amf 241 interest rate models

54
Bonds and Interest Rates Bonds and Interest Rates Timothy Robin Teng Bonds and Interest Rates  1 / 54

Upload: riemann-soliven

Post on 01-Jun-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 1/54

Bonds and Interest Rates

Bonds and Interest Rates

Timothy Robin Teng

Bonds and Interest Rates   1 / 54

Page 2: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 2/54

Page 3: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 3/54

Bonds and Interest Rates

Zero coupon bonds and related interest rates

DefinitionA zero coupon bond  (or pure discount bond) with maturity date  T ,also called a  T -bond, is a contract which guarantees its holder the

payment of one unit of currency at time  T , with no intermediatepayments. The contract value at time  t  < T   is denoted by  P (t , T ).

The convention that the payment at the maturity date, known as the

principal value or face value, equals one is made for computationalconvenience.Note that  P (t , t ) = 1 for all  t .

Bonds and Interest Rates   3 / 54

Page 4: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 4/54

Bonds and Interest Rates

The bond price  P (t , T ) is also a stochastic object with two variablest  and  T , and for each outcome  ω   in the underlying sample space, thedependence upon these variables is different.

Fix  t ;  P (t , T ) is a function of maturity date  T 

The function provides prices for bonds of all possible maturitiesat a fixed time  t . The graph of the function is called “bondprice curve at  t ”, or “the term structure of  t ”. Typically it willbe a very smooth graph, i.e. for each  t ,  P (t , T ) is differentiablewith respect to  T .

Bonds and Interest Rates   4 / 54

B d d I R

Page 5: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 5/54

Bonds and Interest Rates

Fix  T ;  P  (t , T ) (as a function  t ) will be scalar stochastic processThe process gives the prices, at different times, of the bond with

fixed maturity  T , and the trajectory will typically be veryirregular.

Bonds and Interest Rates   5 / 54

B d d I t t R t

Page 6: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 6/54

Bonds and Interest Rates

Interest Rates

We may now define a number of interest rates based on the zerocoupon bond, and the basic construction is as follows:

We consider three time instants, namely the time  t  at which the rateis considered, and two other points in time  T   and  S , such thatt  < S  < T . The goal is to write a contract at time  t  which allows usto make an investment of one (dollar) at time  S , and to have adeterministic rate of return (determined at the contract time  t ) overthe interval [S , T ].

Bonds and Interest Rates   6 / 54

Bonds and Interest Rates

Page 7: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 7/54

Bonds and Interest Rates

At time  t  we sell one  S -bond. This will give us  P (t , S ) dollars.

We use this income to buy exactly  P (t , S )/P (t , T )  T -bonds.Thus our net investment at time  t  equals zero.

At time  S , the  S -bond matures, so we are obliged to pay outone dollar.

Bonds and Interest Rates   7 / 54

Bonds and Interest Rates

Page 8: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 8/54

Bonds and Interest Rates

At time  T , the  T -bonds mature at one dollar a piece, so we willreceive the amount  P (t , S )/P (t , T ) dollars.

The net effect of all this is that, based on a contract at  t , aninvestment of one dollar at time  S  has yielded  P (t , S )/P (t , T )

dollars at  T .

Thus, at time  t , we have made a contract guaranteeing ariskless rate of interest over the future interval [S , T ]. Such aninterest rate is called a forward rate.

Bonds and Interest Rates   8 / 54

Bonds and Interest Rates

Page 9: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 9/54

Bonds and Interest Rates

We now proceed to compute the relevant interest rates implied bythe construction above. The simple forward rate (or LIBOR rate)  L,is the solution to equation

1 + (T  − S )L =  P (t , S )

P (t , T )

whereas the continuously compounded forward rate  R   is the solutionto the equation

e R (T −S ) =  P (t , S )

P (t , T )

The simple rate notation is the one used in the market, whereas thecontinuously compounded notation is used in theoretical contexts.

