what is $E(X\mid Y)-E(X)$?

86 Views Asked by At

I do not have a background of Maths, but I met an equation when I read a paper,

$$E(X\mid Y)-E(X)=(Y-E(Y))\frac{\operatorname{cov}(X,Y)}{E(Y^2)}$$

Could anyone tell me how to prove this? I have tried a lot but failed...


Sorry that, I did not put all the original info from the paper.

This is the original context in the paper "Endogenous versus exogenous shocks in systems with memory":

"To quantify the response in such case, we recall a standard result of stochastic processes with finite variance and covariance that the expectation of some process $X(t)$ conditioned on some variable $Y$ taking a specific value $A_0$ is given by [22]

$$E[X(t)\mid Y=A_0]-E[X(t)]=(A_0-E[Y])\frac{\operatorname{Cov}(X(t),Y)}{E[Y^2]}$$"

Citation[22] is the book "Limit Theorems for Stochastic Processes" which is too huge for me to find this out.

1

There are 1 best solutions below

2
On

It's certainly not true in general. Some context might help. Where did you get this?

EDIT: For example, consider a case where $X = Y^2$, $E[Y] = 0$, $E[Y^2] = 1$. Then the equation says $$ A_0^2 - 1 = A_0 E[Y^3]$$ which is true for at most two possible values of $A_0$.

Perhaps this is supposed to be for Gaussian processes?