Professional Documents
Culture Documents
Convexity Adjustment: A User's Guide: Yan Zeng Version 1.0.1, Last Revised On 2015-02-14
Convexity Adjustment: A User's Guide: Yan Zeng Version 1.0.1, Last Revised On 2015-02-14
Yan Zeng
Version 1.0.1, last revised on 2015-02-14
Abstract
Elements of convexity adjustment.
Contents
1 Introduction 3
1
∫∞ ∫ K0
D Computation of K0
C(K)dK and 0
P (K)dK 30
2
1 Introduction
In this note, we summarize various results on convexity adjustment. The exposition is based on Boenkost
and Schmidt [1], [2], Hagan [3], Hull [5], Hunt and Kennedy [4], Lesniewski [6], Pelsser [8], and Piza [9].
Denote by P (t, T ) (0 ≤ t ≤ T ) the time-t value of a zero coupon bond with maturity T . τ (S, T ) is the
year fraction between time S and time T (S < T ). The simply-compounded forward interest rate F (t; S, T )
is defined as ( )
1 P (t, S)
F (t; S, T ) = −1 .
τ (S, T ) P (t, T )
Suppose Tα < Tα+1 < · · · < Tβ is a set of future times such that the LIBOR rate is reset at Tα , · · · , Tβ−1
and is paid at Tα+1 , · · · , Tβ for a floating-rate note. The forward swap rate Sα,β (t) at time t for the set of
times T = {Tα , Tα+1 , · · · , Tβ } (t ≤ Tα ) and year fractions τ = {τα+1 , · · · , τβ } (τi = τ (Ti−1 , Ti )) is defined as
P (t, Tα ) − P (t, Tβ )
Sα,β (t) = ∑β .
i=α+1 τi P (t, Ti )
These two rates often appear as the underlyings in interest rate derivatives, and will serve as the prototype
for convexity adjustment.
Under the forward measure QT for which P (·, T ) is the numeraire, F (t; S, T ) is a martingale and therefore
E QT [L(S, T )] = F (0; S, T ). This leads to the pricing formula of a floater, which resets LIBOR at time S and
makes payment at time T .
2.1.1 LIBOR-in-arrears
For LIBOR-in-arrears, we need to evaluate E QS [L(S, T )], where QS is the forward measure for which
P (·, S) is the numeraire. The goal is to express E QS [L(S, T )] in terms of the forward rate F (0; S, T ) plus
some “convexity” adjustment (recall E QT [L(S, T )] = F (0; S, T )):
[ ]
QS QT P (S, S)/P (0, S)
E [L(S, T )] = E L(S, T )
P (S, T )/P (0, T )
[ ]
P (0, T )
= E QT L(S, T ) · (1 + τ (S, T )L(S, T )) ·
P (0, S)
[ ]
1 + τ (S, T )L(S, T )
= E QT L(S, T ) ·
1 + τ (S, T )F (0; S, T )
F (0; S, T ) + τ (S, T )E QT [L2 (S, T )]
=
1 + τ (S, T )F (0; S, T )
3
Under the so-called market model which is the model underlying the market valuation for caps, the
LIBOR L(S, T ) is lognormal under QT with volatility σ,
{ }
1 2
L(S, T ) = F (S; S, T ) = F (0; S, T ) exp σWS − σ S ,
2
2
where W is a standard Brownian motion. In this case, VarQT (L(S, T )) = F 2 (0; S, T )(eσ S
− 1) and formula
(1) becomes
[ 2
]
QS τ (S, T )F (0; S, T )(eσ S − 1)
E [L(S, T )] = F (0; S, T ) 1 + . (2)
1 + τ (S, T )F (0; S, T )
2.1.2 LIBOR paid at arbitrary time under the linear rate model
Suppose the payment is made at an arbitrary time T ′ ∈ [S, T ]. This is the case of Asian floater, where
S and T are the starting time and ending time of a coupon period, respectively. Then
[ ′ ′
]
QT ′ QT P (S, T )/P (0, T )
E [L(S, T )] = E L(S, T )
P (S, T )/P (0, T )
The linear rate model assumes
P (S, T ′ )
= a + b(T ′ )L(S, T ), ∀T ′ ∈ [S, T ]
P (S, T )
L(T ′ , T ) F (0; T ′ , T )
= , ∀T ′ ∈ [S, T ] (3)
L(S, T ) F (0; S, T )
The ratio of LIBOR rates over the interval [T ′ , T ] and [S, T ] is equal to the ratio of time-zero forward
rates over the same intervals.
Note the case of LIBOR-in-arrears, where T ′ = S, satisfies the assumption of linear rate model.
Under the linear rate model assumption, we easily deduce that
[ ]
1 − P (0, T )/P (0, T ′ )
E QT ′
[L(S, T )] = F (0; S, T ) 1 + VarQT (L(S, T )) , ∀T ′ ∈ [S, T ] (4)
F 2 (0; S, T )
Remark 1. The original motivation for the linear rate model is probably the consideration that the “nat-
ural rate” under T -forward measure QT is L(S, T ). So one would like to use L(S, T ) to approximate
P (S, T ′ )/P (S, T ), and linear function is obviously the simplest. This idea can be generalized to that of
making the Radon-Nikodym derivative a function of the payout rate.
4
Remark 2. For T ′ = S, formula (4) reduces to formula (1).
Under the market model where L(S, T ) is lognormal under QT with volatility σ,
2.2 CMS
∑β
From the definition of forward swap rate Sα,β (t), if we choose the annuity Ntα,β = i=α+1 τi P (t, Ti )
as numeraire and denote by Qα,β the associated martingale measure (the “swap measure”), we have by
martingale property
α,β
E Q [Sα,β (Tα )] = Sα,β (0).
If the payment is to be paid at some time T ′ > Tα , we need to compute under the T ′ -forward measure QT ′
[ ] [ ]
′ ′ ′
QT ′ Qα,β P (Tα , T )/P (0, T ) N0α,β Qα,β P (Tα , T )
E [Sα,β (Tα )] = E Sα,β (Tα ) = E Sα,β (Tα ) .
NTα,β /N0α,β P (0, T ′ ) NTα,β
α α
The goal is to express E QT ′ [Sα,β (Tα )] in terms of the time-zero swap rate S α,β (0) plus some “convexity”
adjustment.
2.2.1 CMS paid at arbitrary time under the linear swap rate model
∑β
Under the swap measure Qα,β associated with the annuity numeraire Ntα,β = i=α+1 τi P (t, Ti ), the
entity most convenient for computation is the swap rate Sα,β (Tα ). Therefore, a natural assumption for the
so-called linear swap rate model is
P (Tα , T ′ )
= a + b(T ′ )Sα,β (Tα ), T ′ ≥ Tα .
NTα,β
α
To determine a and b, we first take expectation of both sides under the swap measure and use the
martingale property to get
P (0, T ′ )
= a + b(T ′ )Sα,β (0).
N0α,β
[ ′
]
This gives b(T ′ ) = Sα,β1 (0) PN(0,T
α,β
)
− a . To deduce the second equation for a and b, we note
0
∑β ( ) ∑
i=α+1 τi P (Tα , Ti )
∑
β
Sα,β (Tα )
β
Sα,β (Tα )
1 = = τi [a + b(Ti )Sα,β (Tα )] = a 1 − τi + .
NTα,β
α i=α+1
S α,β (0) i=α+1
Sα,β (0)
1
Therefore, we can solve for a: a = ∑β
τi
. In summary, the linear swap rate model makes the
i=α+1
assumption
P (Tα ,T ′ )
= a + b(T ′ )Sα,β (Tα ), T ′ ≥ Tα
NTα,β
α
a = ∑β 1 τ (5)
i=α+1 i [ ]
′
′
b(T ) = Sα,β1 (0) P (0,T )
− ∑
1
, T ′ ≥ Tα
N α,β 0
β
τ
i=α+1 i
5
and consequently
P (0,Tα )−P (0,Tβ )
1− ∑
Sα,β (0)P (0,T ′ ) β
i=α+1 τi
E QT ′ [Sα,β (Tα )] = Sα,β (0) 1 + 2 (0) VarN (Sα,β (Tα )) (6)
Sα,β
where VarN (Sα,β (Tα )) is the variance of the swap rate Sα,β (Tα ) under the swap measure Qα,β .
Under the so-called market model for swpation, it’s assumed the swap rate Sα,β (t) satisfies
where W α,β is a standard Brownian motion under the( swap measure α,β
) Q . The variance of the swap rate
2
2
Sα,β (Tα ) under the swap measure is therefore Sα,β (0) e σα,β Tα
− 1 . Then
[ ( ) ]
P (0, Tα ) − P (0, Tβ ) ( 2
)
E QT ′
[Sα,β (Tα )] = Sα,β (0) 1 + 1− ∑β e σα,β Tα
−1 .
Sα,β (0)P (0, T ′ ) i=α+1 τi
Remark 3. The linear rate model for Libor and CMS can be generalized as follows. Write YS for a floating
rate which is set at time S. Let N , QN denote the natural (“market”) numeraire pair associated with YS
and all we need is
E QN [YS ] = Y0 ,
where Y0 is known and a function of the yield curve P (0, ·) today.
