Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Suppose we have the BS model with time depending coefficient. This means we have to processes $dX_t=rX_tdt$ and $dU_t=U_t(\mu(t) dt +\sigma(t) dW_t)$, where $ W$ is a Brwonian Motion. Moreover we define $Y_t:=\frac{U_t}{X_t}$. Both, $\mu,\sigma$ are continuous function from $[0,T]$ to the reals and $\sigma(t)>0,\forall t\in[0,T]$. Then I should find an equivalent probability measure $Q$ on $\mathcal{F}_T$ such that $Y$ is $Q$ martingale and satisfies the SDE $$dY_t=Y_t\sigma(t)dB_t$$ for a Brownian Motion $B$ under the measure $Q$.

Motivatet from the non time-depedning case, I defined $L_t:=-\int_0^t \lambda(s) dW_s$ and defined the measure $Z_T:=\frac{dQ}{dP}=\mathcal{E}(L)_T$. We have proved in the lecture that $\mathcal{E}(L)$ is a martingale, hence we have $E[Z_T]=E[Z_0]=1$. Since $Z_T$ is also positive, we can define a probability measure $Q$ equivalent to $P$ through this density. I was also able to prove, that $Y_t$ satisfies the SDE above under the new measure $Q$ with this Brownian Motion $B$. It is just applying Girsanov. However, I do not see why $Y$ should be a martingale under $Q$. The problem is, that $\sigma(t)$ appears in the SDE under $Q$. So I am not sure how $Y$ looks like under the measure $Q$. In the non-time depending case, we have that $Y$ has under $Q$ the form

$$Y_t=e^{(\sigma B_t-\frac{1}{2}\sigma^2 t)}$$

which is obviously a martingale. However, in this time depending I am not sure how to prove that $Y$ is a martingale.

Moreover, I want to prove the following inequality: Let $H=((U_T)^2-K)^+$. Then it should be true that

$$E_Q[\frac{H}{X_T}]\ge \big(s^2e^{(\int_0^T(\sigma(u)^2+r)du)}-Ke^{(-rT)}\big)^+$$

I am not sure if $s$ is a mistake in this inequality, since it does not appear so far in this exercise. However, I thought just proving the inequality, will show me what $s$ has to be. So far, I did the following:

$$E_Q[\frac{H}{X_T}]=e^{-rT}E_Q[((U_T)^2-K)^+]\ge e^{-rT}\big(e^{2rT}E_Q[Y_T^2]-K\big)^+=\big(e^{rT}E_Q[Y_T^2]-e^{-rT}K\big)^+$$

Hence, all I have to show is $e^{rT}E_Q[Y^2_T]=s^2e^{\int_0^T(\sigma(u)^2+r)du}$. Even more it is enough to prove $E_Q[Y^2_T]=e^{\int_0^T\sigma(u)^2du}$

It would be very helpful, if someone could explain the things I did not get.

share|cite|improve this question
Isn't $\sigma$ a deterministic function? Consult Section 4 of these notes – Ilya Jan 21 '13 at 12:57
@Ilya Ah nice, thanks! Now everything is working. So the $s$ should be $1$, right? If you want, you can just give a short answer. I will accept it. – user20869 Jan 21 '13 at 13:20
up vote 1 down vote accepted

For the change of measure of a diffusion with time-dependent drift and volatility you shall check Section 4 of these notes. There you also have a formula for $Y_T$ through $\displaystyle{\int_0^t}\sigma_s\mathrm dB_s$ which quickly leads you to the formula of the expectation.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.