Bonds and Interest Rates   9 / 54

Bonds and Interest Rates

Page 10: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 10/54

Bonds and Interest Rates

The  simple forward rate  for [S , T ] contracted at  t ,  henceforthreferred to as the  LIBOR forward rate, is defined as

L(t ; S , T ) = −   1

T  − S 

P (t , T ) − P (t , S )

P (t , T )

The  simple spot rate  for [S , T ],  henceforth referred to as theLIBOR spot rate, is defined as

L(S , T ) := L(S ; S , T ) =

−  1

T  − S P (S , T ) − 1

P (S , T )

Bonds and Interest Rates   10 / 54

Bonds and Interest Rates

Page 11: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 11/54

The  continuously compounded forward rate  for [S , T ]contracted at  t   is defined as

R (t ; S , T ) = − log P (t , T ) − log P (t , S )

T  − S 

The  continuously compounded spot forward rate,  R (S , T ),for the period [S , T ] is defined as

R (S , T ) := R (S ; S , T ) =

−log P (S , T )

T  − S 

Bonds and Interest Rates   11 / 54

Bonds and Interest Rates

Page 12: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 12/54

The   instantaneous forward rate  with maturity  T , contractedat  t , is defined by

f   (t , T ) =

  ∂ 

∂ T 

  log P (t , T )

The  instantaneous short rate  at time  T   is defined by

r t  = f   (t , t )

Bonds and Interest Rates   12 / 54

Page 13: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 13/54

Bonds and Interest Rates

Page 14: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 14/54

The short rate of interest  r t , is the (annualized) interest rate at whichan entity can lend or borrow money over an infinitesimally smallinterval [t , t  + dt ]. As opposed to stock prices, short rates tend tostay within a certain range, hence they are often described as mean

reverting processes. From the short rate  r t , we can define animportant instrument in the market which is the money marketaccount, representing a (locally) riskless investment where profit isaccrued continuously at rate  r t .

Bonds and Interest Rates   14 / 54

Bonds and Interest Rates

Page 15: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 15/54

Remark: One should note that in practice, the process  r t   is notobservable. The shortest maturity rate available is the overnight rate,which is conceptually quite different from an instantaneous spot rate.

Nevertheless, it is quite frequent that one-night and even one-monthor three-month rates are used as proxies for  r t  in the empirical termstructure literature

Bonds and Interest Rates   15 / 54

Bonds and Interest Rates

Page 16: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 16/54

Definition(Money-market account). We define  B t  to be the value of a bankaccount at time  t  ∈ R+, and its given by

B t  = e   t 

0  r (u )du 

that is, it evolves according to the following initial value problem:

dB t  = r t B t dt ,   B 0  = 1

Bonds and Interest Rates   16 / 54

Bonds and Interest Rates

Page 17: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 17/54

As an immediate consequence of the definitions we have the followinguseful formula.

LemmaThe price of the bond can be expressed as 

P (t , T ) = exp−    T 

t f   (t , u )du 

and for  t  ≤ s  ≤ T , we have 

P (t , T ) = P (t , s ) · exp−   T 

f   (t , u )du 

Bonds and Interest Rates   17 / 54

Bonds and Interest Rates

Page 18: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 18/54

Relations between  df   (t , T ), dP (t , T ) and  dr (t )

We will consider dynamics of the following form:Short rate dynamics

dr (t ) =  a(t )dt  + b (t )dW t    (1)

Bond price dynamics

dP (t ,T ) =  P (t ,T )m(t ,T )dt  + P (t ,T )v (t ,T )dW t    (2)

Forward rate dynamics

df   (t ,T ) = α(t ,T )dt  + σ(t ,T )dW t    (3)

The processes a

(t 

) and b 

(t 

) are scalar adapted processes, whereasm(t ,T ), v (t ,T ), α(t ,T ) and  σ(t ,T ) are adapted processes paramaterized bytime of maturity  T . The interpretation of the bond price equation (2) and theforward rate equation (3) is that these are scalar stochastic differential equations(in the   t -variable) for each fixed time of maturity  T .