We are interested in today’s price of the rate YS to be paid at some time T ′ ≥ S,
[ ′
]
′ QT ′ QN P (S, T )
P (0, T )E [YS ] = N0 E YS .
NS
P (S, T ′ )
= a + b(T ′ )YS (7)
NS
with some deterministic a, b(T ′ ) which have to be determined accordingly to make the model consistent. We
then have
[ ]
b(T ′ )
E QT ′ [YS ] = Y0 1 + VarQ (Y S ) (8)
Y0 (a + b(T ′ )Y0 ) N
[ ]
b(T ′ )Y0 2
E QT ′ [YS ] = Y0 1 + (e σY S
− 1) .
a + b(T ′ )Y0
Under the linear rate model for Libor, YS = L(S, T ) and NS = P (S, T ); under the linear rate mode for
CMS, YS = Sα,β (Tα ) and NS = NTα,β α
.
As a last comment, the linear approximation of linear rate model does seem very crude at first, but can be
justified by the following argument. Convexity corrections only become sizeable for large maturities. However,
for large maturities the term structure almost moves in parallel. Hence, a change in the level of the long end
′
of the curve is well described by the rate Y . Furthermore, for parallel moves in the curve, the ratio P (S,T
NS
)
is closely approximated by a linear function of Y , which is exactly what the linear rate model does. Hence,
exactly for long maturities the assumptions of the linear rate model become quite accurate. This leads to a
good approximation of the convexity correction for long maturities.
6
2.2.2 CMS paid at arbitrary time under Hagan’s model
As seen in the previous section, the key to the convexity adjustment involving CMS rate, is to express
P (Tα ,T ′ )
NTα,β
as a function of swap rate Sα,β (Tα ):
α
P (Tα , T ′ )
= G(Sα,β (Tα )).
NTα,β
α
∑
β ∑
β−α [ ]
P (t, Ti ) 1/q P (t, Tα ) 1
Ntα,β = P (t, Tα ) τi ≈ P (t, Tα ) = 1−
i=α+1
P (t, Tα ) j=1
[1 + Sα,β (t)/q]j Sα,β (t) (1 + Sα,β (t)/q)n
Here the forward swap rate Sα,β (t) is used as the discount rate, since it represents the average rate over the
life of the reference swap. In a similar spirit, the zero coupon bond for the pay date T ′ is approximated as
P (t, Tα )
P (t, T ′ ) ≈
(1 + Sα,β (t)/q)∆
T ′ −Tα
where ∆ = Tα+1 −Tα . Combined, the standard “bond math model” leads to the approximation
Model 2: “Exact yield” model. We can account for the reference swap’s schedule and day count
exactly by approximating
P (t, Ti ) ∏
i
1
≈
P (t, Tα ) j=α+1 1 + τj Sα,β (t)
and
P (t, Tα )
P (t, T ′ ) ≈
(1 + τα+1 Sα,β (t))∆
T ′ −Tα
where ∆ = Tα+1 −Tα . Therefore
( )
∑
β ∏
i
1 P (t, T ) ∏β
1
τi = α
Ntα,β ≈ P (t, Tα ) 1−
i=α+1 j=α+1
1 + τj Sα,β (t) Sα,β (t) i=α+1
1 + τi Sα,β (t)
and
P (t, T ′ ) Sα,β (t) 1
G(Sα,β (t)) = ≈ ∏β
Ntα,β ∆
(1 + τα+1 Sα,β (t)) 1 − i=α+1 1
1+τi Sα,β (t)
This approximates the yield curve as flat and only allows parallel shifts, but has the schedule right.
Model 3: Parallel shifts. This model takes into account the initial yield curve shape, which can be
significant in steep yield curve environments. We still only allow parallel yield curve shifts, so we approximate
P (t, Ti ) P (0, Ti ) −(Ti −Tα )s
= e
P (t, Tα ) P (0, Tα )
7
where s is the amount of the parallel shift to be determined. To determine s, note
∑
β
P (t, Ti ) ∑β
P (0, Ti ) −(Ti −Tα )s
Ntα,β = P (t, Tα ) τi = P (t, Tα ) τi e
i=α+1
P (t, Tα ) i=α+1
P (0, Tα )
and hence
P (t, Tα ) − P (t, Tβ ) P (0, Tα ) − P (0, Tβ )e−(Tβ −Tα )s
Sα,β (t) = = ∑β .
Ntα,β i=α+1 τi P (0, Ti )e
−(Ti −Tα )s
and
′ ′
P (t, T ′ )/P (t, Tα ) e−(T −Tα )s
Sα,β (t)e−(T −Tα )s
G(Sα,β (t)) = =[ ] =
Ntα,β /P (t, Tα ) 1−
P (0,Tβ ) −(Tβ −Tα )s
1−
P (0,Tβ ) −(Tβ −Tα )s
P (0,Tα ) e
P (0,Tα ) e /Sα,β (t)
∑
β
Sα,β (t) τi P (0, Ti )e−(Ti −Tα )s = P (0, Tα ) − P (0, Tβ )e−(Tβ −Tα )s .
i=α+1
This model’s limitations are that it allows only parallel shifts of the yield curve and it presumes perfect
correlation between long and short term rates.
Model 4: Non-parallel shifts. We can allow non-parallel shifts by approximating
P (t, Ti ) P (0, Ti ) −[h(Ti )−h(Tα )]s
= e
P (t, Tα ) P (0, Tα )
Then similar to Model 3, we have
′
Sα,β (t)e−[h(T )−h(Tα )]s
G(Sα,β (t)) = P (0,Tβ ) −[h(Tβ )−h(Tα )]s
1− P (0,Tα ) e
∑
β
Sα,β (t) τi P (0, Ti )e−[h(Ti )−h(Tα )]s = P (0, Tα ) − P (0, Tβ )e−[h(Tβ )−h(Tα )]s .
i=α+1
To complete the model, we need to select the function h(·) which determines the shape of the non-parallel
shift. This is often done by postulating a constant mean reversion
1[ ]
h((T ) − h(Tα ) = 1 − e−κ(T −Tα ) .
κ
Alternatively, one can choose h(·) by calibrating the vanilla swaptions which have the same start date Tα
and varying end dates to their market prices.
P (Tα ,T ′ )
In either case, under the assumption NTα,β
= G(Sα,β (Tα )), we have
α
{[ ] }
G(Sα,β (Tα ))
E QT ′ [Sα,β (Tα )] = Sα,β (0) + E Qα,β − 1 Sα,β (Tα ) (9)
G(Sα,β (0))
8
2.3 Hull’s approach to convexity adjustment (LIBOR-in-arrears)
This section is based on Hull [5], Chapter 20. We recall the relation between forward LIBOR rate
F (t; S, T ) and zero coupon bond price P (t, ·) is given by
P (t, T ) 1
= .
P (t, S) 1 + τ (S, T )F (t; S, T )
1
Write yt for F (t; S, T ) and define G(y) = 1+τ (S,T )y . Then Taylor expansion gives
P (t, T ) 1
= G(yt ) ≈ G(y0 ) + G′ (y0 )(yt − y0 ) + G′′ (y0 )(yt − y0 )2
P (t, S) 2
[ ]
Under the S-forward measure QS , E QS PP (t,T ) P (0,T )
(t,S) = P (0,S) = G(y0 ). So taking expectation of both sides of
( )
the Taylor expansion, we have G(y0 ) ≈ G(y0 ) + G′ (y0 ) E QS [yt ] − y0 + 12 G′′ (y0 )E QS [(yt − y0 )2 ]. This gives
1 G′′ (y0 ) QS
E QS [yt ] ≈ y0 − E [(yt − y0 )2 ].
2 G′ (y0 )
Let t = S and approximate E QS [(yS − y0 )2 ] by σ 2 y02 S with σ the volatility of y. We then have
[ ]
τ (S, T )F (0; S, T )σ 2 S
E [L(S, T )] ≈ F (0; S, T ) 1 +
QS
1 + τ (S, T )F (0; S, T )
This is the first order approximation of convexity adjustment formula (2). Note the approximation of
E QS [(yS − y0 )2 ] by σ 2 y02 S is more or less equivalent to assuming yS is lognormally distributed.
2.4 Option on interest rates paid at arbitrary time under linear rate model
In this section, we investigate European options on interest rates like LIBOR L(S, T ) for period [S, T ]
or CMS rates Sα,β (Tα ) for tenure structure T = {Tα , Tα+1 , · · · , Tβ }. The payment date of the option is an
arbitrary time point T ′ with T ′ ≥ S or T ′ ≥ Tα , respectively.
Of particular interest are caps and floors or binaries. For standard caps and floors on LIBOR, we have
T ′ = T and the standard market model postulates a lognormal distribution of L(S, T ) under the forward
measure QT . For standard option on a swap rate Sα,β (Tα ), i.e. swaptions, the market uses a lognormal
distribution for Sα,β (Tα ) under the swap measure Qα,β . However in the general case, i.e. for options on
LIBOR or CMS with arbitrary payment date T ′ , a lognormal model would be inconsistent with the market
model for standard options. For example, if L(S, T ) is lognormal under QT , it cannot be lognormal under
QS in general: [ ]
QT P (0, S) QS f (L(S, T ))
E [f (L(S, T ))] = E .
