Professional Documents
Culture Documents
Nicolas PRIVAULT
Department of Mathematics
City University of Hong Kong
Tat Chee Avenue
Kowloon Tong, Hong Kong
Xiao WEI
School of Insurance
Central University of Finance and Economics
100081, Beijing, P.R. China
Abstract
This paper reviews the BGM model for the parameterization of LIBOR
forward interest rate curves, and presents a C++ implementation in PREMIA
of the calibration algorithm of [7] using the market prices of caps and swaptions
in this model.
Keywords: LIBOR market model, BGM model, interest rates, caps, swaps, calibration.
Introduction
This work was carried out while both authors were at INRIA Rocquencourt, Projet MathFi,
BP 105, 78153 Le Chesnay Cedex, France.
t R+ ,
Z
P (t, T ) := E exp
t
rs ds rt .
Figure 2.1 presents a random simulation of t 7 P (t, T ) in the Vasicek model, combined with the graph of the corresponding deterministic interest rate compounding.
0.9
0.8
P(t,T)
0.7
0.6
0.5
0.4
0.3
0
10
t
15
20
log P (t, T )
,
T
f(t,T)
2
0
10
T-t
15
20
0 t T < S.
The forward rate agreement at time t gives its holder the right to an interest rate
F (t, T, S) on the time period [T, S]. Next (Figure 2.3) is a simulation of the simply
compounded spot rates L(t, T ) defined as L(t, T ) = F (t, t, T ), and computed from
the sample graphs of Figure 2.1:
8.5
7.5
L(t,T)
6.5
5.5
4.5
0
10
t
15
20
5.5
F(0,T,T+delta)
4.5
3.5
2.5
2
0
10
T
15
20
TimeSerieNb
AsOfDate
2D
1W
1M
2M
3M
1Y
2Y
3Y
4Y
5Y
6Y
7Y
8Y
9Y
10Y
11Y
12Y
13Y
14Y
15Y
20Y
25Y
30Y
5
Forward interest rate
4.5
percentage
3.5
2.5
2
0
10
15
years
20
25
30
505
7mai03
2,55
2,53
2,56
2,52
2,48
2,34
2,49
2,79
3,07
3,31
3,52
3,71
3,88
4,02
4,14
4,23
4,33
4,4
4,47
4,54
4,74
4,83
4,86
P (t, Ti ) P (t, Tj )
,
P (t, Ti , Tj )
t [0, Ti ],
1 i < j n,
is defined from
0 = P (t, Tj ) P (t, Ti ) + S(t, Ti , Tj )
j1
X
t [0, Ti ],
k P (t, Tk+1 ),
1 i < j n,
k=i
where
P (t, Ti , Tj ) =
j1
X
t [0, Ti ],
k P (t, Tk+1 ),
1 i < j n,
k=i
t [0, Ti ],
1 i < j n.
When j = i+1, the swap rate S(t, Ti , Ti+1 ) coincides with the forward rate F (t, Ti , Ti+1 ),
and we recover the discount factor P (t, Ti ) from S(t, Ti , Ti+1 ) = F (t, Ti , Ti+1 ) using
the relations
P (t, Ti+1 ) =
P (t, Ti )
,
1 + i S(t, Ti , Ti+1 )
t [0, Ti ],
0 i n 1.
From now on we assume that the dynamics of P (t, Ti ) under the risk neutral probability measure is given by
dP (t, Ti )
= rt dt + i (t) dWt ,
P (t, Ti )
i = 1, . . . , n,
where (rt )tR+ and (i (t))tR+ , i = 1, . . . , n, are adapted with respect to the filtration
(Ft )tR+ generated by the Rd -valued Brownian motion (Wt )tR+ , under the assumption
(absence of arbitrage condition) that
t 7 e
Rt
0
rs ds
P (t, Ti ),
t [0, Ti ],
i = 1, . . . , n,
is an Ft -martingale under P.
Definition 3.1. Let the probability measure Pi be defined as
R Ti
dPi
1
=
e 0 rs ds ,
dP
P (0, Ti )
i = 1, . . . , n.
t [0, Ti ].
dPi|Ft
e t rs ds
=
,
dP|Ft
P (t, Ti )
t [0, Ti ].
