Independent solution

How to solve this Normal Distribution question

Setup

Setup

Let d be the deductible and q the requested percentile among losses above d.

d=15000,zd=15000200004500=1.1111111d=15000,\qquad z_d=\frac{15000-20000}{4500}=-1.1111111
FX(d)=Φ(zd)=0.1332603F_X(d)=\Phi(z_d)=0.1332603

Model

Model

Translate the conditional percentile equation into an unconditional cumulative probability.

Pr(XqX>d)=FX(q)FX(d)1FX(d)=0.95\Pr(X\le q\mid X>d)=\frac{F_X(q)-F_X(d)}{1-F_X(d)}=0.95
FX(q)=FX(d)+0.95(1FX(d))F_X(q)=F_X(d)+0.95\left(1-F_X(d)\right)

Compute

Compute

Evaluate the adjusted normal percentile and transform back to the loss scale.

FX(q)=0.9566630F_X(q)=0.9566630
zq=Φ1(0.9566630)=1.7132097z_q=\Phi^{-1}(0.9566630)=1.7132097
q=20000+4500(1.7132097)=27709.44q=20000+4500(1.7132097)=27709.44

Answer

Answer

Rounded to the nearest hundred, the conditional percentile is 27,700.

27700(B)\boxed{27700\quad\text{(B)}}