Following Expected Value of Maximum of Two Lognormal Random Variables . I made an additional assumption, that $X,Y$ are jointly normal with parameters given above.

I tried to derive the simmilar result by first finding the density function of $max(X,Y)$, and I have the following CDF: $P(\max(X,Y) < m ) = P(X < m, Y < m) = \Phi(\frac{(\log(m) - \mu)} {\sigma}, \frac{(\log(m) - \nu)} {\tau})$, where $\Phi$ is the CDF of multivariate standard normal.

Then, I computed the density function which is $\phi_{max} = \frac{1}{\sigma \times m}\phi_1(\frac{(\log(m) - \mu)} {\sigma}) + \frac{1}{\tau \times m} \phi_2(\frac{(\log(m) - \nu)} {\tau})$, where $\phi_1$ and $\phi_2$ are marginal distribution functions of joint multivariate normal, which in turn are just densities of related normal.

Since I know, the distribution function at each $m$, then I can compute its expectation. I form $E(max(X,Y)) = \int_0^{\infty}(m \times \phi_{max} dm) = \int_0^{\infty} m \times \frac{1}{\sigma \times m}\phi_1(\frac{(\log(m) - \mu)} {\sigma}) dm + \int_0^{\infty} m \times \frac{1}{\tau \times m} \phi_2(\frac{(\log(m) - \nu)} {\tau}) dm = \int_0^{\infty} \frac{1}{\sigma}\phi_1(\frac{(\log(m) - \mu)} {\sigma}) dm + \int_0^{\infty} \frac{1}{\tau} \phi_2(\frac{(\log(m) - \nu)} {\tau}) dm = \exp(\mu + \frac{\sigma^2}{2}) + \exp(\nu + \frac{\tau^2}{2})$. Seemingly, I have made mistake when considering marginal distributions but cannot see where exactly.

Where I have got a flaw in my derivations?


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.