FX forward with stochastic interest rates pricing
I would like to extend the following question about FX Forward rates in stochastic interest rate setup: “Expectation” of a FX Forward
We consider a FX process $ X_t = X_0 \exp( \int_0^t(r^d_s-r^f_s)ds -\frac{\sigma^2}{2}t+ \sigma W_t) $ where $ r^d $ and $ r^f $ are stochastic processes not independent of the Brownian motion $ W $ . As we know the FX Forward rate is $ F^X(t,T) = E_t^d\left[X_T \right] $ under the domestic risk-neutral measure.
The question is how to show that $ F^X(t,T) = X_t \frac{B_f(t,T)}{B_d(t,T)} $ where $ B_d(t,T) $ and $ B_f(t,T) $ are respective the domestic and foreing zero-coupon bond prices of maturity $ T $ 有時 $ t $ .
自從 $ X_T = X_t \exp\left( \int_t^T(r^d_s-r^f_s)ds+ \sigma (W_T-W_t)\right) $
$$ \begin{align} F^X(t,T) &= X_t E_t^d\left[\exp\left( \int_t^T(r^d_s-r^f_s)ds-\frac{\sigma^2}{2}(T-t) + \sigma (W_T-W_t)\right)\right] \& =X_t E_t^d\left[\exp\left( \int_t^T(r^d_s-r^f_s)ds \right) \frac{\mathcal E_T(\sigma W )}{\mathcal E_t(\sigma W )}\right] \&= X_t E_t^d\left[\exp\left( \int_t^T(r^d_s-r^f_s)ds \right) \frac{d\mathcal Q^f}{d\mathcal Q^d} \frac{1}{E_t^d \left[\frac{d\mathcal Q^f}{d\mathcal Q^d}\right]}\right] \&= X_t E_t^f\left[\exp\left( \int_t^T(r^d_s-r^f_s)ds \right) \right] \end{align} $$ 現在如何得出結論 $ r^d $ 和 $ r^f $ 不一定彼此獨立,因為它們都依賴於布朗運動 $ W $ (順便假設我們在自然過濾中工作 $ W $ )?
編輯
我想將我的問題擴展到無本金交割外匯遠期的定價。我在這裡發布了一個新問題:FX 遠期定價與 FX 和 Zero-Cupon 之間的相關性。
公式 $ F^X(t,T) = E_t^d\left(X_T \right) $ ,在國內風險中性措施下,是有問題的。請注意,有時 $ t $ , 遠期匯率 $ F^X(t,T) $ , 成熟度 $ T $ , 是匯率使得收益 $ X_T-F^X(t,T) $ 有一個零值 $ t $ . 那是,
$$ \begin{align*} B_t^d E_d\left(\frac{X_T-F^X(t,T)}{B_T^d} \mid \mathcal{F}_t\right)=0, \end{align*} $$ 在哪裡 $ E_d $ 是國內風險中性測度下的預期 $ Q_d $ . 這裡, $ B_t^d $ 和 $ B_t^f $ 分別表示國內和國外貨幣市場賬戶價值。然後, $$ \begin{align*} F^X(t,T) &= \frac{1}{B_d(t, T)}B_t^d E_d\left(\frac{X_T}{B_T^d} \mid \mathcal{F}_t\right) \tag{1}\ &\neq E_d(X_T \mid \mathcal{F}_t), \end{align*} $$ 在隨機利率假設下。 讓 $ Q_d^T $ 做國內的 $ T $ - 前向測量,和 $ E_d^T $ 是對應的期望運算元。那麼,對於 $ 0 \le t \le T $ ,
$$ \begin{align*} \frac{dQ_d}{dQ_d^T}\big|_t = \frac{B_t^d B_d(0, T)}{B_d(t, T)}. \end{align*} $$ 從 $ (1) $ , $$ \begin{align*} F^X(t,T) &= \frac{1}{B_d(t, T)}B_t^d E_d\left(\frac{X_T}{B_T^d} \mid \mathcal{F}_t\right)\ &=\frac{1}{B_d(t, T)}B_t^d E_d^T\left(\frac{X_T}{B_T^d} \frac{\frac{dQ_d}{dQ_d^T}\big|_T}{\frac{dQ_d}{dQ_d^T}\big|_t}\mid \mathcal{F}_t\right)\ &=\frac{1}{B_d(t, T)}B_t^d E_d^T\left(\frac{X_T}{B_T^d} \frac{B_T^dB_d(t, T)}{B_t^d}\mid \mathcal{F}_t\right)\ &= E_d^T\left(X_T\mid \mathcal{F}_t\right).\tag{2} \end{align*} $$ 即是即期匯率在到期時的預期 $ T $ , 在下面 $ T $ 前向測量而不是風險中性測量。 返回公式 $ (1) $ . 讓 $ Q_f $ 是外國風險中性措施和 $ E_f $ 是對應的期望運算元。那麼,對於 $ t\ge 0 $ ,
$$ \begin{align*} \frac{dQ_d}{dQ_f}\big|_t = \frac{B_t^d X_0}{B_t^f X_t}. \end{align*} $$ 而且, $$ \begin{align*} F^X(t,T) &= \frac{1}{B_d(t, T)}B_t^d E_d\left(\frac{X_T}{B_T^d} \mid \mathcal{F}_t\right)\ &=\frac{1}{B_d(t, T)}B_t^d E_f\left(\frac{X_T}{B_T^d} \frac{\frac{dQ_d}{dQ_f}\big|_T}{\frac{dQ_d}{dQ_f}\big|_t}\mid \mathcal{F}_t \right)\ &=\frac{1}{B_d(t, T)}B_t^d E_f\left(\frac{X_T}{B_T^d} \frac{B_T^d}{B_T^f X_T} \frac{B_t^fX_t}{B_t^d}\mid \mathcal{F}_t\right)\ &=\frac{X_t}{B_d(t, T)}E_f\left(\frac{B_t^f}{B_T^f}\mid \mathcal{F}_t \right)\ &=X_t\frac{B_f(t, T)}{B_d(t, T)}. \end{align*} $$
附加資訊。
結合公式 $ (2) $ ,
$$ \begin{align*} E_d^T(X_T \mid \mathcal{F}_t) &= E_d^T\left(X_T\frac{B^f(T, T)}{B_d(T, T)} \mid \mathcal{F}_t\right)\ &=X_t\frac{B_f(t, T)}{B_d(t, T)}. \end{align*} $$ 即遠期匯率過程 $ \left{X_t\frac{B_f(t, T)}{B_d(t, T)}, 0 \le t \le T \right} $ 是國內下的鞅 $ T $ -前向措施。