(3.1)
t [0, Ti ],
for all integrable random variables F . The next proposition will be useful to determine
the dynamics of interest rate processes under Pi .
6
0 t Ti ,
(3.2)
dPi
P (t, Ti ) R t rs ds
i (t) = E
,
e 0
Ft =
dP
P (0, Ti )
t [0, Ti ],
0 t Ti Tj ,
(3.3)
P (t, Tj )
,
P (t, Ti )
0 t Ti Tj ,
is an Ft -martingale under Pi , 1 i, j n.
Proof. For all bounded and Ft -measurable random variables F we have
dPj
dPj
Ei F
= E F
dPi
dP
h
i
R Tj
1
=
E F e 0 r d
P (0, Tj )
h
i
Rt
1
=
E F e 0 r d P (t, Tj )
P (0, Tj )
RT
1
0 i r d P (t, Tj )
=
E Fe
P (0, Tj )
P (t, Ti )
P (0, Ti )
P (t, Tj )
=
Ei F
,
P (0, Tj )
P (t, Ti )
which shows (3.3).
7
1
P (t, Tj ) Y
=
,
P (t, Ti )
1
+
k F (t, Tk , Tk+1 )
k=i
t [0, Ti ],
and if 1 j i n,
i1
P (t, Tj ) Y
=
(1 + k F (t, Tk , Tk+1 )),
P (t, Ti ) k=j
t [0, Tj ].
The aim of this section is to describe the Brace-Gatarek-Musiela (BGM) model, and
in particular the BGM system of stochastic differential equations
n1
X
dF (t, Ti , Ti+1 )
j F (t, Tj , Tj+1 )
=
i (t) j (t)dt + i (t) dWtn ,
F (t, Ti , Ti+1 )
1
+
F
(t,
T
,
T
)
j
j
j+1
j=i+1
(4.1)
P (t, Ti )
P (t, Ti+1 )
1 P (t, Ti )
(i (t) i+1 (t)) (dWt i+1 (t)dt)
i P (t, Ti+1 )
1
(1 + i F (t, Ti , Ti+1 )) (i (t) i+1 (t)) (dWt i+1 (t)dt)
=
i
1
=
(1 + i F (t, Ti , Ti+1 )) (i (t) i+1 (t)) dWti+1 .
i
i = 0, . . . , n 1, (4.2)
we get
dF (t, Ti , Ti+1 )
= i (t) dWti+1
F (t, Ti , Ti+1 )
= i (t) dWtk + (k (t) i+1 (t))dt
= i (t)
k1
X
dWtk
!
(j (t) j+1 (t))dt
j=i+1
!
k1
X
j j (t)F (t, Tj , Tj+1 )
dWtk
dt ,
1 + j F (t, Tj , Tj+1 )
j=i+1
= i (t)
i (t)
dWtk +
i
X
!
(j (t) j+1 (t))dt
j=k
dWtk
i
X
j j (t)F (t, Tj , Tj+1 )
j=k
1 + j F (t, Tj , Tj+1 )
!
dt .
!
n1
X
(t)F
(t,
T
,
T
)
j j
j
j+1
dWtn
dt ,
1
+
F
(t,
T
,
T
)
j
j
j+1
j=i+1
dF (t, Ti , Ti+1 )
= i (t) dWti+1
F (t, Ti , Ti+1 )
9
for some deterministic Rd -valued function i (t), i = 1, . . . , n1. From Proposition 3.3
we have
i (t) := Ei+1
P (0, Ti+1 )
dPi
P (0, Ti+1 ) P (t, Ti )
=
(1 + i F (t, Ti , Ti+1 )),
F t =
dPi+1
P (t, Ti+1 ) P (0, Ti )
P (0, Ti )
P (0, Ti+1 )
F (t, Ti , Ti+1 )
F (t, Ti , Ti+1 )i (t)dWti+1 = i i (t)
(t)dWti+1 ,
P (0, Ti )
1 + i F (t, Ti , Ti+1 ) i
which eventually recovers (4.1), and also implies (4.2) in view of (3.2).
i E e
R Ti+1
t
rs ds
i
(L(Ti , Ti+1 ) ) Ft = i P (t, Ti+1 )Ei+1 (L(Ti , Ti+1 ) )+ | Ft
+
with
1
(x) =
2
ey
2 /2
x R,
dy,
where
log(F (t, Ti , Ti+1 )/) + i2 (t)(Ti t)/2
d1 =
,
i (t) Ti t
p
d2 = d1 i (t) Ti t,
and
Z Ti
1
|i (t)| =
i (s) i (s)ds.
