Percentiles of sums of random variables

146 Views Asked by At

Heads up, I'm asking this question after having read Percentile of sum of random variables and other questions. Please don't close this as a duplicate. It's not the same question.

Suppose I have two positive random variables $X_1$ and $X_2$, and the following percentiles or upper bounds on them. The joint probability assumes independence.

\begin{aligned} P(X_1<35) &= 0.95 \\ P(X_2<40) &= 0.95 \\ P(X_1<35, X_2<40) &= 0.95^2 \end{aligned}

$X_1$ is concentrated beneath $35$, and $X_2$ beneath $40$. They're both under their respective limits with probability $0.0925$. That means that their sum can not reach $\ge70$ with $p=0.0925$, right? How can I prove this? I have a feeling I'm wrong. I know I'm skipping steps, yet don't know the steps.

$$P(X_1+X_2<35+40) = 0.95^2$$

1

There are 1 best solutions below

1
On BEST ANSWER

No. You can't. Because it is not true.

Because it is possible to get less than 75 in other ways also. For example, 10+65 = 75.

$$P(X_1+X_2<35+40) = 0.95^2 $$ $$\implies P(X_1+X_2<35+40 \land X_1<35 \land X_2<40) $$$$+ P(X_1+X_2<35+40 \land X_1<35 \land X_2\ge40) $$$$+ P(X_1+X_2<35+40 \land X_1\ge35 \land X_2<40) $$$$+ P(X_1+X_2<35+40 \land X_1\ge35 \land X_2\ge40) $$$$= 0.95^2$$

In this, the fourth term is obviously 0.

In the fist term, $X_1<35 \land X_2<40 \implies X_1+X_2<35+40$. So the first term becomes $P(X_1<35 \land X_2<40)$ which is nothing but $ 0.95^2$.

Substituting the first and fourth term, we get $$ 0.95^2 + P(X_1+X_2<35+40 \land X_1<35 \land X_2\ge40) $$$$+ P(X_1+X_2<35+40 \land X_1\ge35 \land X_2<40) +0 $$$$ = 0.95^2 $$

Which means $$ P(X_1+X_2<35+40 \land X_1<35 \land X_2\ge40) + P(X_1+X_2<35+40 \land X_1\ge35 \land X_2<40) =0 $$

Which obviously not true. What could be true is $$P(X_1+X_2<35+40) > 0.95^2 $$