We say that ζ is a p-percentile of the distribution F if F(ζ) = p. Show that if ζ is a p-percentile of the IFRA distribution F, then
\bar{F} (x)= ≤ e^{−θx}, x ≥ζ\\ \bar{F} (x) ≥ e^{−θx}, x ≤ζwhere
θ = \frac{ −log(1 −p)}{ζ}For x \geqslant \xi,
1-p=\bar{F}(\xi)=\bar{F}(x(\xi / x)) \geqslant[\bar{F}(x)]^{\xi / x}
since IFRA. Hence, \bar{F}(x) \leqslant(1-p)^{x / \xi}=e^{-\theta x}.
For x \leqslant \xi,
\bar{F}(x)=\bar{F}(\xi(x / \xi)) \geqslant[\bar{F}(\xi)]^{x / \xi}
since IFRA. Hence, \bar{F}(x) \geqslant(1-p)^{x / \xi}=e^{-\theta x}.