framework for pricing derivatives
DESCRIPTION
Framework for pricing derivatives. Zvi Wiener 02-588-3049 http://pluto.mscc.huji.ac.il/~mswiener/zvi.html. The Market Price of Risk. Observable underlying process, for example stock, interest rate, price of a commodity, etc. Here dz is a Brownian motion. We assume that m(x, t) and s(x, t). - PowerPoint PPT PresentationTRANSCRIPT
Ch. 19J. Hull, Options, Futures and
Other Derivatives
Zvi Wiener
02-588-3049http://pluto.mscc.huji.ac.il/~mswiener/zvi.html
Framework for pricing derivatives
Zvi Wiener Hull - 19 slide 2
The Market Price of Risk
Observable underlying process, for example stock, interest rate, price of a commodity, etc.
sdzmdtx
dx
Here dz is a Brownian motion.
We assume that m(x, t) and s(x, t).
x is not necessarily an investment asset!
Zvi Wiener Hull - 19 slide 3
dzdtf
df
dzdtf
df
222
2
111
1
Suppose that f1 and f2 are prices of two
derivatives dependent only on x and t. For example options. We assume that prior to maturity f1 and f2 do not provide any cashflow.
The same!
Zvi Wiener Hull - 19 slide 4
zftff
zftff
22222
11111
Form an instantaneously riskless portfolio consisting of 2f2 units of the first derivative
and 1f1 units of the second derivative.
211122 ffff
Zvi Wiener Hull - 19 slide 5
An instantaneously riskless portfolio must earn a riskless interest rate.
211122 ffff
tffff 12222121
tr
211122 ffff
Zvi Wiener Hull - 19 slide 6
This implies
121221 rr
or
2
2
1
1 rr
is called the market price of risk of x.
Zvi Wiener Hull - 19 slide 7
If x is a traded asset we must also have
But if x is not a financial asset this is not true.
For all financial assets depending on x and time a similar relation must held.
s
rm
Zvi Wiener Hull - 19 slide 8
Volatility
Note that can be positive or negative, depending on the correlation with x.
| | is called the volatility of f.
If s>0 and f and x are positively correlated, then >0, otherwise it is negative.
Zvi Wiener Hull - 19 slide 9
Example 19.1
Consider a derivative whose price is positively related to the price of oil. Suppose that it provides an expected annual return of 12%, and has volatility of 20%. Assume that r=8%, then the market price of risk of oil is:
2.02.0
08.012.0
Zvi Wiener Hull - 19 slide 10
Example 19.2
Consider two securities positively dependent
on the 90-day IR. Suppose that the first one
has an expected return of 3% and volatility of
20% (annual), and the second has volatility of
30%, assume r=6%. What is the market price
of interest rate risk? What is the expected
return from the second security?
Zvi Wiener Hull - 19 slide 11
Example 19.2
15.02.0
06.003.0
015.03.015.006.0
The market price of IR risk is:
The expected return from the second security:
Zvi Wiener Hull - 19 slide 12
Differential Equation
Finally leading to:
2
222
2
1
x
fxs
t
f
x
fmxf
x
fsxf
rfx
fxs
x
fxsm
t
f
2
222
2
1)(
f is a function of x and t, we can get using Ito’s lemma:
Zvi Wiener Hull - 19 slide 13
Differential Equation
rfx
fxs
x
fxsm
t
f
2
222
2
1)(
Comparing to BMS equation we see that it is similar to an asset providing a continuous dividend yield q=r-m+s.
Using Feynman-Kac we can say that the expected growth rate is r-q and then discount the expected payoff at the risk-free rate r.
Zvi Wiener Hull - 19 slide 14
Risk-neutral approach
True dynamics sdzmdtx
dx
Risk-neutral dynamics
*)( sdzdtsmx
dx
Zvi Wiener Hull - 19 slide 15
Example 19.3
Price of copper is 80 cents/pound. Risk free r=5%. The expected growth rate in the price of copper is 2% and its volatility is 20%. The market price of risk associated with copper is 0.5. Assume that a contract is traded that allows the holder to receive 1,000 pounds of copper at no cost in 6 months. What is the price of the contract?
