Exponential of martingale is also martingale?

420 Views Asked by At

Let $X_t$ be a standard Brownian. Put $Y_t := \exp (\int _0^t udX_u)$ and determine, whether $Y$ is a martingale.

We know $M_t :=\int _0^t u dX_u$ is a martingale, therefore $\langle M \rangle _t = \int _0^t u^2 du $. By Ito's formula we have $$ Y_t = 1 + \int _0^t Y_u dM_u + \frac{1}{2} \int _0^t Y_u d\langle M \rangle _u, $$ where the $dM_u$ term is a martingale. Therefore, $Y$ is martingale if and only if $$ E \left ( \int _0^t Y_u d\langle M \rangle _u \ \Bigg\vert\ s \right ) = 0 \tag{1} $$ for every $s< t$ and this is where I'm stuck. How do we make sense of quantities like (1) and why (not) it would be zero?

1

There are 1 best solutions below

0
On BEST ANSWER

Well unfortunately it is never the case. This comes from the fact that $\int _0^t u dX_u$ follows a Gaussian law (as this is a Wiener integral) and so $Y_t$ is a log-normal random variable for all $t$ with expectation equal to $e^{\mu_t+\sigma_t^2/2}$ where $\mu_t=0=E[\int _0^t u dX_u]$ and $\sigma_t^2=E[(\int _0^t u dX_u)^2]=\int_0^t u^2du=t^3/3$ so the expectation of $Y_t$ is a strictly increasing function of time and it's not a martingale.