Consider the random variables $Y,X$. Can we always write $$ Y=E(Y|X)+\epsilon\tag 1 $$ with $\epsilon$ independent of $X$?

Note: from this answer here, we know that we can always write (1) with $E(\epsilon|X)=0$. However, here I am asking if we strengthen the relationship between $\epsilon $ and $X$ to stochastic independence.


1 Answer 1


Suppose $$ Y=X^2+u $$ where $u|X\sim(0,X^2)$ has conditional heteroskedasticity. Then, $$ \epsilon=Y-E(Y|X)=X^2+u-E(Y|X)=u, $$ which has conditional mean zero but is not independent of $X$, as its second moment depends on $X$.

  • 2
    $\begingroup$ I'm guessing there is an N missing in $u|X\sim(0,X^2)$ $\endgroup$
    – Stef
    Commented Dec 21, 2022 at 11:14
  • 2
    $\begingroup$ Not necessarily, the notation is to indicate that $u$ is, conditionally on $X$, distributed with mean 0 and variance $X^2$. It does not have to be a normal distribution. $\endgroup$ Commented Dec 22, 2022 at 7:45

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.