Bonds and Interest Rates   18 / 54

Bonds and Interest Rates

Page 19: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 19/54

We will study the formal relations which must hold between bondprices and interest rates, and for this we establish the following

assumptions:1 For each fixed  ω, t  all the objects  m(t , T ),  v (t , T ), α(t , T ) and

σ(t , T ) are assumed to be continuously differentiable in theT -variable. This partial  T -derivative is sometimes denoted by

mT (t , T ) etc.2 All processes are such that we can differentiate under the

integral sign as well as interchange the order of integration.

The main result is as follows. Note that the results below hold,regardless of the measure under consideration, and in particular, wedo not assume that markets are free of arbitrage.

Bonds and Interest Rates   19 / 54

Bonds and Interest Rates

Page 20: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 20/54

Proposition.

1 If  P (t , T )  satisfies 

dP (t , T ) = P (t , T )m(t , T )dt  + P (t , T )v (t , T )dW t 

the the forward rate dynamics have 

df   (t , T ) = α(t , T )dt  + σ(t , T )dW t 

where  α  and  σ  are given by   α(t , T ) = v T (t , T ) · v (t , T ) − mT (t , T )σ(t , T ) = −v T (t , T )

Bonds and Interest Rates   20 / 54

Bonds and Interest Rates

Page 21: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 21/54

Proposition

2.   If   f   (t , T )  satisfies 

df   (t , T ) = α(t , T )dt  + σ(t , T )dW t 

then the short rate satisfies 

dr (t ) = a(t )dt  + b (t )dW t 

where      a(t ) = f  T (t , t ) + α(t , t )b (t ) = σ(t , t )

Bonds and Interest Rates   21 / 54

Bonds and Interest Rates

Page 22: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 22/54

Proposition

3.   If    f   (t ,T )  satisfies 

df   (t ,T ) = α(t ,T )dt  + σ(t ,T )dW t 

then  P (t ,T )  satisfies 

dP (t ,T ) =  P (t ,T )

r (t ) +  A(t ,T ) +

 12S 

2(t ,T )

dt +P (t ,T )S (t ,T )dW t 

where 

    A(t ,T ) = −  T 

t   α(t , s )ds 

S (t ,T ) = −  T t    σ(t , s )ds 

Bonds and Interest Rates   22 / 54

Bonds and Interest Rates

Page 23: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 23/54

Short rate models

Interest models behave differently from stock prices and require thedevelopment of specific models to account for properties such aspositivity, boundedness and return to equilibrium. Refer to the other

handouts for a list of short rate models

For this section, we consider classical time homogenous short ratemodels, i.e. the assumed short rate dynamics depended only constantcoefficients. The focus would be on the Vasicek (1977), the Dothan

(1978) and the Cox, Ingersoll and Ross (1985) models.

Bonds and Interest Rates   23 / 54

Bonds and Interest Rates

Page 24: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 24/54

The Vasicek Model (1977)

Vasicek introduced the first model to capture the mean reversionproperty of interest rates. In the Vasicek model, which is based onthe Ornstein-Uhlenbeck process, the short term interest rate processr t  satisfies the SDE

dr t  = k  (θ − r t ) dt  + σdW t 

where  k , θ  and  σ  are positive constants. The model has the propertyof being statistically stationary in time, i.e. the distribution of  r t 

 −r s 

depends only on  t  − s . However, for each time  t , the short rate  r t can be negative with positive probability.