(5.1)
Ti t t
This formula can be used to recover the caplet volatilities iB (t) from market data as
2
in the following table, where the time to maturity Ti t is in ordinate and the period
Tj Ti is in abscissa, cf. Figure 5.1 below. This table and other data used in this
paper has been communicated to the authors in 2005 by IXIS Corporate Investment
Bank, Paris.
2D
1M
2M
3M
6M
9M
1Y
2Y
3Y
4Y
5Y
6Y
7Y
8Y
9Y
10Y
12Y
15Y
20Y
25Y
30Y
VolCapAttheMoney
1M
3M
6M
9,25
9
15,35
15,1
15,75
15,5
15,55
15,3
17,55
17,3
18,35
18,1
19,25
19
17,85
17,6
16,8
16,55
15,6
15,35
14,65
14,4
13,8
13,55
13,35
13,1
13,1
12,85
12,75
12,5
12,4
12,15
11,85
11,6
11,25
11
10,45
10,2
9,7
9,45
9,05
8,8
12M
8,85
14,95
15,35
15,15
17,15
17,95
18,85
17,45
16,4
15,2
14,25
13,45
13
12,75
12,4
12,05
11,5
10,9
10,1
9,35
8,7
2Y
18,6
17,6
18,1
18,6
18,7
18,3
17,9
16,3
15,2
14,4
13,4
12,85
12,3
11,97
11,63
11,3
10,8
10,2
9,5
8,8
8,1
3Y
18
18,03
18,41
18,79
18,28
17,76
17,25
15,96
15,38
14,79
14,5
14,19
13,88
13,65
13,43
13,5
13,22
13
11,9
11,68
11,45
4Y
16,8
16,83
17,11
17,39
16,98
16,56
16,15
15,16
14,58
14,19
13,97
13,66
13,35
13,15
12,96
13,02
12,75
12,55
11,55
11,33
11,1
5Y
15,7
15,73
16,01
16,29
15,88
15,51
15,15
14,46
13,98
13,69
13,53
13,17
12,81
12,65
12,49
12,53
12,28
12,1
11,2
10,98
10,75
7Y
14,7
14,73
15,01
15,29
14,98
14,66
14,35
13,86
13,58
13,29
13,2
12,89
12,58
12,42
12,26
12,25
12,01
11,85
11,05
10,83
10,6
10Y
13
13,03
13,26
13,49
13,48
13,31
13,15
12,96
12,88
12,79
12,8
12,54
12,28
12,12
11,96
11,89
11,69
11,57
11,03
10,88
10,72
11,3
11,33
11,56
11,79
11,98
12,01
12,05
12,06
12,18
12,29
12,4
12,14
11,88
11,75
11,63
11,5
11,3
11,15
10,8
10,55
10,3
k (L(Tk , Tk+1 ) )+
k=i
since they can be decomposed into a sum of caplets and are priced at time t [0, Ti ]
as
j1
X
k=i
11
!+
k P (Ti , Tk+1 )(F (Ti , Tk , Tk+1 ) )
k=i
the positive part can not be taken out of the sum, and in general the price of the
swaption is smaller than the value of the corresponding cap.