Zvi Wiener Hull - 19 slide 16
Example 19.3
m=0.02, =0.5, s=0.2, r=0.05;
the risk-neutral expected growth rate is
08.02.05.002.0 sm
The expected (r-n) payoff from the contract is
63.76880.0000,1 5.008.0 e
Discounting for six months at 5% we get
65.74963.768 5.005.0 e
Zvi Wiener Hull - 19 slide 17
Derivatives dependent on several state variables
State variables (risk factors):
iiii
i dzsdtmx
dx
Traded security
n
iiidzdt
f
df
1
Zvi Wiener Hull - 19 slide 18
Multidimensional Risk
Here i is the market price of risk for xi.
This relation is also derived in APT (arbitrage pricing theory), Ross 1976 JET.
n
iiir
1
Zvi Wiener Hull - 19 slide 19
Pricing of derivatives
To price a derivative in the case of several risk factors we should
• change the dynamics of xi to risk neutral
• derive the expected (r-n) discounted payoff
TrT feEprice ˆ
TrT fEeprice ˆIf r is deterministic
Zvi Wiener Hull - 19 slide 20
Example 19.5
Consider a security that pays off $100 at time T if stock A is above XA and stock B is above
XB. Assume that stocks A and B are
uncorrelated.
The payoff is $100 QA QB, here QA are QB are
r-n probabilities of stocks to be above strikes.
Zvi Wiener Hull - 19 slide 21
Example 19.5
T
TrXSNQ
T
TrXSNQ
B
BBBB
A
AAAA
2ln
2ln
2
2
Zvi Wiener Hull - 19 slide 22
Derivatives on Commodities
The big problem is to estimate the market price of risk for non investment assets. One can use futures contracts for this.
Assume that the commodity price follows (no mean reversion and constant volatility)
dzdttS
dS )(
Zvi Wiener Hull - 19 slide 23
Derivatives on Commodities
The expected (r-n) future price of a commodity is its future price F(t).
)(ln)(
)(
tFt
t
dzdttS
dS
Zvi Wiener Hull - 19 slide 24
Example 19.6Futures pricesAugust 99 62.20Oct 99 60.60Dec 99 62.70Feb 00 63.37Apr 00 64.42Jun 00 64.40
The expected (r-n) growth rate between Oct
and Dec 99 is ln(62.70/60.60)=3.4%, or
20.4% annually.
Zvi Wiener Hull - 19 slide 25
Convenience Yield
y - convenience yield
u - storage costs, then then r-n growth rate is
m - s = r - y + u
Zvi Wiener Hull - 19 slide 26
Martingales and Measures
A martingale is a zero drift stochastic process
for example: dx = s dz
an important property E[xT] = x0,
fair game.
Zvi Wiener Hull - 19 slide 27
Martingales and Measures
Real world fdzfdtdf
Risk-neutral world *fdzrfdtdf In the risk-neutral world the market price
of risk is zero, while in the real world it is
r
Zvi Wiener Hull - 19 slide 28
Martingales and Measures
By making other assumptions we can define other “worlds” that are internally consistent. In a world with the market price of risk * the drift (expected growth rate) * must be
zfdfdtrdf ~)*(
**
r
Zvi Wiener Hull - 19 slide 29
Equivalent Martingale Measures
Suppose that f and g are price processes of
two traded securities dependent on a single
source of uncertainty. Define x=f/g.
This is the relative price of f with respect to g.
g is the numeraire.
Zvi Wiener Hull - 19 slide 30
Equivalent Martingale Measures
The equivalent martingale measure result states that when there are no arbitrage opportunities, x is a martingale for some choice of market price of risk.
For a given numeraire g the same market price of risk works for all securities f and the market price of risk is equal to the volatility of g.
