Let $(W_t)_{0\leq t\leq 1}$ be a standard Wiener process on $[0,1]$, and let $\mathcal{F}_t$ be the natural filtration. Consider a BSDE $$ dX_t=f(t,X_t)dt+\sigma(t,X_t) dW_t $$ with terminal condition $X_1=x$, where $f(t,\cdot)$ and $\sigma(t, \cdot)$ are $\mathcal{F}_t$-adapted square integrable processes.

My question: is it possible for the BSDE to be well defined if $\sigma(t,X)=0$ for all $t\in [0,1]$ and all $X$? Also, it seems unlikely to me that I can treat such a case as an ODE since reversing time would screw with the progressive measurability of $f$. Am I wrong?

  • $\begingroup$ You can treat it as an ODE. Since the resulting function is a deterministic function of time, progressive measurability amounts to the function itself being Borel measurable. $\endgroup$ Jun 30, 2020 at 6:29
  • $\begingroup$ Sorry, I was imagining that the drift is not necessarily deterministic. I will edit my question accordingly. $\endgroup$
    – tsm
    Jun 30, 2020 at 18:09

1 Answer 1


I am not sure if I understand your question correctly. A typical Brownian BSDE has the form

$$dY_t = f(\omega, t, Y_t, Z_t)dt - Z_t dW_t$$

with terminal condition

$$Y_T = \xi \in \mathcal{F}^{W}_T$$

where $Y$ and $Z$ are two parts of the solution and required to be adapted to the Brownian filtration. If your question boils down to if there is a BSDE with deterministic terminal condition $\xi = x$ and (second part of the) solution process constant zero, the answer is yes. E.g., take

$$f(\omega, t, Y_t, Z_t) = W_{\frac{T}{2}}1_{(\frac{T}{2},\frac{3T}{4}]} - W_{\frac{T}{2}}1_{(\frac{3T}{4},T]}$$

and $x=0$.


Your Answer

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

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