Law of total variance on function of random variables

90 Views Asked by At

Let's take the following random variables $X,Y,Z_1,...,Z_K$, on the same probability space (or associated with the same experiment), where we know that $$ Y = \sum_{k=1}^K\beta_kZ_k$$

Let's say I am interested in computing the conditional variance of $X$ knowing $Y$, similarly as in this question. Applying the law of total variance, we get:

$$ V[X|Y]=V[E[X|Z_1,...,Z_K,Y]|Y]+E[V[X|Z_1,...,Z_K,Y]|Y] $$

In this special case, since when we know $Z_1,...,Z_K$, we also know $Y$, can we reduce the law of total variance to:

$$ V[X|Y]=V[E[X|Z_1,...,Z_K]|Y]+E[V[X|Z_1,...,Z_K]|Y] $$

Any help or suggestions would be much appreciated.