Zvi Wiener Hull - 19 slide 31
Equivalent Martingale Measures
fdzfdtrdf ffg )(
gdzgdtrdg gfg )( 2
Using Ito’s lemma
dzdtrfd fffg )2/(ln 2
dzdtrgd gg )2/(ln 2
Zvi Wiener Hull - 19 slide 32
Equivalent Martingale Measures
dzdtg
fd gfgffg )(25.0ln 22
dzg
f
g
fd gf )(
A martingale
Zvi Wiener Hull - 19 slide 33
Forward risk neutral wrt g
T
Tg g
fE
g
f
0
0
Since f/g is a martingale
T
Tg g
fEgf 00 (19.19) in Hull
Zvi Wiener Hull - 19 slide 34
Money market as a numeraire
Money market account dg = rgdt
zero volatility, so the market price of risk will be zero and we arrive at the standard r-n world. g0=1 and
T
Tg g
fEgf 00
Trdt
T eg 0
T
rdt
Tg efEf 00
Zvi Wiener Hull - 19 slide 35
Zero-Coupon Bond as a Numeraire
Define P(t,T) the price at time t of a zero-coupon bond maturing at T. Denote by ET the
appropriate measure.
gT = P(T,T)=1, g0 = P(0,T) we get
TT fETPf ),0(0
Zvi Wiener Hull - 19 slide 36
Zero-Coupon Bond as a Numeraire
Define F as the forward price of f for a contract maturing at time T. Then
TTT fETP
fF
),0(
In a world that is forward risk neutral with respect to P(t,T) the forward price is the expected future spot price.
Zvi Wiener Hull - 19 slide 37
Important Conclusion
We can value any security that provides a payoff at time T by calculating its expected payoff in a world that is forward risk neutral with respect to a bond maturing at time T and discounting at the risk-free rate for maturity T.
In this world it is correct to assume that the expected value of an asset equals its forward value.
Zvi Wiener Hull - 19 slide 38
Interest Rates With a Numeraire
Define R(t, T1, T2) as the forward interest rate as
seen at time t for the period between T1 and T2
expressed with a compounding period T1- T2.
The forward price of a zero coupon bond lasting between T1 and T2 is
),(
),(
1
2
TtP
TtP
Zvi Wiener Hull - 19 slide 39
Interest Rates With a Numeraire
),(
),(
),,()(1
1
1
2
2112 TtP
TtP
TTtRTT
A forward interest rate implied forthe corresponding period is
),(
),(),(1),,(
2
21
1221 TtP
TtPTtP
TTTTtR
Zvi Wiener Hull - 19 slide 40
Interest Rates With a Numeraire
Setting
),(,),(),(
212
21 TtPgTT
TtPTtPf
We get that R(t, T1, T2) is a martingale in
a world that is forward risk neutral with
respect to P(t,T2).
),,(),,( 211221 TTTRETTtR
Zvi Wiener Hull - 19 slide 41
Annuity Factor as a Numeraire
Consider a swap starting at time Tn with payment
dates Tn+1, Tn+2, …, TN+1. Principal $1. Denote the
forward swap rate Sn,N(t). The value of the fixed
side of the swap is
N
niiiiNn
NnNn
TtPTTtA
tAtS
),()(
)()(
11,
,,
Zvi Wiener Hull - 19 slide 42
Annuity Factor as a Numeraire
The value of the floating side is ),(),( 1 Nn TtPTtP
The first term is $1 received at the next payment date and the second term corresponds to the principal payment at the end.
The swap rate can be found as),(),()()( 1,, NnNnNn TtPTtPtAtS
)(
),(),()(
,
1, tA
TtPTtPtS
Nn
NnNn
Zvi Wiener Hull - 19 slide 43
Annuity Factor as a Numeraire
We can apply an equivalent martingale measure by setting P(t,Tn)-P(t,TN+1) as f and
An,N(t) as g. This leads to
)()( ,, TSEtS NnANn For any security f we have
)()(
,,0 TA
fEtAf
Nn
TANn
Zvi Wiener Hull - 19 slide 44
Multiple Risk Factors
n
iiig
n
iiif
dzrgdtdg
dzrfdtdf
1,
1,
Zvi Wiener Hull - 19 slide 45
Multiple Risk Factors
n
iiig
n
iigi
n
iiif
n
iifi
dzgdtrdg
dzfdtrdf
1,
1,
*
1,
1,
*
Equivalent world can be defined as
Where i* are the market prices of risk
Zvi Wiener Hull - 19 slide 46
Multiple Risk Factors
Define a world that is forward risk neutral with respect to g as a world where i*=g,i. It
can be shown from Ito’s lemma, using the fact that dzi are uncorrelated, that the process
followed by f/g in this world has zero drift.