This swaption is priced at time t [0, Ti ] as
!+
j1
X
R Ti
E e t rs ds
k P (Ti , Tk+1 )(F (Ti , Tk , Tk+1 ) )
F t
k=i
= E e
!+
j1
R Ti
t
rs ds
P (Ti , Ti ) P (Ti , Tj )
k P (Ti , Tk+1 )
Ft
k=i
"
= E e
R Ti
t
rs ds
j1
X
k P (Ti , Tk+1 ) (S(Ti , Ti , Tj ) )+ Ft
k=i
= P (t, Ti )Ei
" j1
X
j P (Ti , Tk+1 ) (S(Ti , Ti , Tj ) ) Ft
i
h k=i
= P (t, Ti )Ei P (Ti , Ti , Tj ) (S(Ti , Ti , Tj ) )+ Ft
i
h
= P (t, Ti , Tj )Ei,j (S(Ti , Ti , Tj ) )+ Ft ,
(5.2)
R Ti
rs ds
P (Ti , Ti , Tj )
,
P (t, Ti , Tj )
t [0, Ti ],
dPi,j|Ft
P (t, Ti )P (Ti , Ti , Tj )
=
,
dPi|Ft
P (t, Ti , Tj )
t [0, Ti ],
i.e.
1 i < j n.
Swaption prices can be computed using the dynamics of F (t, Tk , Tk+1 ) under Pi , 1
i k < j n, but the market practice is to use approximation formulas. When
j = i + 1 we have Pi,i+1 = Pi , the swaption price equals
i
h
P (t, Ti )Ei (L(Ti , Ti+1 ) )+ Ft
12
and is approximated by
P (t, Ti )Bl(, F (t, Ti , Ti+1 ), i (t), 0, Ti t).
The swaption approximation formula extends to general indices 1 i < j n as
P (t, Ti , Tj )Bl(, S(t, Ti , Tj ), i,j (t), 0, Ti t),
(5.3)
where
Z Ti
j1 i,j
X
vl (t)vli,j
0 (t)F (t, Tl , Tl+1 )F (t, Tl0 , Tl0 +1 )
|i,j (t)| =
l (s) l0 (s)ds,
2
(T
t)|S(t,
T
,
T
)|
i
i
j
t
l,l0 =i
2
(5.4)
t [0, Ti ], and
vli,j (t) =
l P (t, Tl+1 )
.
P (t, Ti , Tj )
2D
1M
2M
3M
6M
9M
1Y
2Y
3Y
4Y
5Y
6Y
7Y
8Y
9Y
10Y
12Y
15Y
20Y
25Y
30Y
VolSwaptionAtTheMoney
1Y
2Y
3Y
18,6
18
17,6
18
18,1
18,35
18,6
18,7
18,7
18,1
18,3
17,5
17,9
16,9
16,3
15,2
15,2
14,2
14,4
13,2
13,4
12,4
12,85
11,95
12,3
11,5
11,97
11,13
11,63
10,77
11,3
10,4
10,8
10,04
10,2
9,5
9,5
8,8
8,8
8,1
8,1
7,4
4Y
16,8
16,8
17,05
17,3
16,8
16,3
15,8
14,4
13,4
12,6
11,9
11,45
11
10,67
10,33
10
9,58
9,1
8,5
7,9
7,3
5Y
15,7
15,7
15,95
16,2
15,7
15,25
14,8
13,7
12,8
12,1
11,5
11
10,5
10,2
9,9
9,6
9,28
8,8
8,2
7,6
7
6Y
14,7
14,7
14,95
15,2
14,8
14,4
14
13,1
12,4
11,7
11,2
10,75
10,3
10
9,7
9,4
9,02
8,6
8
7,4
6,8
7Y
13,8
13,8
14
14,2
13,9
13,6
13,3
12,6
12
11,5
11
10,55
10,1
9,8
9,5
9,2
8,92
8,5
8
7,5
7
8Y
13
13
13,2
13,4
13,3
13,05
12,8
12,2
11,7
11,2
10,8
10,4
10
9,7
9,4
9,1
8,76
8,4
8
7,6
7,2
9Y
12,3
12,3
12,55
12,8
12,7
12,55
12,4
11,9
11,5
11
10,7
10,25
9,8
9,53
9,27
9
8,66
8,3
8
7,7
7,4
10Y
11,8
11,8
12
12,2
12,2
12,1
12
11,6
11,2
10,8
10,5
10,1
9,7
9,43
9,17
8,9
8,56
8,2
7,9
7,6
7,3
25Y
11,3
11,3
11,5
11,7
11,8
11,75
11,7
11,3
11
10,7
10,4
10
9,6
9,33
9,07
8,8
8,46
8,1
7,9
7,7
7,5
9,3
9,3
9,45
9,6
9,7
9,7
9,7
9,3
9,2
8,8
8,6
8,3
8
7,83
7,67
7,5
7,38
7,2
6,9
6,6
6,3
0.2
0.18
0.16
0.14
0.12
9
8
7
0.1
6
5
0.08
4
0
3
2
2
5
1
8
9 0
j1
X
(5.5)
l,l0 =i
vli,j (t)vli,j
0 (t)F (t, Tl , Tl+1 )F (t, Tl0 , Tl0 +1 ) B
l (t)lB0 (t)Cort (F (Ti , Tl , Tl+1 ), F (Ti , Tl0 , Tl0 +1 )),
2
|S(t, Ti , Tj )|
1 i < j n, where iB (t) are the Black caplet volatilities and
Covt (F, G) = E[F G | Ft ] E[F | Ft ]E[G | Ft ].
