2
$\begingroup$

I'm trying to prove two statements:

1.) Suppose $X,Y$ are random variables on $(\Omega,\mathcal F, P)$ such that $P(X \leq Y)=1$. Show that $F_Y(x) \leq F_X(x)$ for every $x \in \mathbb R$. (Show that $F_X$ is stochastically smaller than $F_Y$.) Here $F_X=P(X \leq x)$ is the distribution function of X.

2.) Suppose $F$ is stochastically smaller than $G$. Show that there exist random variables $X,Y$ on some probability space $(\Omega,\mathcal F, P)$such that $F_X = F, F_Y =G$, and $P(X \leq Y)=1$.

Try at 1.): For any $x \in \mathbb R$, let $A_1$ and $A_2$ sets of $\Omega$ such that: $$A_1= \{\omega: X(\omega) \le x\}, A_2 = \{\omega: Y(\omega) \le x\}.$$ Now, $$P(X \le x)=P(\{\omega: X(\omega) \le x\}) = F_X(x)$$ and $$P(Y \le x)=P(\{\omega: Y(\omega) \le x\}) = F_Y(x)$$ by defintion. Then since $P(X\leq Y)=1$, we can write $X(\omega)\leq Y(\omega)$ for every $\omega \in \Omega$, thus $X(\omega)\leq Y(\omega)\leq x$. It follows that $A_2\subset A_1$, and $$F_Y(x)=P(Y\leq x)=P(A_2)\leq P(A_1)=P(X\leq x)=F_X(x)$$ by monotonicity of probability measures, thus $F_X$ is stochastically smaller than $F_Y$. As I was asking I actually found my proof to be very similar to the OP's proof here. I just wanted to check if the proofs were ok.

Try at 2.): Since $F,G$ are distribution functions, there exists a probability space $(\Omega,\mathcal F, P)$, and random variables $X:\Omega \to \mathbb R, Y:\Omega \to \mathbb R$ such that $F=F_X,G=F_Y$. But I'm not really sure where to go from this point on.

Any advice or help would be appreciated!

$\endgroup$
2
$\begingroup$

1) is OK. For 2) there is a standard construction. Consider the probability space $(0,1)$ with Borel sets and Lebesgue measure. Let $X(\omega)=\inf \{t: F(t) \geq \omega \}$ and $Y(\omega)=\inf \{t: G(t) \geq \omega \}$. It is trivial to check that $X(\omega) \leq Y(\omega)$ for all $\omega$. To show that the distribution of $X$ is $F$ show that the statements $X(\omega) \leq t$ and $F(t) \geq \omega$ are equivalent. [You will need right continuity of $F$ for this]. Taking probability of $\{\omega: X(\omega) \leq t\}$ you will see that $X$ indeed has distribution $F$. Similarly, $Y$ has distribution $F$.

$\endgroup$
  • $\begingroup$ "To show that the distribution of $X$ is $F$ show that the statements $X(\omega) \leq t$ and $F(t) \geq \omega$ are equivalent." Why is this so? $\endgroup$ – Sank Sep 30 '18 at 1:59
  • $\begingroup$ Also would it make more sense to use $X(\omega)$=$sup${$t:F(t)<\omega$}? $\endgroup$ – Sank Sep 30 '18 at 2:20
  • $\begingroup$ I guess this would give you a left continuous function instead of a right continuous one. $\endgroup$ – Kabo Murphy Sep 30 '18 at 4:16
  • $\begingroup$ Would it matter which one I use? $\endgroup$ – Sank Sep 30 '18 at 5:18
  • $\begingroup$ It depends how distribution functions are defined. Most books define them to be right continuous. $\endgroup$ – Kabo Murphy Sep 30 '18 at 11:31

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.