2
$\begingroup$

Let $\{W_t,t \geq 0\}$ be a standard Brownian motion under $\mathbb P$. Let $T_a$ be the hitting time of level $a$, that is: $$T_a= \text{inf}\{t \geq 0:W_t=a\}.$$ From a proposition, we know that $$\mathbb E[\exp(-\theta T_a)] = \exp(-a \sqrt{2\theta}).$$

How do we make use of the above preposition to calculate $\mathbb E[T_a]$?

I am supposed to obtain $\infty$ as the expectation.

$\endgroup$
2
  • $\begingroup$ Are you sure this isn't the characteristic function, and so should be $E[\exp(-i\theta T_a)]$ where $i$ is the imaginary unit? because the mgf only exists if all moments exist, and as @vinux's answer shows, using this to calculate the first moment doesn't work... $\endgroup$ Feb 21, 2012 at 12:33
  • $\begingroup$ @prob: $T_a$ is nonnegative almost surely, so for any $\theta \geq 0$, we have $\exp(-\theta T_a) \leq 1$ almost surely, so the expectation definitely exists (for such $\theta$)! $\endgroup$
    – cardinal
    Feb 21, 2012 at 12:41

1 Answer 1

2
$\begingroup$

This is similar to calculating expectation from M.G.F. Since $ e^x = 1 + x + {x^2 \over 2!} + {x^3 \over 3!} + {x^4 \over 4!} + \cdots. $ use differentiation technique for deriving expectation.

$$ -\dfrac{\partial}{\partial \theta} E[\exp(-\theta T_a)]_{\theta=0} =-\dfrac{\partial}{\partial \theta} \exp(-\sqrt{2\theta}) \Big \vert_{\theta=0} = E[T_a]$$

This works when the moments are finite. It seems the expectation is not finite.

$\endgroup$
6
  • $\begingroup$ Thank you for the respone. I do not understand the first part though. Could you kindly elaborate. $\endgroup$ Feb 21, 2012 at 5:42
  • $\begingroup$ Also How will i compute the partial differentiation with E(exp(-theta*Ta)) with E being a part of the equation. If you could provide me a head start on how to differentiate I would be able to proceed. $\endgroup$ Feb 21, 2012 at 5:44
  • $\begingroup$ I will modify my solution. $\endgroup$
    – vinux
    Feb 21, 2012 at 5:46
  • $\begingroup$ Hmm. Do you not need $\mathbb E(e^{−\theta T_a})$ to exist in an open interval around zero to apply such a result? I think you can get at it by simpler means. $\endgroup$
    – cardinal
    Feb 21, 2012 at 15:54
  • $\begingroup$ I agree cardinal. I initially wrote the approach. later I added the actual function and realised it is not that simple. $\endgroup$
    – vinux
    Feb 21, 2012 at 16:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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