Pricing Interest Rate Derivatives: The Vasicek and Hull - PowerPoint PPT Presentation

1 / 25
About This Presentation
Title:

Pricing Interest Rate Derivatives: The Vasicek and Hull

Description:

Term Structure Models versus Black's Model. Black's model is concerned ... expression of the same structure as the Black & Scholes model for equity options. ... – PowerPoint PPT presentation

Number of Views:634
Avg rating:3.0/5.0
Slides: 26
Provided by: sim126
Category:

less

Transcript and Presenter's Notes

Title: Pricing Interest Rate Derivatives: The Vasicek and Hull


1
Pricing Interest Rate Derivatives The Vasicek
and Hull White Models
2
Term Structure Models versus Blacks Model
  • Blacks model is concerned with describing the
    probability distribution of a single variable at
    a single point in time
  • Example
  • When valuing a swap option, we assume that the
    swap rate at the maturity date of the option is
    lognormal
  • We dont assume anything about swap rate at any
    other date
  • A term structure model describes the evolution of
    the whole yield curve

3
The Zero Curve
  • In a world in which the spot rate account is the
    numeraire, the value f(t) at time t of a
    derivative with payoff fT at time T is
  • where is the average interest rate over the
    period from t to T
  • For the price P(t,T) at t of a zero-coupon bond
    with maturity date T we get
  • If P(t,T) would not be given by the above
    expression there would be an arbitrage
    opportunity.
  • R(t,T) is the continuously compounded interest
    rate at time t for a maturity of T-t

4
  • So we have
  • Which leads to
  • This equation shows that the complete term
    structure of interest rates at some time t is
    determined by the risk-neutral dynamics of the
    spot rate
  • US zero rates at 20/02/2008

5
One-factor Models
  • Consider the following physical dynamics for the
    spot rate r
  • The spot rate is driven by a single Brownian
    motion z.
  • Such models are called one-factor models
  • The risk-neutral dynamics are then given by
  • Example Rendleman Bartter Model
  • The physical dynamics are
  • The risk-neutral dynamics are
  • This is geometrical Brownian motion, as for a
    stock

6
Mean Reversion
  • The assumption that the spot rate follows a
    geometrical Brownian motion is not realistic.
  • Over time interest rates are pulled back to some
    long-run average level
  • The Rendleman Bartter model does not include
    this feature

7
  • Mean reversion of interest rates is linked to the
    business cycle.
  • When the economy is at its peak, demand for funds
    is high, leading to high interest rates.
  • As the economy cools down, demand for funds
    decreases, and interest rates go down.
  • Central banks try to prevent the economy from
    overheating by increasing short term rates, and
    they try to stimulate the economy during a
    slowdown by reducing short-term rates.

8
The Vasicek Model
  • The physical dynamics of the spot rate are given
    by
  • We assume that the market price of risk is a
    constant ?. The risk-neutral dynamics are then
    given by
  • where
  • Calibrating the model to data, one always finds
    that ?0, and thus that .
  • Both with the physical and the risk-neutral
    dynamics, the future spot rate has a normal
    distribution.

9
  • On can show that in the Vasicek model one has
  • where A(t,T) and B(t,T) are given by
  • From the expression for P(t,T) we get
  • We see that all zero rates are linear functions
    of the spot rate r.
  • Weak point of the Vasicek model interest rates
    can become negative. Although for realistic
    values for the parameters this is very unlikely.

10
The Cox-Ingersoll-Ross Model
  • The physical dynamics of the spot rate are given
    by
  • We assume that the market price of risk is of the
    form
  • The risk-neutral dynamics are then given by
  • where
  • Calibrating the model to data, one always finds
    that ?0, and thus that .
  • Both with the physical and the risk-neutral
    dynamics, the future spot rates do not have a
    normal distribution.

11
  • Both for the physical ad the risk-neutral
    dynamics, we have an equation f the following
    type
  • If r becomes very small (close to zero), the
    diffusion term goes to zero as well, and the
    drift term is strictly positive.
  • As such, r is pushed away from zero, and can not
    become negative (or zero).
  • Also in the CIR model one has
  • But now, A(t,T) and B(t,T) are given by
  • where

12
Possible Term Structuresin Vasicek CIR Models
  • In both models the zero rates that make up the
    term structure are linear functions of the spot
    rate.
  • Therefore, the shape of he term structure at some
    future date t, is not driven by the value of
    r(t), but is mainly determined by the shape of
    the functions A(t,T) and B(t,T)
  • r(t) mainly drives the level of the term structure