P (0, T ) 1 + τ (S, T )L(S, T )
We follow the general setup of linear rate model. YS is a floating interest rate which is set at time S and
(N, QN ) denotes the “market” numeraire pair associated with YS . We assume that the distribution of YS
under QN is lognormal with volatility σY ,
{ }
1 2
YS = Y0 exp σY WS − σY S ,
2
where W is a standard Brownian motion. For a payment date T ′ ≥ S, we further assume a linear rate model
of the form (7)
P (S, T ′ )
= a + b(T ′ )YS .
NS
Recall that for the case of YS = L(S, T ), NS = P (S, T ), and T ′ = S, i.e. LIBOR-in-arrears, the assumption
of a linear rate model is trivially satisfied and there is no restriction.
9
We first consider the valuation of standard options on the rate YS but the option payout is at some
arbitrary time T ′ ≥ S. The value of a call option with strike K is then
[ ]
′
[ ] ′
+ P (S, T )
[ ]
P (0, T )E QT ′
(YS − K)+
= N0 E QN
(YS − K) = N0 E QN (YS − K)+ (a + b(T ′ )YS )
NS
√
Y0 Φ(d1 )(a − b(T ′ )K) − aKΦ(d2 ) + b(T ′ )Y02 eσY S Φ(d1 + σY S)
2
= P (0, T ′ )
a + b(T ′ )Y0
It is desirable to be able to use standard valuation formula (i.e. Black’s formula) also for options on
interest rates which are irregularly paid. For this purpose, we assume YS is lognormally distributed under
QT ′ , with adjusted volatility. [ ]
b(T ′ )Y0 2
Using formula (8), i.e. E QT ′ [YS ] = Y0 1 + a+b(T ′ )Y (e
0
σY S
− 1) , and moment matching, we can derive
{ }
cS − 1 (σY∗ )2 S ,
YS ≈ E QT ′ [YS ] exp σY∗ W
2
with E QT ′ [Yi ] given by formula (4), (6), or (8) and W i standard Brownian motion under QT ′ . Suppose the
instantaneous correlation between W 1 and W 2 is ρ, the fair price of the exchange option is then given by
where
( ) ( )
Y20 Y20
ln Y10
+ 12 (σ12 S1 + σ22 S2 − 2σ1 σ2 ρS1 ) ln Y10
− 12 (σ12 S1 + σ22 S2 − 2σ1 σ2 ρS1 )
b1 = √ , b2 = √
σ12 S1 + σ22 S2 − 2σ1 σ2 ρS1 σ12 S1 + σ22 S2 − 2σ1 σ2 ρS1
For details of the computation, see Boenkost and Schmidt [1], Section 4.4, Proposition 7.
Remark 4. The market standard method of valuing options on convexity adjusted rates is to apply the Black
formula using the convexity adjusted rate as the forward rate. But to be conceptually correct, we should also
convexity adjust volatility by formula (10). These two adjustments combined is equivalent to assuming the
rate is lognormal under the T ′ -forward measure QT ′ .
10
3 Convexity adjustment with volatility smile: pricing by replica-
tion
As shown in Section 2 and Section 2.4, pricing irregular interest cash flows such as LIBOR-in-arrears or
CMS requires a convexity correction on the corresponding forward rate. This convexity correction involves
the volatility of the underlying rate as traded in the cap/floor or swaption market. In the same spirit, an
option on an irregular rate, such as an in-arrears cap or a CMS cap, is often valued by applying Black’s
formula with a convexity adjusted forward rate and a convexity adjusted volatility. However, to a large
extent, this approach ignores the volatility smile, which is quite pronounced in the cap/floor market. This
section solves this problem by replicating the irregular interest flow or option with liquidly traded options
with different strikes. As an important consequence, one obtains immediately the respective simultaneous
delta and vega hedges in terms of liquidly traded options.
3.1 Motivation
Example 1. (LIBOR and caplet paid at arbitrary time) Let Y = L(S, T ) be the LIBOR rate for
the period [S, T ]. Suppose our goal is to price an “exotic” payoff f (L(S, T )) at time S, or more generally,
at time T ′ ≥ S. For example, f (L(S, T )) = τ (S, T )L(S, T ) is the payoff of a LIBOR, and f (L(S, T )) =
τ (S, T )[L(S, T ) − K]+ is the payoff of a caplet. How can we use liquid instruments to hedge this exotic
payoff?
In the interest rate derivative market, caplets are quite liquid for (more or less) any strike K ≥ 0.
Therefore, we intend to use caplets as hedging instruments. One difficulty is that by default, a caplet pays
in arrears (i.e. at time T ).
To hedge the exotic payoff, we first need to “match” the payoff times. Since the exotic payoff is either
paid at time S or at time T ′ ≥ S, we discount everything to time S. First, the standard caplet paying at
time T is equivalent to a payoff of
τ (S, T )
P (S, T )τ (S, T )[L(S, T ) − K]+ = [L(S, T ) − K]+
1 + τ (S, T )L(S, T )
( )
P (0,T ′ )
P (S,T ′ ) P (0,T )
−1
at time S. Second, under the linear rate model P (S,T ) = 1+ F (0;S,T ) L(S, T ), the exotic payoff f (L(S, T ))
at time T ′ is equivalent to a payoff of
( )
P (0,T ′ )
P (0,T )
−1
1+ L(S, T )
P (S, T ′ )f (L(S, T )) =
F (0;S,T )
f (L(S, T )) := f1 (L(S, T ))
1 + τ (S, T )L(S, T )
( )
P (0,T ′ )
−1
P (0,T )
1+ y
at time S. Here f1 (y) := F (0;S,T )
1+τ (S,T )y f (y).
In summary, the exotic payoff can be written as f (L(S, T )), paid at time S; and the hedging instruments
can be generalized to the form of g(L(S, T ))[L(S, T ) − K]+ (K ≥ 0), also paid at time S. Here g(y) :=
τ (S,T )
1+τ (S,T )y . We are interested in finding a representation
∫ ∞
f (y) = C + g(y)(y − K)+ dµ(K)
0
with some locally finite signed measure µ on R and some constant C. If such a representation exists, the
+
where V0caplet (K) is the time-zero price of a caplet for the period [S, T ], with strike K.
11
Example 2. (CMS swap and CMS caplet paid at arbitrary time) Let Y = Sα,β (Tα ) be the
swap rate for the tenure structure T = {Tα , Tα+1 , · · · , Tβ }. Suppose our goal is to price an “exotic” payoff
f (Sα,β (Tα )) at time Tα . For example, f (y) = y is a cash flow in a CMS swap (but paid in advance), and
f (y) = (y − K)+ gives a CMS caplet.
In case the payoff is scheduled from a time T ′ ≥ Tα , we can approximately think of a derivative with
payoff
y
for a CMS swap
(1+y)T ′ −Tα
f (y) = +
(y−K) for a CMS caplet;
(1+y)T ′ −Tα
at time Tα .
How can we use liquid instruments to hedge this exotic payoff?
In the interest rate derivative market, swaptions are quite liquid for any strike K ≥ 0. Therefore, we
intend to use swaptions as hedging instruments. There are two types of swaptions: the payoff of a cash-settled
payer swaption at time Tα is
( β )
∑ τ (Ti−1 , Ti )
Ti −Tα
(Sα,β (Tα ) − K)+
i=α+1
[1 + S α,β (Tα )]
and the payoff of a physically-settled payer swaption at time Tα is
( β )
∑
τ (Ti−1 , Ti )P (Tα , Ti ) (Sα,β (Tα ) − K)+ .
i=α+1
The market does not make a significant difference in pricing a cash- or physically-settled swaption, and both
are priced under a lognormal Black model.
Example 3. (CMS swap and CMS caplet paid at arbitrary time under linear rate model)
In Example 2, we approximated CMS swap rate and CMS caplet paid at arbitrary time by
Sα,β (Tα )
T ′ −T for a CMS swap
[1+Sα,β (Tα )] α
In the current example, we want to avoid this approximation but we then need the additional assumption
of linear rate model. The so-called linear swap rate model is
P (Tα ,T ′ )
= a + b(T ′ )Sα,β (Tα ), T ′ ≥ Tα
NTα,β
α
a = ∑β 1 τ
i=α+1 i [ ]
P (0,T ′ )
′
b(T ) = Sα,β (0)1
− ∑β 1
, T ′ ≥ Tα
N α,β 0 τi=α+1 i
∑β
where N·α,β is the annuity numeraire Ntα,β = i=α+1 τi P (t, Ti ),
Consequently, an exotic derivative paying f (Sα,β (Tα )) at some time T ′ ≥ Tα is equivalent to a payoff of
P (Tα , T ′ )f (Sα,β (Tα )) = NTα,β
α
[a + b(T ′ )Sα,β (Tα )]f (Sα,β (Tα ))
at time Tα . Under the swap measure Qα,β associated with the annuity numeraire N·α,β , the price of the
contingent claim is given by
[(a + b(T ′ )Sα,β (Tα ))f (Sα,β (Tα ))]
α,β
N0α,β E Q
The price of physically-settled payer swaption can be written as
α,β [ ]
V0swpt (K) = N0α,β E Q (Sα,β (Tα ) − K)+ .
Comparing these two formulas leads us to looking for a representation of the form
∫
(a + b(T ′ )Y )f (Y ) = C + (Y − K)+ dµ(K).
