Suppose that $(\Omega,\Sigma,\mathbb P)$ is a probability space and let $Y$ and $Z$ be two real-valued, $\Sigma$-to-Borel measurable functions on this space.

Suppose that the distribution of $Y$ conditional on $Z$ satisfies $$\mathbb P(Y\leq y\lvert\rvert Z)=F(y-Z)\quad\text{almost surely, for each $y\in\mathbb R$},\tag{$\clubsuit$}$$ where $F:\mathbb R\to[0,1]$ is some non-decreasing (a fortiory Borel-measurable) distribution function.

The form of the conditional probabilities ($\clubsuit$) leads me to formulate the following two conjectures, which I have had a hard time proving:

Conjecture 1: The unconditional distribution of $Y-Z$ satisfies \begin{align*} \mathbb P(Y-Z\leq y)=F(y)\quad\text{for each $y\in\mathbb R$}; \end{align*}


Conjecture 2: The variables $Y-Z$ and $Z$ are independent.

Any hints would be greatly appreciated.


(A) Let $G$ be the distribution function of $Z$. Then, $$P(Y-Z\leq y)=\int P(Y\leq z+y|Z=z)dG(z)=\int F(y-z+z)dG(z)=F(y)\int dG(z)=F(y)$$

(B) $P[Y-Z\leq y|Z]=P[Y\leq Z+y|Z]=F(Z+y-Z)=F(y)$ independent of $Z$. Hence independence follows.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.