I have been reading a finance paper that claims that the following function, which is a value for a financial derivative (1): $$V(s,t)=E_{Q} \left[\zeta\big(S(T)\big)e^{-\int_t^T r_F(\nu) d\nu}\\+\int_t^T C\big(S(u),u\big) e^{-\int_t^u r_F(\nu) d\nu}\big(r_F (u)-r_C (u)\big)du\ \middle\vert\ S(t)=s \right]$$

can be rearranged to be expressed as (2):

$$V(s,t)=E_{Q}\left[\zeta\big(S(T)\big)e^{-\int_t^T r_C(\nu) d\nu} \\-\int_t^T\big(V(S(u),u)-C(S(u),u)\big) e^{-\int_t^u r_C(\nu) d\nu}\big(r_F (u)-r_C (u)\big)du\ \middle\vert\ S(t)=s \right]$$

where $S(t)$ (which is the price of the underlying - eg a stock) follows a Wiener process with SDE:

$$\frac{dS(t)}{S(t)} =\big(r_R(t)-r_D(t)\big) dt+\sigma_S(t)dW_Q(t)$$

where $W_Q(t)$ is a Brownian Motion under measure $Q$.

The representation (1) is obtained by using the Feynman-Kac technique to solve the following PDE: $$ \frac{\partial V(S,t)}{\partial t}+\big(r_R(t)-r_D(t)\big)\frac{\partial V(S,t)}{\partial S}S(t)+\frac{\sigma_S(t)^2}{2} S(t)^2 \frac{\partial ^2 V(S,t)}{\partial S^2}\\=r_F V(S,t)-\big(r_F(t)-r_C(t)\big)C(S,t) $$

with the boundary condition: $$V(S(T),T)=\zeta\big(S(T)\big)$$

I can see no way to derive (2) from (1). I have tried several, but those that worked proved to have invalid steps. The paper itself just presents (2) as a fait accompli, with no attempt to justify it. On an intuitive level, (2) looks plausible, and seems like the sort of thing that may well be true. But a vague sense of plausibility is not the same as a rigorous demonstration of equivalence.

I am still working away at it, to see if it can be justified, but I was hoping that somebody might have seen something similar and be able to suggest a derivation, or alternatively point to a counterexample that invalidates the claimed equivalence.

I have the feeling that I may be overlooking something obvious, but I can't think what it is.

Thank you for any suggestions or comments anybody can make.

Background: the formula is intended as an adjustment to the classical derivative valuation formula, allowing for the trader's funding costs. $r_C(t), r_R(t),r_F(t)$ are respectively the interest rates for risk-free, repo and unsecured borrowing at time $t$ and $r_D(t)$ is the dividend rate. $C(s,t)$ is the collateral posted under the derivative agreement. $\zeta$ is the option payoff function and $V$ is the derivative value.


Your Answer

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

Browse other questions tagged or ask your own question.