[0,∞)
12
3.2 Problem formulation and its solution
Consider a financial underlying with price Y at time T . Suppose there is a liquid market for plain vanilla
options on this underlying with all possible strikes K. Somewhat more generally, we suppose that for all
K, the time-zero price V0vanilla (K) of the “plain vanilla” derivative with payoff g(Y )(Y − K)+ at time T is
known. Here g is a deterministic function.
Our goal is to price an exotic contingent claim with payoff f (Y ) at time T , where f is a deterministic
function. The idea is to replicate the exotic payoff f (Y ) by a portfolio of traded derivatives g(Y )(Y − K)+
with different strikes K. If this is possible, then the replication is the key to incorporating the volatility
smile of the liquid options into the pricing of the exotic derivative.
So we are looking for a representation
∫ ∞
f (Y ) = C + g(Y )(Y − K)+ dµ(K) (11)
0
with some locally finite signed measure µ on R+ and some constant C. By risk-neutral pricing, this repre-
sentation will give the time-zero price of the contingent claim f (Y ):
∫ ∞
V0 = C · P (0, T ) + V0vanilla (K)dµ(K) (12)
0
Remark 5. The importance of formula (12) goes beyond the issue of just pricing a derivative with payoff
f (Y ), since it provides us with an explicit strategy for a simultaneous delta and vega hedge of the derivative
f (Y ) in terms of liquidly traded products g(Y )(Y − K)+ . In practice one would discretize the integral
appropriately to get an (approximate) hedging strategy in a finite set of products g(Y )(Y − Ki )+ with
different strikes Ki , i = 1, · · · .
Proposition 3.1. The exotic payoff f (Y ) allows for a replication (11) with some locally finite signed measure
µ on [0, ∞) and some constant C if and only if
(i) limx↓0 f (x) = C;
(ii) the function f −C
g , extended to the domain of definition [0, ∞) by setting
f (0)−C
g(0) = 0, is a difference of
convex functions on [0, ∞).
The measure µ is then generated by the following right continuous generalized dsitrbution function (function
of locally bounded variation) ( )
f (y) − C
dµ(y) = dD+
g(y)
( )
with D+ denoting the right derivative and defining D+ f −C g (0−) = 0.
where µ+ and µ− are the positive and negative parts of µ, respectively. For sufficiency, we apply the integral
representation of convex functions (Proposition B.1)
∫ y ∫ y ∫ y ∫ ∞
f (y) − C f (0) − C
− = µ(K)dK = µ(y)y − Kdµ(K) = (y − K)dµ(K) = (y − K)+ dµ(K)
g(y) g(0) 0 0 0 0
13
where FY (·) is the cumulative distribution function of Y under QT . Since
we have
∫ ∞
D+ V0vanilla (x) = P (0, T )E QT [g(Y )D+ (Y − x)+ ] = −P (0, T )E QT [g(Y )1{x<Y } ] = −P (0, T ) g(y)dFY (y)
x
and
dD+ V0vanialla = P (0, T )g(x)dFY (x).
Therefore by integrating by parts twice, we have (define C = limx↓0 f (x))
∫ ∞
f (x) − C
V0 = C · P (0, T ) + dD+ V0vanilla (x)
g(x)
∫0 ∞
= C · P (0, T ) − D+ V0call (x)df (x)
∫0 ∞
= C · P (0, T ) + V0call (x)d2 f (x).
0
1 + b(T ′ )L(S, T )
τ L(S, T )
1 + τ L(S, T )
( )
P (0,T ′ )
−1 1+b(T ′ )y
′ P (0,T )
at time S, where b(T ) = F (0;S,T ) . Therefore f (y) = 1+τ y τ y and C = limx↓0 f (x) = 0. The standard
caplet paying at time T is equivalent to a payoff of
τ
[L(S, T ) − K]+
1 + τ L(S, T )
f (y) − C
= y[1 + b(T ′ )y],
g(y)
In view of dµ({0}) = µ(0) − µ(0−) = 1, we obtain the formula for the price of a LIBOR expressed in terms
of caplet prices V0caplet (K) with different strikes K:
∫
V0 (τ L(S, T ) paid at time T ′ ≥ S) = V0caplet (0) + 0∞ V0caplet (K)2b(T ′ )dK,
( )
P (0,T ′ )
−1 (13)
b(T ′ ) = P (0,T )
F (0;S,T )
14
Now consider a caplet paid at time T ′ ≥ S. Under the linear rate model, the payoff is equivalent to a
payoff of
1 + b(T ′ )L(S, T )
τ [L(S, T ) − K]+
1 + τ L(S, T )
1+b(T ′ )y
at time S. Therefore f (y) = 1+τ y τ (y − K)+ and C = limx↓0 f (y) = 0. We then have
f (y) − C
= [1 + b(T ′ )y](y − K)+ ,
g(y)
which is a convex function. Accordingly,
( ) {
f (y) − C 1 − b(T ′ )K + 2b(T ′ )y y≥K
µ(y) = D+ = .
g(y) 0 y<K
In view of dµ({K}) = µ(K) − µ(K−) = 1 + b(T ′ )K, we obtain the formula for the price of a caplet expressed
in terms of caplet prices V0caplet (K) with different strikes K:
∫
V0 (τ [L(S, T ) − K]+ paid at time T ′ ≥ S) = V0caplet (K)(1 + b(T ′ )K) + K∞ V0caplet (K)2b(T
e ′ e
)dK
( )
P (0,T ′ )
−1 (14)
b(T ′ ) = P (0,T )
F (0;S,T )
Remark 7. In case the cap market quotes no smile, it is easy to verify that formula (13) is reduced to
formula (2).
3.3.2 CMS swap and CMS caplet paid at arbitrary time discounted by swap rate yield
Consider a CMS swap rate Sα,β (Tα ) paid at time T ′ ≥ Tα . The payoff is approximated by a payoff of
Sα,β (Tα )
(1 + Sα,β (Tα ))T ′ −Tα
at time Tα . Therefore, f (y) = (1+y)yT ′ −Tα and C = limx↓0 f (x) = 0.
The payoff of a cash-settled payer swaption at time Tα is
( β )
∑ τ (Ti−1 , Ti )
Ti −Tα
(Sα,β (Tα ) − K)+
i=α+1
[1 + S α,β (Tα )]
∑
β
τi
g(y) = Ti −Tα
i=α+1
(1 + y)
We then have
f (y) − C y y
= ∑β = ,
g(y) τi
′
DV01(y)
i=α+1 (1+y)Ti −T
∑β τi
where DV01(y) := i=α+1 (1+y)Ti −T ′ , the so-called present (or dollar) value of one basis point factor. It is
f (y)−C
kind of hard to determine the convexity of g(y) , or to find the two convex functions whose difference is
f (y)−C f (y)−C
g(y) . But we can easily see that is absolutely continuous and its derivative is of finite variation
g(y)
locally. Therefore, by Proposition B.2, Proposition 3.1 applies. Accordingly,
( ) DV01(y)−yDV01′ (y)
f (y) − C y≥0
µ(y) = D+ = DV012 (y)
g(y) 0 y<0
15
In view of dµ({0}) = µ(0) − µ(0−) = DV01 1
(0)
= ∑β 1 τ , we obtain the formula for the price of a CMS
i=α+1 i
swpt
swap rate expressed in terms of swaption price V0 (K) with different strikes K:
∫∞ ( ′ )′
V0 (Sα,β (Tα ) paid at time T ′ ≥ Tα ) = V0swpt (K) DV01(K)−K2DV01 (K) dK
V swpt (0)
∑0β +
i=α+1 τi
0 DV01 (K) (15)
DV01(y) = ∑β τi
′
i=α+1 (1+y)Ti −T
Now consider a CMS caplet [Sα,β (Tα ) − K]+ paid at time T ′ ≥ Tα . The payoff is approximated by a
payoff of
[Sα,β (Tα ) − K]+
[1 + Sα,β (Tα )]T ′ −Tα
+
(y−K)
at time Tα . Therefore f (y) = (1+y) T ′ −Tα and C = limx↓0 f (x) = 0.
Like the case of CMS swap rate, by using cash-settled payer swaption as hedging instruments, we have
Accordingly,
( ) ′
DV01(y)−(y−K)DV01 (y)
f (y) − C y≥K
µ(y) = D+ = DV01 2
(y)
g(y) 0 y<K
of a CMS caplet expressed in terms of swaption price V0swpt (K) with different strike K:
V0 ([Sα,β (Tα ) − K]+ paid at time T ′ ≥ Tα ) = V0 (K) + ∫ ∞ V swpt (K)
swpt
e · h(K)
e · dK
e
∑β DV01 (K) K 0
(16)
DV01(y) = τ i
T −T ′
i=α+1 (1+y) i
where
( )′
DV01(y) − (y − K)DV01′ (y)
h(y) =
DV012 (y)
[ ]
(DV01′ (y))2 DV01′′ (y) DV01′ (y)
= (y − K) 2 3 − 2 −2 .
DV01 (y) DV01 (y) DV012 (y)
3.3.3 CMS swap and CMS caplet paid at arbitrary time under linear rate model
Consider a CMS swap rate Sα,β (Tα ) paid at time T ′ ≥ Tα . As explained in Example 3 of Section 3.1, we
are looking for integral representation of the form
∫
[a + b(T ′ )Sα,β (Tα )]Sα,β (Tα ) = C + (Sα,β (Tα ) − K)+ dµ(K).
