Let $X_t$ follows a Ornstein-Uhlenbeck process

$dX_t = -\theta X_t dt + dW_t$

for $\theta < 1$ and $X_0 = 0$.

I am interested in the stochastic integral

$Y_T = \int_0^T X_s dX_s$ and I wonder if what I do to solve it is the right approach. I am asking because I am aware of the many conditions under which solution to stochastic integral exist, but haven't studied those in detail.

I recall the solution to

$Z_T = \int_0^T W_s dW_s$ is $\frac{1}{2} (W_T^2 - [W, W]_T)$ = $\frac{1}{2} (W_T^2 - T)$, which follows a gamma distribution.

If I use the same trick for the case of OU process above, I get

$Y_T = \frac{1}{2}(X_T^2 - [X,X]_T) = \frac{1}{2}(X_T^2 - T)$, where we know $X_T \sim N(0, \frac{1}{2\theta}(1-e^{2\theta T}))$, so $X_T^2$ also follows a gamma distribution.

However, I feel somewhat unsettled with this solution. It is with the fact that $W_s$ and $dW_s$ are independent but $X_s$ and $dX_s$ are not independent, and somehow I think that would make the solution of $Y_T$ different from $Z_T$ (in some profound way.)

My question is, what is the solution of $Y_T$ and does the dependence of increments affect the stochastic integral?


Note that \begin{align*} d\left(X_t^2\right) &= 2X_t dX_t + d\langle X, X\rangle_t \\ &= 2X_t dX_t + dt. \end{align*} Then \begin{align*} \int_0^T X_s\, dX_s = \frac{1}{2}\left(X_T^2 -T\right). \end{align*}

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.