This formula can be used to reduce the numerical instabilities observed in the calibration procedure, see the next section.
LIBOR calibration
1 i n,
14
appearing in the BGM model from the data of caps and swaptions prices observed on
the market. This involves several computational and stability issues. Let
gi2 (t) = i (t) i (t),
i = 1, . . . , n,
and
i,j (t) =
i (t) j (t)
1/2
1/2
i (t) i (t)
j (t) j (t)
i (t) j (t)
,
=
gi (t)gj (t)
i, j = 1, . . . , n.
i = 1, . . . , n,
where
g(t) = g + (1 + at g )ebt ,
a, b, g > 0,
1
=
Ti t
Z
t
Ti
i (s)
1
i (s)ds =
Ti t
Ti
gi2 (s)ds
t)|S(t,
T
,
T
)|
i
i
j
t
l,l0 =i
v
u j1
u X l,l0 l,l0 (t) B (t) B0 (t)v i,j (t)v i,j
l
l
l
l0 (t)F (t, Tl , Tl+1 )F (t, Tl0 , Tl0 +1 )
= t
(Ti t)|S(t, Ti , Tj )|2
l,l0 =i
15
Tl t Tl0 t 0 i (g + (1 g )ebs )2 ds
qR
= qR
,
Tl t
Tl0 t
bs
2
bs
2
(g + (1 g )e ) ds t
(g + (1 g )e ) ds
0
and a is set equal to 0.
Following again [7] we minimize the mean square distance
v
u
2
k
n B
X
X
u
(t)
(t)
2
i,j
i,j
,
RMS(b, g , 1 , 2 , ) := t
B
(n 1)(n 2) i=1 j=i+1
i,j
(t)
where n is the number of tenor dates (in multiples of one year) and k is the maximum
number of swaption maturities used in the calibration, with non-available data treated
as zero in the sum. The data of discount factors and swap rates are interpolated with
a fixed tenor = half year.
Minimization is done using the Broyden-Fletcher-Goldfarb-Shanno (BFGS) gradient
descent method for nonconvex objective functions. The volatilities computed in this
way are given in Figure 6.1, where the index i refers to Ti t and j refers to Tj Ti :
0.2
0.18
0.16
0.14
0.12
9
8
7
0.1
6
5
0.08
4
0
3
2
2
5
1
8
9 0
0.2
0.18
0.16
0.14
0.12
9
8
7
0.1
6
5
0.08
4
0
3
2
2
5
1
8
9 0
M SF
which yields an expression of i,j
(t) as a function
M SF
i,j
(t, b, g , 1 , 2 , )
of b, g , 1 , 2 , , cf. [7]. The mean square distance associated to the MSF formula
is then defined as
v
u
u
M SF
RMS
(b, g , 1 , 2 , ) = t
2
k
n B
M SF
X
X
i,j (t) i,j
(t)
2
,
B
(n 1)(n 2) i=1 j=i+1
i,j
(t)
used in each calibration is denoted by UpToMat. The total number of swaptions used
is bounded by nk k(k + 1)/2.