Bonds and Interest Rates   24 / 54

Bonds and Interest Rates

Page 25: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 25/54

Solving the above equation will give us,

r t    =   θ + e −kt 

r 0 − θ +

   t 

0

σe ku dW u 

=   r 0e −kt  + θ 1 − e −kt  + σ    t 

0

e −k (t −u )dW u 

and for  s  ≤ t 

r t  = r s e −k (t −s ) + θ 1 − e −k (t −s ) + σ    t 

e −k (t −u )dW u 

Bonds and Interest Rates   25 / 54

Bonds and Interest Rates

Page 26: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 26/54

Hence,   r t  conditional on F s   is normally distributed with mean and

variance given respectively by

E  [ r t | F s ] =   E 

r s e −k (t −s ) + θ

1 − e 

−k (t −s )

 + σ

   t 

e −k (t −u )

dW u 

F s 

=  r 

s e −k (t −s )

+ θ

1 −e −k (t −s )

 + E 

σ   t s 

e −k (t −u )

dW u F s 

=   r s e −k (t −s ) + θ

1 − e 

−k (t −s )

 +  E 

σ

   t 

e −k (t −u )

dW u 

=   r s e −k (t −s ) + θ 1

−e −k (t −s )

Bonds and Interest Rates   26 / 54

Bonds and Interest Rates

Page 27: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 27/54

Var  [ r t | F s ] =   E (r t  − E  [ r t | F s ])2

F s =   E 

σ2

   t 

e −k (t −u )dW u 

2F s 

=   σ2E    t 

e −k (t −u )dW u 2

=   σ2E 

   t 

e −2k (t −u )du 

=   σ2

E    t 

s  e −2k (t −u )

du 

=  σ2

2k 

1 − e −2k (t −s )

Bonds and Interest Rates   27 / 54

Bonds and Interest Rates

Page 28: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 28/54

Based on the conditional mean and variance, the short rate  r t   ismean reverting, since the expected rate approaches the value θ  as  t 

goes to infinity. The fact that  θ  can be regarded as a long term

average rate could be inferred from the dynamics of the SDE itself.Notice that the drift of the process  r t   is positive whenever the shortrate is below  θ  and negative otherwise, so that  r   is pushed, at everytime  t , closer towards the average level θ.

Bonds and Interest Rates   28 / 54

Bonds and Interest Rates

Page 29: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 29/54

The Dothan Model (1978)

The Dothan model basically follows the GBM dynamics for the shortrate

dr t  = ar t dt  + σr t dW t 

in which case

r t  = r s  exp

a − 1

2σ2

(t  − s ) + σ (W t  −W s )

for s 

 ≤t .

Bonds and Interest Rates   29 / 54

Bonds and Interest Rates

Page 30: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 30/54

Hence,  r t  conditional on F s   is lognormally distributed with mean and

variance given by

E  [ r t | F s ] =   E 

r s  exp

a −  1

2σ2

(t  − s ) + σ (W t  −W s )

F s 

=   r s e (a− 1

2σ2)(t −s )

e σ(W t −W s )

|F s =   r s e (a−

12σ2)(t −s )E 

e σ(W t −W s )

  W t  −W s  ∼ N (0, t  − s )

=   r s e (a−12σ2)(t −s )e 

12σ2(t −s )

=   r s e a(t −s )

Bonds and Interest Rates   30 / 54

Bonds and Interest Rates

Page 31: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 31/54

Var  [ r t | F s ] =   E 

r 2t  |F s − (E  [r t |F s ])2

=   E 

r 2s  e (2a−σ2)(t −s )+2σ(W t −W s )|F s − r 2s  e 2a(t −s )

=   r 2s  e (2a−σ2)(t −s )E e 2σ(W t −W s )

|F s − r 2s  e 2a(t −s )

=   r 2s  e (2a−σ2)(t −s )E 

e 2σ(W t −W s )− r 2s  e 2a(t −s )

=   r 2s  e (2a−σ2)(t −s )e 12

(2σ)2(t −s ) − r 2s  e 2a(t −s )

=   r 2s  e 2a(t −s ) e σ2(t −s )

−1

Bonds and Interest Rates   31 / 54

Bonds and Interest Rates

Page 32: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 32/54

The lognormal distribution implies that  r t   is always positive for eacht , so that the main drawback of Vasicek is addressed here. However,the process  r t  is mean-reverting if and only if  a < 0, with thereversion level that must be necessarily equal to 0.