13
Two-factor Models
  • The assumption that a single factor (dz) drives
    the entre term structure is quite restrictive.
  • As an immediate consequence we have that all zero
    rates are completely determined by the spot rate.
  • Two possible generalizations
  • Assume that the mean-reversion level b is
    stochastic and follows its on stochastic process
  • Assume that the volatility s is stochastic
  • In both cases, the term structure is driven by
    two factors
  • The spot rate and the mean-reversion level of it
  • The spot rate and its volatility

14
No-Arbitrage Models
  • In an equilibrium model todays term structure
    is an output
  • The term structure observed in the market can not
    be fitted by the model.
  • The Vasicek and CIR models only have thee
    parameters
  • As a result, the zero-coupon bond prices implied
    by the model will not be the same as those
    obtained from market prices.
  • This could potentially lead to arbitrate
    opportunities, as prices of options would not be
    consistent with prices of traded bonds.
  • In a no-arbitrage model todays term structure
    is an input
  • This is achieved by making one of the parameters
    time-depend

15
Developing No-Arbitrage Model for r
  • A model for r can be made to fit the initial term
    structure by including a function of time in the
    drift
  • The Ho-Lee Model
  • The risk-neutral dynamics are
  • dr ?(t)dt sdz
  • The variable ?(t) determines the average
    direction of the movement of the spot rate r at
    time t.
  • Just as a implied volatilities allow the BS
    model to replicate observed market prices for
    European options on stocks, the parameter ?(t)
    allows the Ho-Lee model to be calibrated to the
    observed term structure.
  • Interest rates normally distributed
  • One volatility parameter, s

16
  • A time t0 the variable ?(t) can be calculated
    as
  • The average direction that the spot rate will
    move to at a future date t is given by the
    current slope of the forward curve.
  • Many analytic results for bond prices and option
    prices
  • For zero-coupon bonds we have
  • where
  • In these equations time zero is today, and t and
    T and dates in the future with tT.

17
The Hull and White Model
  • The risk-neutral dynamics are
  • dr q(t) ar dt sdz
  • or
  • dr aq(t )/a r dt sdz
  • The Hull-White model is the no-arbitrage version
    of the Vasicek model.
  • Many analytic results for bond prices and option
    prices
  • Two volatility related parameters, a and s
  • Interest rates normally distributed
  • The function q(t) is given by

18
  • Bond prices in the Hull and White model are again
    given by
  • where now
  • We still have that all zero rates are determined
    by the spot rate
  • The slope of he yield curve at some future date t
    is not driven by the level of r(t) but by the
    functions A(t,T) and B(t,T).

19
Options on Zero-Coupon Bonds
  • In Vasicek and Hull-White model, price of call
    maturing at T on a bond lasting to s is given by
    the following expression
  • LP(0,s)N(h)-KP(0,T)N(h-sP)
  • The price of put is
  • KP(0,T)N(-h s P)-LP(0,s)N(-h)
  • The variable s P is the standard deviation of
    the log of the bond price at time T.

20
  • This shows that we obtain an expression of the
    same structure as the Black Scholes model for
    equity options.

21
  • For the CIR model the expressions are more
    complicated and involve complicated integrals

22
Options on Coupon Bearing Bonds
  • In any one-factor model a European option on a
    coupon-bearing bond can be expressed as a
    portfolio of options on zero-coupon bonds.
  • Consider a call with strike K and maturity date
    T.
  • We first calculate the critical value for the
    spot rate r(T) at the option maturity for which
    the coupon-bearing bond price equals the strike
    price K at maturity
  • The strike price for each zero-coupon bond is set
    equal to its value when the interest rate equals
    this critical value
  • The value of the option on the coupon bond is
    then given by the sum of the options on the
    corresponding zero-coupon bonds, with strike
    prices as calculated above.

23
Calibrating the Vasicek Model
  • The physical dynamics
  • The physical dynamics of the spot rate are given
    by
  • or
  • for some fixed time-step ?t
  • From this we get
  • We can obtain estimates for the three parameters
    by running a regression.

8
24
  • We need to estimate the intercept and the slope
    of the regression, as well as the standard
    deviation of the errors.
  • At this point we have the parameters for the
    physical dynamics.
  • The risk-neutral dynamics
  • The risk-neutral dynamics, used for pricing, are
    given by
  • The values for a and s we know already.
  • We need to estimate the value of bQ

25
  • The parameter bQ is given that value which
    minimizes the pricing errors for selection of
    bonds.
  • We minimize the mean squared error
  • The risk-neutral dynamics An alternative
  • In order to allow the model to better fit the
    current term structure, one can re-estimate the
    parameter a.
  • In this case, only the estimate from the
    regression analysis that is used for pricing is
    the estimate for s.
  • Typically, the two estimates for a are different
Write a Comment
User Comments (0)
About PowerShow.com