Zvi Wiener Hull - 19 slide 47
An Option to Exchange Assets
Consider an option to exchange an asset worth U to an asset worth V. Assume that the correlation between assets is and they provide no income. Setting g=U, fT=max(VT-UT,0) in
19.19 we get.
T
TTU U
UVEUf
)0,max(00
Zvi Wiener Hull - 19 slide 48
An Option to Exchange Assets
0,1max00
T
TU U
VEUf
VUVU 2222
The volatility of V/U is
This is a simple option.
Zvi Wiener Hull - 19 slide 49
An Option to Exchange Assets
Tdd
T
TqqU
V
d
dNUdNVf
VU
12
2
0
0
1
20100
2ln
)()(
VUVU 2222
Assuming that the assets provide anincome at rates qU and qV.
Zvi Wiener Hull - 19 slide 50
Change of Numeraire
Dynamics of asset f with forward risk neutral measure wrt g and h we have
n
iiif
n
iifih
n
iiif
n
iifig
dzfdtrdf
dzfdtrdf
1,
1,,
1,
1,,
When changing numeraire from g to h we update drifts by
n
iifigih
1,,,
Zvi Wiener Hull - 19 slide 51
Change of Numeraire
Set v be a function of traded securities. Define v,i as the i-th component of v volatility. The
rate of growth of v responds to a change of numeraire in the same way. Define q=h/g, then h,i- g,i is the i-th component of
volatility of q. Thus the drift update of v is
qvv
Zvi Wiener Hull - 19 slide 52
Quantos
Quanto provides a payoff in currency X at time T. We assume that the payoff depends on the value of a variable V observed in currency Y at time T.
F(t) - forward value of V in currency Y
PX(t,T) - value (in X) of 1 unit of X paid at T
PY(t,T) - value (in X) of 1 unit of Y paid at T
G(T) forward exchange rate units of Y per X
G(t)= PX(t,T)/PY(t,T)
Zvi Wiener Hull - 19 slide 53
QuantosIn equation 19.19 set g=PY(t,T), f - be a security
that pays VT units of currency Y at time T.
fT=VT/ST, and gT=1/ST
f0= PY(t,T)EY(VT), no arbitrage means
F(0)=f0/PY(0,T), hence
EY(VT)=F(0)
When we move from X world to Y world the expected growth rate increases by FG
Zvi Wiener Hull - 19 slide 54
Quantos
This means that approximately
TYX
GFeTFETFE )]([)]([
TTX
GFeFVE )0()(
Since V(T)=F(T) and EY(VT)=F(0)
Zvi Wiener Hull - 19 slide 55
Example 19.7
Nikkei = 15,000
yen dividend yield = 1%
one-year USD risk free rate = 5%
one-year JPY risk free rate = 2%
The forward price of Nikkei for a contract denominated in yen is
15,000e(0.02-0.01)1=15,150.75
Zvi Wiener Hull - 19 slide 56
Example 19.7
Suppose that the volatility of the one-year forward price of the index is 20%, the volatility of the one-year forward yen/USD exchange rate is 12%. The correlation of one-year forward Nikkei with the one-year forward exchange rate =0.3.
The forward price of the Nikkei for a contract that provides a payoff in dollars is
15,150.75 e0.3*0.2*0.12=15,260.2
Zvi Wiener Hull - 19 slide 57
Siegel’s paradoxConsider two currencies X and Y. Define S an exchange rate (the number of units of currency Y for a unit of X).
The risk-neutral process for S is
tSXY SdBSdtrrdS )(
By Ito’s lemma the process for 1/S is
tSSXY dBS
dtS
rrS
d111 2
Zvi Wiener Hull - 19 slide 58
Siegel’s paradox
The paradox is that the expected growth rate of 1/S is not ry - rX, but has a correction term.
If we change numeraire from currency X to currency Y the correction term is 2=-2.
The process in terms of currency Y becomes
tSXY dBS
dtS
rrS
d111