Remark:  An alternative to this, which is also a lognormal short ratemodel, is the Exponential Vasicek model, where the short rate isalways mean reverting (refer to handout for the formulation andproperties of the model)

Bonds and Interest Rates   32 / 54

Bonds and Interest Rates

Page 33: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 33/54

Cox-Ingersoll-Ross (1985)

The general equilibrium approach developed by Cox, Ingersoll andRoss (1985) led to the introduction of a “square root” term in thediffusion coefficient of the instantaneous short rate dynamics

proposed by Vasicek. The resulting model, which is also meanreverting, has been a benchmark for many years because of itsanalytical tractability and the fact that the short rate is alwayspositive.

Bonds and Interest Rates   33 / 54

Bonds and Interest Rates

Page 34: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 34/54

The model formulation in this case is given by

dr t  = k (θ − r t )dt  + σ√ 

r t dW t 

where  k , θ  and  σ  are positive constants. The condition 2k θ > σ

2

hasto be imposed to ensure that the origin is inaccessible to the processdescribed above, so that we can grant that  r t  remains positive.

Bonds and Interest Rates   34 / 54

Bonds and Interest Rates

Here rt follows a noncentral chi-squared distribution. Denoting pY

Page 35: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 35/54

Here  r t  follows a noncentral chi squared distribution. Denoting  p Y the density function of the random variable  Y ,

p r t (x ) =   p χ2(v ,λt )/c t (x ) = c t p χ2(v ,λt )(c t x )

c t    =  4k 

σ2 (1 − exp(−kt ))  v  =

 4k θ

σ2  λt  = c t r 0 exp(−kt )

where the noncentral chi-squared distribution function χ2(·, v , λ) with

v  degrees of freedom and non-centrality parameter  λ  has density

p χ2(v ,λ)(z ) =∞i =0

e −λ/2 (λ/2)i 

i !  p Γ(i +v /2,1/2)(z )

p Γ(i +v /2,1/2)(z ) =   (1/2)i +v /2

Γ(i  + v /2)z i −1+v /2e −z /2 = p χ2(v +2i )(z )

with  p χ2(v +2i )(z ) denoting the density of a central chi-squareddistribution with  v  + 2i  degrees of freedom.

Bonds and Interest Rates   35 / 54

Bonds and Interest Rates

Page 36: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 36/54

The mean and variance of  r t  conditional on F s  are given by

E  [ r t | F s ] =   r s e −k (t −s ) + θ(1 − e −k (t −s ))

Var  [ r t | F s ] =   r s σ2

e −k (t −s )

− e −2k (t −s 

 + θ

σ2

2k (1 − e −k (t −s )

)2

Bonds and Interest Rates   36 / 54

Bonds and Interest Rates

Page 37: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 37/54

We remark that while the above time homogenous models allow forthe analytical pricing of bond and bond options, these models don’t

tend to fit well with initial observed term structure in the market,regardless of how the parameters are chosen. Consequently, somepractitioners are reluctant to apply such kinds of models. Hence, wemay consider time dependent extensions of the above models. For

example, the Hull White model

dr t  = k  (θt  − r t ) dt  + σdW t 

is a time dependent extension of the Vasicek model. The CIR also

admits a similar time-dependent extension (refer to handout for moremodels).

