What is excess distribution function and mean excess function? How to calculate them?

999 Views Asked by At

I was watching a YouTube video about finance and in the video, the lecturer mentioned excess distribution function and mean excess function.

\begin{align} F_u(x)&=P(X\le u+x|X>u)\\ e(u)&=E(X-u|X>u) \end{align}

where $X$ is a random variable with distribution function $F(x)=P(X\le x)$.

However, I don't really know how to compute these quantities. My background is computer science, and I'm doing some programming about finance. What I've searched and learned is that,

  1. Let's say random variable $X$ follows a uniform distribution on [0, 1], then the $x_F$, which is the upper end-point of $X$, equals to 1.

  2. The Wikipedia page about Pickands–Balkema–de Haan theorem gives a formula, $$F_u(y)=P(X-u\le y|X>u)=\frac{F(u+y)-F(u)}{1-F(u)}$$

  3. The cumulative distribution function for standard uniform random variable $X$ is $F(x)=x$

  4. For mean excess function, I found some materials on this page, on hu-berlin.de, $$e(x)=E(X-x|X>x)=\frac{\int_x^\infty(1-F(u))du}{1-F(x)}$$

Does 2. 3. and 4. suggest that the $F_u(x)$ and $e(u)$ can be computed as the following?

\begin{align} F_u(x)&=\frac{F(u+x)-F(u)}{1-F(u)}=\frac{u+x-u}{1-u}=\frac{x}{1-u}\\ e(u)&=\frac{\int_x^\infty(1-F(u))du}{1-F(x)}=\frac{\int_x^\infty(1-u)du}{1-x} \end{align}

And if we let $u\rightarrow x_F$, then is the following right?

\begin{align} \lim_{u\rightarrow x_F}F_u(x)&=\frac{x}{1-u}=\frac{x}{1-1}=\infty\\ \lim_{u\rightarrow x_F}e(u)&=\frac{\int_x^\infty(1-u)du}{1-x}=\frac{\int_x^\infty(1-1)du}{1-x}=0 \end{align}

I'm now totally lost, could anyone shed some light on this?

Update 1:

\begin{align} F_u(x)&=\frac{F(u+x)-F(u)}{1-F(u)}\\ &\not=\frac{u+x-u}{1-u} \end{align}

$F(u)$ can be approximated by $\frac{n-N_u}{n}$