[0,∞)
Using notation of Proposition 3.1, f (y) = [a + b(T ′ )y]y, g(y) = 1, and C = limx↓0 f (x) = 0. Therefore
( ) {
f (y) − C a + 2b(T ′ )y y ≥ 0
µ(y) = D+ =
g(y) 0 y<0
16
In view of dµ({0}) = µ(0) − µ(0−) = a, we obtain the formula for the price of a CMS swap rate expressed
in terms of swaption price V0swpt (K) with different strikes K:
∫∞
V0 (Sα,β (Tα ) paid at time T ′ ≥ Tα ) = aV0swpt (0) + 0 V0swpt (K) · 2b(T ′ )dK
a = ∑β 1 τ
i=α+1 i [ ] (17)
′
b(T ) = S (0) 1 P (0,T ′ )
− ∑β 1
, T ′ ≥ Tα
α,β N α,β0 τ
i=α+1 i
Now consider a CMS caplet [Sα,β (Tα ) − K]+ paid at time T ′ ≥ Tα . We are looking for integral repre-
sentation of the form
∫
[a + b(T ′ )Sα,β (Tα )][Sα,β (Tα ) − K]+ = C + (Sα,β (Tα ) − K)+ dµ(K).
[0,∞)
Using notation of Proposition 3.1, f (y) = [a+b(T ′ )y](y −K)+ , g(y) = 1, and C = limx↓0 f (x) = 0. Therefore
( ) {
f (y) − C [a − b(T ′ )K] + 2b(T ′ )y y ≥ K
µ(y) = D+ =
g(y) 0 y<K
In view of dµ({K}) = µ(K) − µ(K−) = a + b(T ′ )K, we obtain the formula for the price of a CMS caplet
expressed in terms of swaption price V0swpt (K) with different strikes K:
∫∞
e · 2b(T ′ )dK
V0 ([Sα,β (Tα ) − K]+ paid at time T ′ ≥ Tα ) = [a + b(T ′ )K]V0swpt (K) + K V0swpt (K) e
1
a = ∑β
i=α+1 τi [ ] (18)
b(T ′ ) = S 1 (0) P (0,T
′
)
− ∑β 1
, T ′ ≥ Tα
α,β N α,β
0 τ
i=α+1 i
3.3.4 CMS caplet, floorlet, and floater paid at arbitrary time under Hagan’s model
In Section 2.2.2, we explained several models for the ratio
and
P (0, T ′ )/N0α,β = G(Sα,β (0)),
where G(x) could be
x
· 1
Model 1
1
(1+x/q)∆ 1− (1+x/q)
n
(1+τα+1 x)∆ · 1−∏βi=α+1
x 1
1 Model 2
1+τi x
G(x) = −(T ′ −Tα )s
xe
Model 3
P (0,T )
1− P (0,Tβ ) e−(Tβ −Tα )s
α
′
xe−[h(T )−h(Tα )]s
P (0,T Mdoel 4
β ) −[h(T )−h(Tα )]s
β
1− P (0,T e
α)
∑
β
x τi P (0, Ti )e−(Ti −Tα )s = P (0, Tα ) − P (0, Tβ )e−(Tβ −Tα )s
i=α+1
∑
β
x τi P (0, Ti )e−[h(Ti )−h(Tα )]s = P (0, Tα ) − P (0, Tβ )e−[h(Tβ )−h(Tα )]s .
i=α+1
17
for Model 4.
We first consider CMS caplet. The payoff of [Sα,β (Tα ) − K]+ at time T ′ ≥ Tα is equivalent to a payoff
of
P (Tα , T ′ )[Sα,β (Tα ) − K]+ .
at time Tα . By risk-neutral pricing,
V0 ([Sα,β (Tα ) − K]+ paid at time T ′ ≥ Tα )
[ ]
′
α,β Qα,β P (Tα , T )[Sα,β (Tα ) − K]
+
= N0 E
NTα,β
[ α
]
′
′ Qα,β
P (Tα , T )/NTα,β
= P (0, T )E α
[Sα,β (Tα ) − K] +
P (0, T ′ )/N0α,β
[( ) ]
′ Qα,β
{ } ′ Qα,β
P (Tα , T ′ )/NTα,β
= P (0, T )E [Sα,β (Tα ) − K] +
+ P (0, T )E α
− 1 [Sα,β (Tα ) − K] +
P (0, T ′ )/N0α,β
( )
P (t,T ′ )
where we observe that Ntα,β
is a martingale under swap measure Qα,β and
0≤t≤T ′
[ ]
Qα,β P (Tα , T ′ ) P (0, T ′ )
E =
NTα,β
α
N0α,β
The first term is exactly the price of a European swaption with notional P (0, T ′ )/N0α,β , regardless of
how the swap rate Sα,β (t) is modeled. The second term is the “convexity correction”. Since Sα,β (t) is a
P (Tα ,T ′ )/NTα,β
martingale and α
P (0,T ′ )/N0α,β
− 1 is zero on average, this term goes to zero linearly with the variance of the
swap rate Sα,β (t) and is much smaller than the first term.
Under a general rate model, we assume P (t, T ′ )/Ntα,β = G(Stα,β ). Then for f (x) := [G(x)/G(Sα,β (0)) −
1](x − K), we have the property
∫ ∞ {
′ + ′′ f (x) for x ≥ K
f (K)(x − K) +
+
(x − ζ) f (ζ)dζ =
K 0 for x < K
So ∫ ∞
[G(x)/G(Sα,β (0)) − 1](x − K)+ = f ′ (K)(x − K)+ + (x − ζ)+ f ′′ (ζ)dζ
K
and
[( ) ]
Qα,β
P (Tα , T ′ )/NTα,β
E α
− 1 [Sα,β (Tα ) − K] +
P (0, T ′ )/N0α,β
α,β
{ }
= EQ [G(STα,β
α
)/G(S α,β (0)) − 1](S α,β
Tα − K) +
∫ ∞
= f ′ (K)E Q [(STα,β E Q [(Sα,β (Tα ) − ζ)+ ]f ′′ (ζ)dζ
α,β α,β
α
− K)+ ] +
K
18
α,β
Here C(K) = N0α,β E Q [(Sα,β (Tα ) − K)+ ] is the time zero value of a vanilla payer swaption with strike K.
Repeating the above argument shows that the value of a CMS floorlet is
{ ∫ }
P (0, T ′ ) ′
K
′′
V0 ([K − Sα,β (Tα )]+ paid at time T ′ ≥ Tα ) = [1 + f (K)]P (K) − P (ζ)f (ζ)dζ
N0α,β −∞
α,β
where f (x) is the same as before and P (K) = N0α,β E Q [(K − Sα,β (Tα ))+ ] is the value of the receiver
swaption with strike K.
By the call-put parity and by setting K = Sα,β (0), we have C(Sα,β (0)) − P (Sα,β (0)) = 0. Hence the
value of a CMS floater is
where
fatm (x) = [G(x)/G(Sα,β (0)) − 1](x − Sα,β (0))
As a last word, by call-put parity, we can price an in-the-money caplet or floorlet as a swaplet plus an
out-of-the-money floorlet or caplet.
Analytical formulas: The function G(x) is smooth and slowly varying, regardless of the model used to
obtain it. Since the probable swap rates Sα,β (·) are heavily concentrated around around Sα,β (0), it makes
sense to expand G(x) as
If we take the expansion to the linear term, f (x) becomes a quadratic function
G′ (Sα,β (0))
f (x) ≈ (x − Sα,β (0))(x − K).
G(Sα,β (0))
Then
G′ (Sα,β (0)) G′ (Sα,β (0))
f ′ (K) = [K − Sα,β (0)] , f ′′ (ζ) = 2 .
G(Sα,β (0)) G(Sα,β (0))
P (0,T ′ )
Consequently, we have (note G(Sα,β (0)) = N0α,β
)
and
[∫ ∫ ]
∞ Sα,β (0)
′ ′ ′
V0 (Sα,β (Tα ) paid at time T ≥ Tα ) = P (0, T )Sα,β (0) + 2G (Sα,β (0)) C(ζ)dζ + P (ζ)dζ
Sα,β (0) −∞
19
We note 1
∫ ∞ ∫ ∞
α,β 1 α,β Qα,β { }
C(ζ)dζ = N0α,β EQ [(Sα,β (Tα ) − ζ)+ ]dζ = N0 E ([Sα,β (Tα ) − K]+ )2
K K 2
Therefore, an alternative form of the formulas is
for floater.
To finish the calculation, one needs an explicit model for the swap rate Sα,β (t). There are two simple
models one can use. The first model is Black’s model, which assumes that the swap rate Sα,β (Tα ) is log
normal with a volatility σB :
Sα,β (Tα ) = Sα,β (0)eσB WTα − 2 σB Tα .
