Pricing Calls on Defaultable Bonds with Stochastic Rates and Hazard Rates
Summary
The document sets up a call option on a defaultable bond under a risk-neutral measure. Both the risk-free interest rate and the default hazard rate follow mean-reverting stochastic processes. The option payoff is conditional on the issuer surviving to the option date and discounted by the accumulated risk-free rate. The author then tries to split the payoff and change numeraire using a risk-free zero-coupon bond.
The central difficulty is the survival weighting, expressed through the integrated hazard rate, which remains in the expectation after the numeraire change. The question asks how to proceed toward a pricing formula, possibly by analogy with Black–Scholes, but the document contains no answer or derivation. It therefore illustrates a modeling issue rather than supplying a valuation method. Any closed-form or numerical approach would depend on assumptions about the joint dynamics of rates, hazard rates, default, and the defaultable bond price that are not resolved here.
Key ideas
- The setup models the risk-free rate and default hazard rate as mean-reverting stochastic processes.
- The call payoff depends on the issuer surviving to the option date.
- Changing to a risk-free bond numeraire leaves a survival-weighting term involving the integrated hazard rate.
- The document poses, but does not solve, the challenge of pricing the resulting expectation.
- A Black–Scholes-style formula cannot be inferred from the setup alone without further assumptions or derivation.
Tags
Full text
# Price of a Bond-Call option in the defaultable framework
# Price of a Bond-Call option in the defaultable framework
I would like to compute the price for a Call option written on a defaultable bond as underlying. Suppose you have the following dynamic under the risk free measure $\mathcal{Q}$ for the interest rate: $$ d r_t = ( a^r - b^r r_t) \, dt + \sigma^r dW^r_t \\ $$ with initial point $r_0$, and the following dynamic for the hazard rate: $$ d \gamma_t = ( a^\gamma - b^\gamma \gamma_t) \, dt + \sigma^\gamma dW^\gamma_t \\ $$ with initial point $\gamma_0$. I'm gonna call $p_1(T_1, T_2)$ the price at time $T_1$ for a defaultable bond with maturity $T_2$ and interest rate $r_t$. We can also suppose that there exists a risk-free bond $p_0(T_1, T_2)$ (again with interest rate $r_t$).
I will write $\tau$ in order to denote the random variable of the default time so that its density is
$$ \gamma_t e^{-\int_0^t \gamma_s \, ds} $$
The very classical formula for the price of the call option in this case is:
$$ E^{Q} \left[ I_{\tau > T_1} e^{-\int_0^{T_1} r_s \, ds} \left( p_1(T_1, T_2) - K \right)^+ \right] $$
By using a classical theorem (Theorem 9.23; Quantitative Risk Management; McNeil, Frey, Embrechts) we have the following result:
$$ = E^{Q} \left[ e^{\int_0^{T_1} r_s + \gamma_s \, ds } \left( p_1(T_1, T_2) - K \right)^+ \right] I_{\tau > 0} . $$
Here I have some trouble: I would like to do a change of numeraire (the really standard technique) and use the Black-Scholes formula. The problem is the "spred term", i.e. the exponential with $\gamma$. Infact let me do the usual splitting of the Call payoff: $$ = E^{Q} \left[ e^{-\int_0^{T_1} r_s + \gamma_s \, ds } p_1(T_1, T_2) I_{ p_1(T_1, T_2) - K >0} \right] - K E^{Q} \left[ e^{-\int_0^{T_1} r_s + \gamma(s) \, ds } I_{ p_1(T_1, T_2) - K >0} \right] $$
In order to compute the first term I will change the numeraire by using $p_0$:
$$ p_0(0, T_1)E^{Q^{p_0}} \left[ e^{-\int_0^{T_1} \gamma_s \, ds } I_{ p_1(T_1, T_2) - K >0} \right] = p_0(0, T_1)E^{Q^{p_0}} \left[ e^{-\int_0^{T_1} \gamma(s) \, ds } I_{ \frac{p_1(T_1, T_2)}{p_0(T_1, T_1)} - K >0} \right] $$ where in the last passage I've used the fact that $p_0(T_1, T_1) = 1$. At this point I'm little bit lost. Infact in the case of non defaultable bond everything is fine since there is no $\gamma$-term and I can use the Black-Scholes formula.
My question is: what is, more or less, the idea to compute the price in the defaultable framework?
Ciao! AMShown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.