = l (t), (L, t) = r (t)
(x, 0) = 0 (x)
(x) = 2 x 2 /2,
u(x) = (r 2 )x,
L = Smax ,
0<x <L
t>0
0x L
(x) = r
0 (x) = (K x)+
r = 0
x01
x11
x21
x41 = x02
t dx
x |L0
Z
x42 = Sm
(x)x x dx
0
u(x)x dx +
0
(x)dx
0
x01
x11
K Z
X
k=1
x21
x41 = x02
kt k dx
Xk () = xk1 + k +1
2
xk
kx k |L0
xk1
K Z
X
k=1
K Z
X
k=1
xk
xk
(x)kx kx dx
xk1
u(x)kx k dx
xk1
x42 = Sm
K Z
X
k=1
xk
(x)k dx
xk1
x01
x11
x21
k = 1, . . . , K
x41 = x02
Xk () = xk1 + k +1
2
x42 = Sm
j {1, . . . , N 1}
k . k
k
k
k
j wn = h (), j iN + hu , j iN + h , j iN
2
k = 1, . . . , K 1
k+1 . k+1
k . k
k
k+1
N wN +
0 w0 = rhsN + rhs0
2
2
10 = l , KN = r
10
"./results_call.txt" using 3:4
"./results_call.txt" using 3:6
9
8
7
6
5
4
3
2
1
0
0
10
15
(a) Approximated vs Exact: Put
Figure: X = S0 = 10, r = 8%, = 30%, T = 1
20
-1
20
40
80
160
320
-2
-3
-4
-5
-6
-7
-8
10
20
30
40
50
60
70
80
90
100
110
120
(a) Log-Error in terms of total mesh points N, for several number of

time steps
Figure: X = S0 = 10, r = 8%, = 30%, T = 1
Two dimensions
Problem (2d Black-Scholes)
= ( ) + q + , (x, y ) D, 0 < t < T

(x, y , 0) = 0 (x, y ), (x, y ) D
(x, y , t) = b (x, y , t), (x, y ) D

=
1
2
12 s12
1 2
1 2
22 s22
q=
(r 12 21 1 2 )s1
(r 22 12 1 2 )s2
0 (x, y ) = (k min(x, y ))+
= r
"./results_rainbow_00.txt" using 1:2:6
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1 1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1
2
Figure: Rainbow Payoff K = 0.5, Smax
= Smax
=1
S1
0
x
1
Smax
x
S2
y
0
y
2
Smax
(K min(S1 , S2 ))+
K
K
(K y )+
(K x)+
Table: Boundary for a put option on the minimum
S2
K
X(,) = X x + Y y

X X
J=
Y Y
S1
k = 1, . . . , K , {(i, j)}N
i,j=0 ,
.k
k
k
k
k
k
k
k
ij i j Jij h F , ij iN = hq , ij iN + h , ij iN
i Dk ( k ), j Dk ( k ),
kK(Dk )
o
n.k
k
k
k
ij i j Jij h F , ij iN
kK(Dk )
hqk , kij iN
hk , kij iN
k K(D), i D( k ), j D( k ),
n+1,k
= b (xi , yj , tn+1 )
ij
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1 1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Figure:
K = S01 = S02 = 0.5, r = 2%, 1 = 20%, 2 = 30%, T = 0.5, Ne = 10, N = 16
Future research
Current
Convergence analysis of Spectral Element methods in 2-d.
Add jumps components in the PDE: integral term.
Add dependence structure on jump sizes / frequency.
Solve problem on min0sT Ss rather than ST .
Prospect
Multi-periods.
More that two assets: Monte Carlo.
Comparison to intensity models !
Merci !
PoMax
10
9
8
7
6
5
4
3
2
1
0
-1
CoMax
10
9
8
7
6
5
4
3
2
1
0
-1
14
8
10
12
14
16
18
20 20
18
16
14
12
10
12
10
10
8
6
2
0
4
2
18
20
16
14
12
10
0 0
10
12
14
16
t0
0.35
0.12
0.08
0.06
0.04
0.02
0.18
0.16
0.14
0.1
0.2
0
-1
-0.8
0.3
-0.8
-0.6
0.25
-0.6
0.2
-0.4
0.15
-0.2
0.1
0.05
0.2
9
8
7
6
5
4
3
2
1
0
-1
18
20
18
16
14
12
-0.2
0
10
12
14
16
0.2
0.12
-0.5
0.5
1.5
0.1
0.08
0.06
0.04
0.2
0.8
-1
20
0.16
0.8
1
-1.5
18
0.14
0.8
1.5
-0.2
0.6
0 0
-0.4
0.4
0.5
-0.6
0.6
-0.8
0.4
-0.5
0.12
0.08
0.06
0.04
0.02
0.16
0.14
0.1
-1
0.6
-1
t2
0.2
0.18
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
0.4
1
-1.5
10
t1
0.35
0.05
0.15
0.25
0.1
0.2
0.3
0
-1
-0.4
9
8
7
6
5
4
3
2
1
0
-1
20
4
CoMin
14
12
1
-1.5
0.02
-1
-0.5
0.5
1.5

Basket Cds Pres

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Basket Cds Pres

Uploaded by

Copyright:

Available Formats

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit: A spectral Element

Financial Mathematics Seminar

Structural Models of Credit

Spectral Element Methods

Figure: 20-Year Aaa and Baa bond spreads.

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Figure: MF Global stock price.

Structural Models of Credit

Spectral Element Methods

Table: Modigliani-Miller theorem

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

Default time = inf{t : min0st Vs D}

Structural Models of Credit

Spectral Element Methods

Default time = T .1{VT D} + 1{VT >D}

Structural Models of Credit

Spectral Element Methods

Default time = T .1{VT D} + 1{VT >D}

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

DL = E (1 R)1 (k) <T e

Structural Models of Credit

Spectral Element Methods

Structural Models of Credit

Spectral Element Methods

First to default time = inf{t : min0st Vsi D i , i = 1, 2}

P( > t) = E 1{min0st Vs1 >D 1 ,min0st Vs2 >D 2 }

Structural Models of Credit

Spectral Element Methods

First to default time = T .1min(V 1 ,V 2 )D} + 1{min(V 1 ,V 2 )>D}

P( > T ) = P(min(VT1 , VT2 ) > D)

Structural Models of Credit

Spectral Element Methods

First to default time = T .1min(V 1 ,V 2 )D} + 1{min(V 1 ,V 2 )>D}

P( > T ) = P(min(VT1 , VT2 ) > D)

Structural Models of Credit

Spectral Element Methods

Black Scholes framework !!

E 1{min(VT1 ,VT2 )>D}

Structural Models of Credit

Spectral Element Methods

(u + Au r ) F (u, x) = 0 t < u < T , x (0, )n

Structural Models of Credit

Spectral Element Methods

(u + Au r ) F (u, x) = 0 t < u < T , x (0, )n