1 2
V0caplet
P (0, T ′ ) [
C(K) + G′ (Sα,β (0))N0α,β (Sα,β (0))2 eσB Tα Φ(d3/2 ) − Sα,β (0)(Sα,β (0) + K)Φ(d1/2 )
2
=
N0α,β
]
+Sα,β (0)KΦ(d−1/2 )
V0f loorlet
P (0, T ′ ) [
P (K) − G′ (Sα,β (0))N0α,β (Sα,β (0))2 eσB Tα Φ(−d3/2 ) − Sα,β (0)(Sα,β (0) + K)Φ(−d1/2 )
2
=
N0α,β
]
+Sα,β (0)KΦ(−d−1/2 )
20
The second model is the normal, or absolute model, which assumes that the swap rate follows
dSα,β (t) = σN dWt .
This yields
V0f loater = P (0, T ′ )Sα,β (0) + G′ (Sα,β (0))N0α,β σN
2
Tα
for the CMS floaters,
( )
P (0, T ′ ) ′ Sα,β (0) − K
V0caplet = C(K) + G (Sα,β (0))N0α,β σN
2
Tα Φ √
N0α,β σN Tα
for CMS caplets, and
( )
P (0, T ′ ) ′ K − Sα,β (0)
V0f loorlet = P (K) − G (Sα,β (0))N0α,β σN
2
Tα Φ √
N0α,β σN Tα
for CMS floorlets. We can obtain the normal vol σN by noting that if σB is the lognormal volatility for a
swaption with forward rate Sα,β (0) and strike K, then the normal volatility σN of this swaption is
Sα,β (0) − K 1
σN = σB ( ) 1 2 1 4 2
S (0) 1+
ln α,β K 24 σB Tα+ 5760 σB Tα
Near the money ((Sα,β (0) − K)/K is less than 20% or so), we can replace this formula with
( ) ( )
√ 1
1 + 24
S
ln2 α,β K
(0) 1
+ 1920
S
ln4 α,β K
(0)
σN = σB Sα,β (0)K · 1 2 1 4 T2
1 + 24 σB Tα + 5760 σB α
N.B. The key concern with Black’s model is that it does not address the smiles and/or skews seen in
the marketplace. This can be partially mitigated by using the correct volatilities. For CMS floaters, the
volatility σatm for at-the-money swpations should be used, since the expected value includes high and low
strike swaption equally. For out-of-the-money caplets and floorlets, the volatility σK for strike K should
be used, since the swap rate Sα,β (Tα ) near K provide the largest contribution to the expected value. For
in-the-money options, the largest contributions come from swap rates Sα,β (Tα ) near the mean value Sα,β (0).
Accordingly, call-put parity should be used to evaluate in-the-money caplets and floorlets as a CMS swap
payment plus an out-of-the-money floorlet or caplet.
21
V0 (1{Sα,β (Tα )<K} paid at time T ′ ≥ Tα )
[ ]
P (0, T ′ ) d ′ d
= P (K) − G (Sα,β (0)) (Sα,β (0) − K) P (K) + P (K)
N0α,β dK dK
Under the lognormal model,
C(K) = N0α,β [Sα,β (0)Φ(d1 ) − KΦ(d2 )]
(S )
ln α,β
(0) 2
+ 21 σK Tα √
and d2 = d2 (σK ) = d1 (σK ) − σK Tα . Since
K
where d1 = d1 (σK ) = √
σK Tα
d ∂ ∂ dσK
[Sα,β (0)Φ(d1 ) − KΦ(d2 )] = [Sα,β (0)Φ(d1 ) − KΦ(d2 )] + [Sα,β (0)Φ(d1 ) − KΦ(d2 )]
dK ∂K ∂σ dK
1 √ dσK
′ ′ ′
= − √ [Sα,β (0)Φ (d1 ) − KΦ (d2 )] − Φ(d2 ) + Sα,β (0)Φ (d1 ) Tα
KσK Tα dK
and
√ √
1 (d2 +σK Tα )2
Sα,β (0)Φ′ (d1 ) = Sα,β (0) · √ e− = Sα,β (0)e−d2 σK Tα − 2 σK Tα Φ′ (d2 ) = KΦ′ (d2 ),
1 2
2
2π
α,β [ √ ]
d
we conclude dK C(K) = N0 −Φ(d2 ) + Φ′ (d2 )K dσK
dK Tα . Therefore, the value of a digital call under
lognormal model is
V0 (1{Sα,β (Tα )>K} paid at time T ′ ≥ Tα )
[ ]
dσK √
= P (0, T ′ ) Φ(d2 ) − Φ′ (d2 )K Tα + G′ (Sα,β (0))N0α,β Sα,β (0) {Φ(d1 ) − Φ(d2 )
dK
√ }
dσK
−[Φ′ (d1 ) − Φ′ (d2 )]K Tα
dK
By the call-put parity, the value of a digital put under lognormal model is
V0 (1{Sα,β (Tα )<K} paid at time T ′ ≥ Tα )
= P (0, T ′ ) − V0 (1{Sα,β (Tα )>K} paid at time T ′ ≥ Tα )
[ ]
dσK √
= P (0, T ′ ) Φ(−d2 ) + Φ′ (d2 )K Tα + G′ (Sα,β (0))N0α,β Sα,β (0) {Φ(d2 ) − Φ(d1 )
dK
√ }
dσK
−[Φ′ (d2 ) − Φ′ (d1 )]K Tα
dK
Under the normal model,
[ √ ]
C(K) = N0α,β (Sα,β (0) − K)Φ(d1 ) + Φ′ (d1 )σ Tα
Sα,β (0)−K
where d1 = √
σ Tα
and σ is the normal vol. Since
d [ √ ]
(Sα,β (0) − K)Φ(d1 ) + Φ′ (d1 )σ Tα
dK
∂ [ √ ] ∂ [ √ ] dσ
= (Sα,β (0) − K)Φ(d1 ) + Φ′ (d1 )σ Tα + (Sα,β (0) − K)Φ(d1 ) + Φ′ (d1 )σ Tα
∂K ∂σ dK
√ dσ
′
= −Φ(d1 ) + Φ (d1 ) Tα ,
dK
[ √ dσ ]
d
we conclude dK C(K) = N0α,β −Φ(d1 ) + Φ′ (d1 ) Tα dK . Therefore, the value of a digital call under normal
model is
V0 (1{Sα,β (Tα )>K} paid at time T ′ ≥ Tα )
[ √ dσ ] √ [ ]
′ ′ ′ α,β ′ dσ
= P (0, T ) Φ(d1 ) − Φ (d1 ) Tα + G (Sα,β (0))N0 Φ (d1 ) Tα (Sα,β (0) − K) +σ
dK dK
22
By the call-put parity, the value of a digital put under normal model is
3.3.6 CMS fracl paid at arbitrary time under linear rate model
A floating CMS range accrual pays out
[ ]
leverage · S1 (t10 ) + spread 1{Rmin <S2 (t20 )<Rmax } + fixRate
at time tp ≥ t20 ≥ t10 , where S1 (·) is a swap rate with schedule t10 < t11 < · · · < t1n1 and corresponding coverage
τ11 < · · · < τn11 , and S2 (·) is a swap rate with schedule t20 < t21 < · · · < t2n2 and corresponding coverage
τ12 < · · · < τn22 . The time-zero value of the payoff is
( [ ] [ ])
V0 = P (0, tp ) leverage · E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 ) + spread · E Qtp 1{Rmin <S2 (t20 )<Rmax }
+P (0, tp ) · fixRate
Denote by L1 and L2 the annuities corresponding to S1 and S2 , respectively. Then by formula (5) from
the linear rate model
[ ] [ ]
P (t20 , tp )/P (0, tp )
E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 ) = E L2 · 1 S (t
{Rmin <S2 (t0 )<Rmax } 1 0
2
1
)
L2 (t20 )/L2 (0)
[ ]
= E L2 (1 + a + bS2 (t20 )) · 1{Rmin <S2 (t20 )<Rmax } S1 (t10 )
[ ]
L2 (0) L2 (0)
where a = ∑n2 2
P (0,tp ) i=1 τi
− 1, b = 1
S2 (0) 1− ∑n2 2
P (0,tp ) i=1 τi
. Assume E L2 [S1 (t10 )|S2 (t20 )] is a linear function
of S2 (t20 ):
( )
E L2 [S1 (t10 )|S2 (t20 )] − E L2 [S1 (t10 )] ≈ C S2 (t20 ) − E L2 [S2 (t20 )] = C[S2 (t20 ) − S2 (0)]
[ ] [ ]
By requiring E L2 E L2 [S1 (t10 )|S2 (t20 )] · S2 (t20 ) = E L2 S1 (t10 )S2 (t20 ) , we obtain
√
varL2 S1 (t10 )
C = corrL2 (S1 (t10 ), S2 (t20 ))
varL2 S2 (t20 )
Consequently, we have
[ ]
E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 )
[ ( [ ])]
= E L2 (1 + a + bS2 (t20 )) · 1{Rmin <S2 (t20 )<Rmax } E L2 [S1 (t10 )] + C S2 (t20 ) − S2 (0)
L2 (t)
To compute E L2 [S1 (t10 )], we apply the linear rate model L1 (t) = a1 + b1 S1 (t)
[ ]
L2 (t10 )/L2 (0)
E L2 [S1 (t10 )] = E L1 S (t
1 0
1
)
L1 (t10 )/L1 (0)
[ ]
= E L1 (a1 + b1 S1 (t10 ))S1 (t10 )
= a1 S1 (0) + b1 S12 (0) + b1 varL1 S1 (t10 )
∑n2 2 [ ∑n2 2 ]
L1 (0) i=1 τ L (0) τi
where a1 = ∑n1 i1
L2 (0) j=1 τj
