Let $\Phi(x)$ be a CDF of standard normal distribution and $\Phi^{-1}(p),\,p\in(0,1)$ its inverse.

It is evident that $$ \mathbb{P}(X<\Phi^{-1}(p))=\Phi(\Phi^{-1}(p))=p, $$ where $X\sim N(0,1)$.

Is there possible to get any simplified expression of $$ \mathbb{P}(X<\Phi^{-1}(p),\,Y<\Phi^{-1}(p))= \int_{-\infty}^{\Phi^{-1}(p)}\varphi(y)\Phi\left(\frac{-\varrho y+\Phi^{-1}(p)}{\sqrt{1-\varrho^2}}\right)\,dy, $$ where $X,\,Y\sim N(0,1)$, $\textrm{cov}(X,Y)=\varrho$ and $\varphi(y)=\frac{1}{\sqrt{2\pi}}e^{-y^2/2}$?


1 Answer 1


If $\varrho$ is not zero there isn't much you can do. If you have an implementation of the bi-variate normal CDF $\Phi_2(x,y,\varrho)$ you can "simplify" it to $$ \mathbb P\Big(X<\Phi^{-1}(p),Y<\Phi^{-1}(p)\Big)=\Phi_2\Big(\Phi^{-1}(p),\Phi^{-1}(p),\varrho\Big)\,. $$ This RHS is known as Gauss copula. If $p=1$ and $\Phi^{-1}(p)=+\infty$ the RHS is one.

Needless to say that for $\varrho=0$ the RHS is $\Phi(\Phi^{-1}(p))\,\Phi(\Phi^{-1}(p))=p^2\,.$


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.