UpToMat
1
2
3
4
5
6
7
8
9
10
12
15
#swaptions
10
20
30
40
50
60
70
80
90
100
110
120
b
5.03
5.03
5.04
5.03
5.04
5.03
5.02
5.02
5.02
5.04
5.03
5.03
g
1
0.85 1.95
0.71 2.00
0.73 1.86
0.72 2.00
0.70 1.27
0.65 0.56
0.60 0.15
0.60 0.01
0.72 0.70
0.63 0.01
0.65 0.01
1.00 0.701
0.15 0.010
0.13 0.11
0.13 0.08
0.09 0.09
0.03 0.06
0.00 0.03
0.00 0.01
0.00 0.09
0.00 0.01
0.00 0.01
0.00 0.004
0.00 0.002
RMS
0.008
0.010
0.010
0.010
0.011
0.011
0.012
0.012
0.013
0.012
0.012
0.014
2Y
1Y
5Y
10Y
15Y
20Y
30Y
5
17,65
14,4
9,6
8,1
6,63
5,05
1Y
5Y
10Y
15Y
20Y
30Y
5
17,65
13,5
8,9
6,8
5,4
3,8
2
3,75
1,9
0,4
0,3
0,78
1,3
1
1,25
0,5
1,7
1,9
1,88
2,05
0,5
0,25
1,6
2,6
2,6
2,38
2,15
0,25
0,25
1,9
2,75
2,7
2,48
2,25
2
3,75
1,9
0,1
0,85
1,05
1,55
1
1,25
0,7
1,6
2,1
2,1
2,05
0,5
0,25
1,55
2,3
2,4
2,25
2,05
0,25
0,25
1,9
2,35
2,55
2,3
2,05
ShiftcomparedtoATM
0
0,25
0,35
0,3
2,1
2,1
2,95
2,9
2,85
2,85
2,58
2,58
2,25
2,25
0,5
0,1
2
2,85
2,8
2,53
2,25
1
0,45
1,95
2,8
2,7
2,48
2,25
2
1,55
1,2
2
2,05
2,13
2,2
5
5,65
1,8
0,35
0,15
0,03
0,25
10
11,4
5,9
3,55
3,1
2,73
2,15
20
16,65
10,1
7,4
6,5
5,73
4,8
0,5
0,1
1,85
2,3
2,45
2,2
1,9
1
0,45
1,55
1,95
2,05
1,8
1,7
2
1,75
0,5
1
1,2
1,15
1,05
5
5,9
2,75
1,7
1,4
1,3
1,1
10
11,9
6,95
5,25
4,6
4,2
3,6
20
17,3
11,05
8,8
7,75
7
6
2
1,55
1,2
2
2,05
2,13
5
5,65
1,8
0,35
0,15
0,03
10
11,4
5,9
3,55
3,1
2,73
20
16,65
10,1
7,4
6,5
5,73
10Y
30Y
0
0,35
2
2,4
2,65
2,35
2,05
0,25
0,3
2
2,4
2,5
2,3
2
18
VolCapCMSSmile
1Y
5Y
10Y
15Y
20Y
2Y
5
17,65
14,4
9,6
8,1
6,63
2
3,75
1,9
0,4
0,3
0,78
1
1,25
0,5
1,7
1,9
1,88
0,5
0,25
1,6
2,6
2,6
2,38
ShiftcomparedtoATM
0,25
0
0,25
0,25
0,35
0,3
1,9
2,1
2,1
2,75
2,95
2,9
2,7
2,85
2,85
2,48
2,58
2,58
0,5
0,1
2
2,85
2,8
2,53
1
0,45
1,95
2,8
2,7
2,48
VolCapLiborSmile
1M
1Y
5Y
10Y
30Y
5
20
17,75
15,5
9
1Y
5Y
10Y
30Y
5
20
17,75
15,5
9
1Y
5Y
10Y
30Y
5
20
17,75
15,5
9
2
6,5
4,7
3,75
1,75
1
2,75
1,75
0,9
0,25
0,5
1
0,6
0,3
0,15
0,25
0,5
0,15
0,05
0
ShiftcomparedtoATM
0
0,25
0
0,05
0
0,05
0
0,03
0
0
0,5
0,1
0,05
0,05
0
1
0,15
0,1
0,05
0,05
2
0,5
0,4
0,25
0,1
5
3,5
3,2
2,3
1,5
10
7
7
5,5
4
20
10,5
10,5
8
6
2
6,5
4,7
3,75
1,75
1
2,75
1,75