and b1 = S11(0) 1 − L1(0) ∑ni=1
1 τ1 (see Appendix A). Define
2 j=1 j
[ ]
E0 = E Qtp 1{Rmin <S2 (t20 )<Rmax } ,
23
then
( [ ] )
V0 = P (0, tp ) leverage · E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 ) + spread · E0 + P (0, tp ) · fixRate
Define [ ]
Ee0 = E L2 1{R <S (t2 )<R }
[ min 2 0 max
]
e1 = E L2 1{R <S (t2 )<R } S2 (t2 )
E
[ min 2 0 max 0
]
E e2 = E L2 1{R <S (t2 )<R } S 2 (t2 )
min 2 0
[max 2 0
]
e0 cc = aEe0 + bE e1 = E L2 1{R <S (t2 )<R } (a + bS2 (t2 ))
E
[ min 2 0 max 0
]
e e e L2 2 2
E 1 cc = a E1 + bE 2 = E 1{Rmin <S2 (t0 )<Rmax } S2 (t0 )(a + bS2 (t0 ))
2
g
EA1 = E L1 [(a1 + b1 S1 (t10 ))S1 (t10 )] = E L2 [S1 (t10 )]
[ ]
We can write E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 ) as
[ ]
E Qtp 1{Rmin <S2 (t20 )<Rmax } S1 (t10 )
[ ( [ ])]
= E L2 (1 + a + bS2 (t20 )) · 1{Rmin <S2 (t20 )<Rmax } E L2 [S1 (t10 )] + C S2 (t20 ) − S2 (0)
[ ( [ ])]
= E L2 1{Rmin <S2 (t20 )<Rmax } E L2 [S1 (t10 )] + C S2 (t20 ) − S2 (0)
[ ]( )
+E L2 (a + bS2 (t20 )) · 1{Rmin <S2 (t20 )<Rmax } E L2 [S1 (t10 )] − C · S2 (0)
[ ]
+E L2 (a + bS2 (t20 )) · 1{Rmin <S2 (t20 )<Rmax } C · S2 (t20 )
= g 1 − C · S2 (0)] · E
[EA e0 + C · E
e1 + [EA
g 1 − C · S2 (0)] · E
e0 cc + C · E
e1 cc
= Ee + cc
e
e = [EA
where E g 1 − C · S2 (0)] · E
e0 + C · E
e1 and cc g 1 − C · S2 (0)] · E
e = [EA e0 cc + C · E
e1 cc. Therefore
[ ]
e + cc)
V0 = P (0, tp ) leverage · (E e + spread · E0 + P (0, tp ) · fixRate (19)
The following table shows how to compute the various quantities in the above derivation
Formula Black Model (S > 0 always)
1
σB Wt1 − 12 (σB
1 2 1
) t0
S1 (t10 ) = S1 (0)e 0 ,
2
σB Wt2 − 12 (σB
2 2 2
) t0
[ ] S2 (t20 ) = S2 (0)e 0
Qtp
E0 = E 1{Rmin <S2 (t20 )<Rmax } replication (Appendix C)
( Rmax ) ( Rmin )
ln S (0) √ ln S (0) √
e0 = E L2 [1{R <S (t2 )<R } ]
E Φ √2
+ 1 2
σ t 2 −Φ √
2
+ 1 2
σ t 2
min 2 0 max 2
σB 2 2 B 0 2 2 2 B 0
[ t( 0 )σB t( 0 )]
ln R max √ ln
Rmin √
e1 = E L2 [1{R <S (t2 )<R } S2 (t2 )]
E S2 (0) Φ √
S2 (0)
− 1 2
σ t2 −Φ √
S2 (0)
− 1 2
σ t 2
min 2 0 max 0 σB2 2 2 B 0 2 2 2 B 0
[ t0( Rmax )σB t0
ln √
e2 = E L2 [1{R <S (t2 )<R } S 2 (t2 )]
E
2 2 2
S22 (0)e(σB ) t0 Φ √ 2 − 32 σB
S2 (0) 2
t20
min 2 0 max 2 0 2
( Rmin σB
)]
t0
ln S (0) √
−Φ 2
√2
2
− 3 2
σ
2 B t2
0
σB t0
e0 cc = aE
E e0 + bE
e1
e1 cc = aE
E e1 + bE
e2
g
EA1 = E [(a1 + b1 S1 (t10 ))S1 (t10 )]
L1
a1 S1 (0) + b1 S12 (0)e(σB )
1 2 1
t0
e = [EA
E g 1 − C · S2 (0)] · E
e0 + C · E e1
g e
e = [EA1 − C · S2 (0)] · E0 cc + C · E
cc e1 cc
24
∫y ∫y
By noting Φ′′ (x) = −xΦ′ (x), it’s easy to derive −∞
xΦ′ (x)dx = −Φ′ (y) and −∞
x2 Φ′ (x)dx = Φ(y) −
′
yΦ (y). Therefore we have
Formula Normal Model (S could be negative)
S1 (t10 ) = S1 (0) + σN
1
Wt10 , S2 (t10 ) = S2 (0) + σN
2
Wt20
[ ]
E0 = E Qtp 1{Rmin <S2 (t20 )<Rmax } replication (Appendix C)
( ) ( )
e0 = E L2 [1{R <S (t2 )<R } ]
E Φ Rmax −S2 (0)
√2 −Φ Rmin −S2 (0)
√2
min 2 0 max σN2 2
[ t0( )σN t0( )]
√
e1 = E L2 [1{R <S (t2 )<R } S2 (t2 )]
E 2
σN t0 Φ2 ′ Rmin −S2 (0)
√2 −Φ ′ Rmax −S2 (0)
√2
min 2 0 max 0 2 2
[ ( σN t0 ) ( σN t0 )]
+S2 (0) Φ Rmax2 −S √ 22(0) − Φ Rmin2 −S √ 22(0)
σN t0 σ N t0
e2 = E ′ (Rmax −S2 (0))/(σN
2 2
t )
E L2
[1{Rmin <S2 (t20 )<Rmax } S22 (t20 )] 2 2 2
(σN ) t0[Φ (y) − yΦ (y)]|(Rmin −S2 (0))/(σ2 t20)
N 0
+2S2 (0)Ee1 − S 2 (0)E
e0
2
e0 cc = aE
E e0 + bE
e1
e1 cc = aE
E e1 + bE
e2
g
EA1 = E [(a1 + b1 S1 (t10 ))S1 (t10 )]
L1
a1 S1 (0) + b1 S12 (0) + b1 (σN
1 2 1
) t0
e = [EA
E g 1 − C · S2 (0)] · E
e0 + C · E e1
g 1 − C · S2 (0)] · E
e = [EA
cc e0 cc + C · E e1 cc
where σatm is the at-the-money volatility of the caplet for the period [S, T ] and
{ 2
}
2σatm S
[a + b(S)F (0; S, T )][a + b(S)F (0; S, T )e ]
(σ ∗ )2 = (σatm )2 + ln 2 /S
[a + b(S)F (0; S, T )eσatm S ]2
{ 2
}
2 [1 + τ (S, T )F (0; S, T )][1 + τ (S, T )F (0; S, T )e2σatm S ]
= (σatm ) + ln 2 /S
[1 + τ (S, T )F (0; S, T )eσatm S ]2
The case of “adj Fwd & Vol” is equivalent to assuming L(S, T ) is lognormal under QS , while the case of
“adj Fwd” is kind of halfway from “lognormal under QT ” to “lognormal under QS ”.
Numerical experiments show results by these formulas are quite close and the two naive approaches
(adjusted forward rates without and with additional adjusted volatilities) yield results whose difference to
the correct one from replication are negligible in practice.
25
adjusted volatility (formula (10))
For price in terms of adjusted forward swap rate and adjusted volatility, the adjusted volatility is given by
[ ]
[a + b(T ′ )Sα,β (0)] a + b(T ′ )Sα,β (0)e2(σatm ) Tα
2
(σ ∗ )2 = (σatm )2 + ln /Tα .
[a + b(T ′ )Sα,β (0)eσatm Tα ]2
2
Again, price in terms of adjusted forward rate and adjusted volatility is obtained by assuming the
underlying rate is lognormal under QT ′ and applying Black formula.
The naive approaches, although taking into account the smile by taking the caplet volatility from the
swaption smile, show a significant mispricing relative to the correct valuations based on replication. Com-
paring the results of the two replication formulas, it turns out that the replication based on the idea of
cash-settled swaptions consistently leads to slightly higher CMS caplet prices.
26
5.1 Multi-currency change of numeraire theorem
Suppose we have a domestic economy d and a foreign economy f , together with the exchange rate X that
expresses the value of one unit foreign currency in terms of domestic currency. Let N d denote a numeraire
process with associated martingale measure Qd for the domestic economy, and N f a numeraire process with
associated martingale measure Qf for the foreign economy.
The domestic value today of a domestic payoff ZTd to be paid in foreign units and at time T is by the
general theory [ ]
f Qf ZTd
X0 N0 E .
NTf
On the other hand, the same payoff translated back into domestic currency with the exchange rate at time
T should trade at the same domestic price, therefore
[ ] [ d ]
f Qf ZTd d Qd ZT XT
X0 N0 E = N0 E .