Bonds and Interest Rates   37 / 54

Bonds and Interest Rates

Bond price and risk neutral measure

Page 38: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 38/54

Bond price and risk neutral measure

We highlight the relationship between the bond price  P (t , T ) and theshort rate process {r t }t ∈R+  under the absence of arbitrage conditionby considering the following scenarios:

The short rate is a deterministic constant  r  > 0. In this case,P (t , T ) should satisfy

e r (T −t )P (t , T ) = P (T , T ) = 1,   t  ∈ [0, T ]

thereforeP (t , T ) = e −r (T −t ),   t  ∈ [0, T ]

Bonds and Interest Rates   38 / 54

Bonds and Interest Rates

Page 39: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 39/54

The short rate is time dependent and deterministic function  r t .In this case, it can be shown that

P (t , T ) = e −  T t    r u du ,   t  ∈ [0, T ] (4)

Bonds and Interest Rates   39 / 54

Bonds and Interest Rates

Page 40: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 40/54

The short rate is a stochastic process  r t . We remark that (4)does not make sense since the price  P (t , T ) being set at time  t ,can depend only on information known up to time  t . In fact, theprice for this case would be given by

P (t , T ) = E P e −

  T 

t   r u du F t 

,   t 

 ∈[0, T ]

where the expectation is under the risk neutral measure  P. Itgives us the “best possible estimate” of the future quantity

e −  T 

t   r u du  given information known up to time  t .  Furthermore, as

a conditional expectation with respect to F t , the bond priceP (t , T ) is F t -measurable.

Bonds and Interest Rates   40 / 54

Bonds and Interest Rates

Page 41: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 41/54

We remark that the risk neutral measure  P  is equivalent to  P, underwhich for all  s  ∈ [0, T ], the process {P (t , s )}0≤s ≤T   defined by

P (t , s ) = e −  t 

0  r u du P  (t , s ) =

  P  (t , s )

B t 

is a  P-martingale.

Bonds and Interest Rates   41 / 54

Bonds and Interest Rates

Page 42: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 42/54

For the succeeding discussion on bond pricing, we will assume thatthe short rate process is under the measure P, and that the filtration{F t }   is generated by the P-Brownian motion {W t }. Indeed, the termstructure, as well as prices of all other interest rate derivatives, arecompletely determined by specifying the dynamics of the short rateunder P, and the objective probability measure  P can just be ignored.

Bonds and Interest Rates   42 / 54

Page 43: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 43/54

Bonds and Interest Rates

Page 44: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 44/54

The arbitrage price of the bond is given by

P (t .T ) = E P

e −

  T 

t   r s ds 

F t  = e A(t ,T )−B (t ,T )r t 

where

B (t , T ) = 1b 

1 − e −b (T −t )

and

A(t , T ) = 2ab 

−σ2

2b 2   (B (t , T ) − T  + t ) − σ2

4b B 2(t , T )

Bonds and Interest Rates   44 / 54

Page 45: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 45/54

Bonds and Interest Rates

Page 46: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 46/54

Affine Term Structure Model

Affine term structure models are interest rate models where thecontinuously compounded spot rate  R (t ,T ) is an affine function in theshort rate   r t , that is

R (t ,T ) = α(t ,T ) + β (t ,T )r t 

where  α  and  β  are deterministic functions of time. If this happens, themodel is said to possess an affine term structure.

Bonds and Interest Rates   46 / 54

Bonds and Interest Rates

Page 47: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 47/54

This relationship is always satisfied when the zero-coupon bond pricecan be written in the form

P (t , T ) = C (t , T )e −B (t ,T )r t 

since then clearly it suffices to set

α(t , T ) = − ln C (t , T )/(T  − t )   β (t , T ) = B (t , T )/(T  − t )

Both Vasicek and CIR models seen earlier are affine models, sincetheir corresponding bond prices has the above form. The Dothanmodel is not an affine model.

Bonds and Interest Rates   47 / 54

Bonds and Interest Rates

Page 48: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 48/54

Now assume that we have the risk neutral dynamics for the short rate

given bydr t  = b (t , r t )dt  + σ(t , r t )dW t 

The conditions for which  b  and σ  admits an affine term structure isgiven by

b (t , x ) =   λ(t )x  + η(t )

σ(t , x ) = 

γ (t )x  + δ (t )

that is,  b  and σ2

are also affine functions themselves.