0,9
0,25
0,5
1
0,6
0,3
0,15
0,25
0,5
0,15
0,05
0
0
0
0
0
0
0,5
0,1
0,05
0,05
0
1
0,15
0,1
0,05
0,05
2
0,5
0,4
0,25
0,1
5
3,5
3,2
2,3
1,5
10
7
7
5,5
4
20
10,5
10,5
8
6
2
6,5
4,7
3,75
1,75
1
2,75
1,75
0,9
0,25
5
3,5
3,2
2,3
1,5
10
7
7
5,5
4
20
10,5
10,5
8
6
3M
6M
References
0,25
0,05
0,05
0,03
0
Figure
6.5: Smile
data
for0,5 swaptions.
0,5
0,25
0
0,25
1
1
0,6
0,3
0,15
0,5
0,15
0,05
0
0
0
0
0
0,05
0,05
0,03
0
0,1
0,05
0,05
0
0,15
0,1
0,05
0,05
2
0,5
0,4
0,25
0,1
[1] A. Brace, D. Gatarek, and M. Musiela. The market model of interest rate dynamics. Math.
Finance, 7(2):127155, 1997.
[2]
D. Brigo and
F. Mercurio. Interest rate
modelstheory and practice. Springer Finance. SpringerVolCapLiborSmile
1M
ShiftcomparedtoATM
5
2second1 edition,
0,5 2006.
0,25
0
0,25
0,5
1
2
5
10
20
Verlag, Berlin,
1Y
20
6,5
2,75
0,5
0,05
0,1
0,15
0,5
3,5
10,5
5Y R. A. Carmona
17,75
4,7 and 1,75
0,15
0
0,05 rate 0,05
0,4
3,2
7
10,5
[3]
M. R. 0,6Tehranchi.
Interest
models:0,1 an infinite
dimensional
stochastic
10Y analysis15,5perspective.
3,75
0,9
0,3 Finance.
0,05
0
0,03
0,05 Berlin,
0,05
0,25
2,3
5,5
8
Springer
Springer-Verlag,
2006.
30Y
1,75
0,25
0,15
0,05
0,1
1,5
3M
[5] N. Privault.
An
Elementary
Introduction
to 0,25
Stochastic
Interest
Rate
Modeling,
volume
12 of
5
2
1
0,5
0,25
0
0,5
1
2
5
10
20
Statistical
Science
&0 Applied
Probability.
World
Scientific
Co.
1Y Advanced
20 Series
6,5 on 2,75
1
0,5
0,05
0,1
0,15
0,5
3,5
7 Publishing
10,5
5Y Inc., River
17,75 Edge,
4,7 NJ,
1,752008. 0,6ISBN0,15
0
0,05
0,05
0,1
0,4
3,2
7
10,5
978-981-283-273-3.
10Y
15,5
3,75
0,9
0,3
0,05
0,03
0,05
0,05
0,25
30Y R. Rebonato.
9
1,75
0,25
0,15Option0 Models.
0
0 Wiley
0 & Sons,
0,05
0,1
[6]
Interest-Rate
John
1996.
2,3
1,5
5,5
4
8
6
[7] J. Schoenmakers. Calibration of LIBOR models to caps and swaptions: a way around intrinsic
instabilities via parsimonious structures and a collateral market criterion. WIAS Preprint No
740, Berlin, 2002.
[8] J. Schoenmakers. Robust LIBOR modelling and pricing of derivative products. Chapman &
Hall/CRC Financial Mathematics Series. Chapman & Hall/CRC, Boca Raton, FL, 2005.
19