This post discretizes Hull-White 2-factor model and provide derivations of the simulation equations.
Earlier posts on Hull-White 2-factor model
We try to price an interest derivatives which have cashflows at times \(T_1\),\(T_2\),...,\(T_N\).
When we let \(f(T_j)\) denote a cash flow at time \(T_j\), the price of this product is
\[ P_0 = \displaystyle\sum_{j=1}^{N} E\left[\frac{f(T_j)}{B_{T_j}} \right] \]
This pricing is the risk-neutral pricing and needs cash flows and discount factors from future interest rate simulations.
Since \(dW_1 (t)\) and \(dW_2 (t)\) follow \(\sqrt{\Delta t} N(0,1)\) independently, \(dx(t)\) and \(dy(t)\) can be discretized as follows.
\[\begin{align}
x_{t+\Delta t} &= x_t - a(t)x_t \Delta t+\sigma(t)\epsilon_1 \sqrt{\Delta t} \\
y_{t+\Delta t} &= y_t - b(t)y_t \Delta t
+\eta(t)\{ \rho \epsilon_1 + \sqrt{1-\rho^2}\epsilon_2 \}\sqrt{\Delta t}
\end{align}\]
Here, \(\epsilon_1\) and \(\epsilon_2\) are random numbers from the standard normal distribution.
Like HW 1-factor model, the same discretized time axis is used as follows ( \(\Delta t_i = t_{i+1} - t_i\)).
\[\begin{align}
0 = t_0 &< t_1 < t_2 < t_3 < ... < t_{M_1 -1} < t_{M_1} = T_1 \\
&< t_{M_1 +1} < t_{M_1 +2} < ... < t_{M_2 -1} < t_{M_2} = T_2 \\
&< t_{M_2 +1} < t_{M_2 +2} <...
\end{align}\]
For this time axis, stochastic process of discretized \(x(t)\) and \(y(t)\) has the following form.
\[\begin{align}
x_{t_{i+1}} &= x_{t_i} e^{-\int_{t_i}^{t_{i+1}} a(v)dv} + \epsilon_1 \sqrt{\int_{t_i}^{t_{i+1}}\sigma(u)^2 e^{-2 \int_{u}^{t_{i+1}}a(v)dv}du} \\
y_{t_{i+1}} &= y_{t_i} e^{-\int_{t_i}^{t_{i+1}} b(v)dv}
+ \{ \rho \epsilon_1 + \sqrt{1-\rho^2}\epsilon_2 \} \sqrt{\int_{t_i}^{t_{i+1}}\eta(u)^2 e^{-2 \int_{u}^{t_{i+1}}b(v)dv}du}
\end{align}\]
Since \(x_{t_0}\),\(x_{t_1}\), \(x_{t_2}\), \(x_{t_3}\), ..., and \(y_{t_0}\),\(y_{t_1}\), \(y_{t_2}\), \(y_{t_3}\), ... are easily obtained from this scenario generating equation, discount factors at time \(T_j\) is
\[ \frac{1}{B_{T_j}} = \prod_{i=0}^{M_j-1} P(t_i , t_{i+1}) \]
\[\begin{align}
P(t_i , t_{i+1}) &= \frac{P(0 , t_{i+1})}{P(0 , t_i)} \exp \left( -x_{t_i} B_1(t_i,t_{i+1}) - y_{t_i} B_2(t_i,t_{i+1})
+\frac{1}{2}\Omega(t_i , t_{i+1}) \right) \\
\Omega(t_i , t_{i+1}) &= \int_{0}^{t_i}\sigma(u)^2 \{ B_1(u,t_i)^2-B_1(u,t_{i+1})^2 \}du \\
&+\int_{0}^{t_i}\eta(u)^2 \{ B_2(u,t_i)^2-B_2(u,t_{i+1})^2 \}du \\
+&2\rho\int_{0}^{t_i}\sigma(u)\eta(u) \{ B_1(u,t_i)B_2(u,t_i)-B_1(u,t_{i+1})B_2(u,t_{i+1}) \}du
\end{align}\]
Cash flow at time \(T_j\) is
\[ R(t_i , {t_i}+\tau) = \frac{1}{{\tau}} \left\{ {\frac{1}{P(t_i , {t_i}+\tau)} -1} \right\} \]
\[\begin{align}
P(t_i , {t_i}+\tau) &= \frac{P(0 , {t_i}+\tau)}{P(0 , t_i)} \exp \left(
-x_{t_i} B_1(t_i,{t_i}+\tau)-y_{t_i} B_2(t_i,{t_i}+\tau)
+\frac{1}{2}\Omega(t_i , {t_i}+\tau) \right) \\
\Omega(t_i , {t_i}+\tau) &=
\int_{0}^{t_i}\sigma(u)^2 \{ B_1(u,t_i)^2-B_1(u,{t_i}+\tau)^2 \}du \\
& +\int_{0}^{t_i}\eta(u)^2 \{ B_2(u,t_i)^2-B_2(u,{t_i}+\tau)^2 \}du \\
+& 2 \rho\int_{0}^{t_i}\sigma(u)\eta(u) \{ B_1(u,t_i)B_2(u,t_i)-B_1(u,{t_i}+\tau)B_2(u,{t_i}+\tau) \}du
\end{align}\]
Therefore, for discount factors and cash flows from this scenario, the price of interest derivatives \(P_0\) which has cash flows \(f(T_1)\),\(f(T_2)\),...,\(f(T_N)\) at time \(T_1\),\(T_2\),...,\(T_N\) respectively under the risk-neutral measure is as follows.
\[ P_0 = \displaystyle\sum_{j=1}^{N} E\left[\frac{f(T_j)}{B_{T_j}} \right] \]
The present value is the average from iterating this process with a number of scenario. This is the same case with the Hull-White 1-factor model except for correlated related cross terms. Expressions for this kind of terms are presented as product (or multiplication) terms not as square terms.
This manipulation for cross terms also holds for numerical integration and calculation which are also similar to 1-factor case. So we do not repeat these similar works.
\(\blacksquare\)
No comments:
Post a Comment