Bonds and Interest Rates   48 / 54

Bonds and Interest Rates

Forward Rate and the HJM condition

Page 49: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 49/54

We recall that instantaneous forward rate with maturity  T ,contracted at  t , is defined by

f   (t , T ) = −   ∂ 

∂ T   log P (t , T )

If the short rate folows the Vasicek stochastic interest rate model,that is

dr t  = (a − br t )dt  + σdW t 

then

f   (t , T ) = r t e −b (T −t ) + a

1 − e −b (T −t )

−   σ2

2b 2

1 − e −b (T −t )

2

Bonds and Interest Rates   49 / 54

Bonds and Interest Rates

Now suppose we determine the dynamics of the process f (t T ) of

Page 50: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 50/54

Now suppose we determine the dynamics of the process   f   (t , T ) of forward rates in the Vasicek model

df   (t , T ) =   e −b (T −t )dr t  + be −b (T −t )r t dt  +

be −b (T −t )dt 

dr t 

−ae −b (T −t )dt  + σ2

1 − e −b (T −t )

e −b (T −t )dt 

=   e −b (T −t ) ((a

−br t )dt  + σdW t ) + be −b (T −t )r t dt 

−ae −b (T −t )dt  + σ2

1 − e −b (T −t )

e −b (T −t )dt 

=  σ2

b  1 − e −b (T −t )

e −b (T −t )dt  + σe −b (T −t )dW t 

=   σ2e −b (T −t )   T 

e −b (s −t )ds 

dt  + σe −b (T −t )dW t 

Bonds and Interest Rates   50 / 54

Bonds and Interest Rates

Hence  df   (t , T ) can be written as

( ) ( ) ( )

Page 51: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 51/54

df   (t , T ) = α(t , T )dt  + σ(t , T )dW t 

with

σ(t , T ) = σe −b 

(T −t 

)

and

α(t , T ) =   σ2e −b (T −t )    T 

e −b (s −t )ds =   σe −b (T −t )

   T 

σe −b (s −t )ds 

=   σ(t , T )    T 

σ(t , s )ds 

We note that the above relation between  σ  and  α  is a not acoincidence, but rather a general consequence of the absence of arbitrage hypothesis on the dynamics of forward rates. This will beelaborated in succeeding section.

Bonds and Interest Rates   51 / 54

Bonds and Interest Rates

Page 52: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 52/54

Heath-Jarrow-Morton (HJM) condition

In the HJM model, the dynamics for the instantaneous forward rate isgiven by

df   (t , T ) = α(t , T )dt  + σ(t , T )dW t 

where the date  T  is fixed. At this point, our objective is to determinethe conditions in which the above equation will make sense in thefinancial context (that is, satisfying the absence of arbitragecondition)

Bonds and Interest Rates   52 / 54

Bonds and Interest Rates

Page 53: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 53/54

To determine this, we remark that at this point we have two differentformulas for bond prices

P (t , T ) = e −  T 

t   f   (t ,u )du  and   P (t , T ) = E 

P

e −

  T 

t   r u du 

F t Both of these formulas should hold simultaneously so that there willbe absence of arbitrage.

This will ultimately yield the consistency relation between  α  and σ,given by

α(t , T ) = σ(t , T )   T 

σ(t , s )ds 

Bonds and Interest Rates   53 / 54

Bonds and Interest Rates

Page 54: AMF 241 Interest Rate Models

8/9/2019 AMF 241 Interest Rate Models

http://slidepdf.com/reader/full/amf-241-interest-rate-models 54/54

References

1 Bjork, Tomas.  Arbitrage Theory in Continuous Time, 2nd Edition. Oxford University Press, 2004.

2 Privault, Nicolas.  An Elementary Introduction to Stochastic Interest Rate Modelling . World Scientific Publishing Co. Pte.

Ltd, 20083 Brigo, Damiano and Mercurio, Fabio.   Interest Rate Models -

Theory and Practice (with Smile, Inflation and Credit). SpringerFinance. Springer-Verlag, Berlin, 2006.

Bonds and Interest Rates   54 / 54