NTf NTd
P d (S, T ′ )
= a + b(T ′ )YSd , T ′ ≥ S.
NSd
XS P f (S, T ′ ) fx
= X0∗ eσf x WS − 2 σf x S
1 2
d
P (S, T ) ′
[ ]
f
(S,T ′ )
where W f x is a standard Brownian motion under Qd and X0∗ = E Q XPS Pd (S,T
d
′) . Here σf x is identified
with the implied volatility of a foreign exchange rate option with maturity S, which is a crucial simplification
as long as the payment date T ′ is not close to S.
27
To calculate X0∗ , we note
[ ] [ ]
X0 P f (0, T ′ ) f ′
Qd XS P (S, T )
f ′ d
Qd XS P (S, T ) P (S, T )
′
=E =E = X0∗ [a + b(T ′ )Y0d eρσf x σY S ],
N0d NSd P d (S, T ′ ) NSd
√ √ ]
b(T ′ )(Y0d )2 e(σY +2ρσf x σY )S Φ(d1 + (ρσf x + σY ) S) − aKΦ(d2 + ρσf x S)
2
with ( ) ( )
Y0d Y0d
ln K + 12 σY2 S ln K − 21 σY2 S
d1 = √ , d2 = √ .
σY S σY S
The corresponding put formula is
f
P f (0, T ′ )E QT ′ [(YSd − K)+ ]
P f (0, T ′ ) [ √
= −Y 0
d ρσf x σY S
e Φ(−d 1 − ρσf x S)(a − b(T ′ )K)−
a + b(T ′ )Y0 eρσf x σY S
d
√ √ ]
b(T ′ )(Y0d )2 e(σY +2ρσf x σY )S Φ(−d1 − (ρσf x + σY ) S) + aKΦ(−d2 − ρσf x S)
2
Remark 8. In the special case of Y d being the domestic LIBOR rate Ld (S, T ) for the period of [S, T ] and
T ′ = T , we have N d (·) = P d (·, T ) and a = 1, b ≡ 0 in Assumption 2. The price of a call option on Ld (S, T )
paid at time T in foreign currency units is therefore
f √ √
P f (0, T )E QT [(Ld (S, T ) − K)+ ] = P f (0, T )[Y0d eρσf x σY S Φ(d1 + ρσf x S) − KΦ(d2 + ρσf x S)]
= P f (0, T )[Ye d Φ(de1 ) − KΦ(de2 )] 0
where ( ed ) ( ed )
Y0 Y
+ 12 σY2 S
ln Kln K0 − 21 σY2 S
d1 = √ , d2 = √ .
σY S σY S
This is the market standard method of valuing options on convexity corrected rates: apply the Black formula
using the convexity corrected rate as the forward rate.
28
A Approximation of ratio of annuities under linear rate model
In pricing floating rate CMS range accrual, we need to approximate the ratio of two annuities by swap
rate. More formally, suppose t10 < t11 < · · · < t1n1 and t20 < t21 < · · · < t2n2 are two schedules with their
respective day counting conventions {τ11 , τ21 , · · · , τn11 } and {τ12 , τ22 , · · · , τn22 }. For k = 1, 2, annuity Lk (t) is
defined as
∑nk
Lk (t) = τik P (t, tki ), t ≤ tk0
i=1
L2 (t) ∑
n2
τ 2 P (t, t2 ) ∑
n2
[ ]
i i
= = τi2 a + b(t2i )S1 (t)
L1 (t) i=1
L1 (t) i=1
∑n2 2 ∑ n2
[ ]
τi 1 P (0, t2i ) 1
= ∑n1 1 +
i=1
τi ·
2
− ∑n1 1 · S1 (t)
j=1 τj i=1
S1 (0) L1 (0) j=1 τj
Proof. Clearly, we may restrict ourselves to the case where [a, b] ⊂ intI. If a = t0 < t1 < · · · < tn = b is a
partition of [a, b], then
′ ′ f (tk ) − f (tk−1 ) ′ ′
f− (tk−1 ) ≤ f+ (tk−1 ) ≤ ≤ f− (tk ) ≤ f+ (tk )
tk − tk−1
29
∑n
for every k. As f (b) − f (a) = k=1 [f (tk ) − f (tk−1 )], a moment’s reflection shows that
∫ b ∫ b
′ ′
f (b) − f (a) = f− (t)dt = f+ (t)dt.
a a
′ ′
As f− ≤φ≤ f+ , this forces the equality in the statement.
A property very useful when joint with Proposition B.1 is the following
Proposition B.2. If f : I → R is absolutely continuous and its derivative φ is of finite variation locally,
then f can be written as the difference of two convex functions.
Proof. By Fundamental Theorem of Calculus for absolutely continuous functions, we have
∫ b
f (b) − f (a) = φ(t)dt.
a
By properties of functions of finite variation, φ can be written as the difference of two monotone increasing
functions: φ = φ+ − φ− . Then
∫ x ∫ x
f (x) = f (a) + φ+ (t)dt − φ− (t)dt.
a a
V0 = leverage · [V0caplet (Rmin ) − V0caplet (Rmax )] + V0digital_call (Rmin )(Rmin · leverage + spread)
−V0digital_call (Rmax )(Rmax · leverage + spread)
∫∞ ∫ K0
D Computation of K0 C(K)dK and 0 P (K)dK
We recall C(K) = F Φ(d1/2 (K)) − KΦ(d−1/2 (K)) and P (K) = −F Φ(−d1/2 (K)) + KΦ(−d−1/2 (K)),
√
where dλ (K) = ln(F
√
σ T
/K)
+ λσ T . We shall verify that
∫ ∞
1 2 σ2 T 1
C(x)dx = F e Φ(d3/2 (K)) − F KΦ(d1/2 (K)) + K 2 Φ(d−1/2 (K))
K 2 2
and
∫ K
1 2 σ2 T 1
P (x)dx = F e Φ(−d3/2 (K)) − F KΦ(−d1/2 (K)) + K 2 Φ(−d−1/2 (K))
0 2 2
30
So by integration-by-parts formula, we have
∫ ∞ ∫ ∞ ∫ ∞
∞ d
C(x)dx = xC(x)|K − x C(x)dx = −KC(K) + xΦ(d−1/2 (x))dx
K K dx K
∫ ∞ 2 ( )
x2 ∞ x ′ 1
= −KC(K) + Φ(d−1/2 (x)) − Φ (d−1/2 (x)) − √ dx
2 K K 2 xσ T
∫ ∞
K2 1
= −KC(K) − Φ(d−1/2 (K)) + √ xΦ′ (d−1/2 (x))dx
2 2σ T K
∫ ∞
K2 1
= −KF Φ(d1/2 (K)) + Φ(d−1/2 (K)) + √ xΦ′ (d−1/2 (x))dx
2 2σ T K
By the change-of-variable y = ln Fx , we have
( √ )2
∫ ∞ ∫ ∞ − 12 −y
√ − 12 σ T √
e σ T
xΦ′ (d−1/2 (x))dx =
2
F 2 e2y √ dy = σ T · F 2 eσ T Φ(d3/2 (K))
K ln K
F
2π
Combined, we have obtained
∫ ∞
1 2 1
C(x)dx = F 2 eσ T Φ(d3/2 (K)) − F KΦ(d1/2 (K)) + K 2 Φ(d−1/2 (K))
K 2 2
By the put-call parity C(K) − P (K) = F − K and
∫ ∞ ∫ ∞
1 2
C(x)dx = lim C(x)dx = F 2 eσ T ,
0 K↓0 K 2
we conclude
∫ K ∫ K ∫ ∞ ∫ ∞
K2
P (x)dx = [C(x) − F + x]dx = C(x)dx − C(x)dx − F K +
0 0 0 K 2
1 2 σ2 T 1 1 2 1
= F e − F K + K 2 − F 2 eσ T Φ(d3/2 (K)) + F KΦ(d1/2 (K)) − K 2 Φ(d−1/2 (K))
2 2 2 2
1 2 σ2 T 1 2
= F e Φ(−d3/2 (K)) − F KΦ(−d1/2 (K)) + K Φ(−d−1/2 (K))
2 2
References
[1] W. Boenkost and W. Schmidt. Notes on convexity and quanto adjustments for interest rates and related
options. Working paper. October 2003. 3, 10
[2] W. Boenkost and W. Schmidt. Interest rate convexity and the volatility smile. Working paper. May
2006. 3
[3] P. Hagan. Convexity conundrums: Pricing CMS swaps, cpas, and floors. Wilmott Magazine, March,
p.38-44. 3, 7
[4] P. J. Hunt and J. E. Kennedy. Financial derivatives in theory and practice. Revised Edition, Wiley,
2004. 3
[5] J. Hull. Options, futures, and other derivatives. Fourth Edition. Prentice-Hall, 2000. 3, 9
[6] A. Lesniewski. Convexity. 3
[7] Constantin P. Niculescu and Lars-Erik Persson. Convex functions and their applications: A contempo-
rary approach. Springer. 29
[8] A. Pelsser. Mathematical foundation of convexity correction. Wroking paper. April 8, 2001. 3
[9] M. Piza. Convexity correction. January 18, 2007. 3
31