Re[2]: Study Manual 4b, Fall 1998

Cathy McEvoy ( Cathy_McEvoy@watsonwyatt.com )
Wed, 9 Sep 1998 09:37:21 -0400

I wonder if you could help. I'm having a mental block on decimal
factorials.

Specifically, I was doing a Pearson's Chi-square problem with a fitted
negative binomial. The K parameter was 1.76. So, how do I go about
solving for different values of x:

( x+k-1)
( ) p^k q^x ?
( x )

Thanks!
______________________________ Reply Separator _________________________________
Subject: RE: Study Manual 4b, Fall 1998
Author: RMcCollough@sfbcic.com (McCollough; Russell) at Internet
Date: 08/28/1998 12:48 PM

Dimitra,

The formula you are referring to is formula 8.3.12 on page 158 of
Hossack,Pollard, and Zehnwirth. However,
this formula gives the full credibility criterion for Pure Premiums. We are
interested in the full credibility criterion
for Frequency. For the Poisson case, this formula is N = (y/k)^2 , which is
at the bottom of page 158.

The confusion lies in the fact that the text refers to full credibility
criterion for Frequency as N0 and the study manual
refers to this same quantity as NF , which happens to be the notation that
Hossack, et al. uses for the full credibility
criterion for Pure Premium.

Russell McCollough
rmccollough@sfbcic.com

> -----Original Message-----
> From: Dimitra Roidakis [SMTP:dr@g-g-a.com]
> Sent: Thursday, August 27, 1998 1:43 PM
> To: studygroup4b@lists.casact.org
> Subject: Study Manual 4b, Fall 1998
>
> I have a question from the study manual, fall 1998.
>
> It's concerning full credibility and partial credibility.
> Page 77 question C14, and C15 deal with the same type of problem.
> My question is why do they use the following equation for the full
> credibility:
>
> Nf=(y/k)^2
> and they assume that sigma over m the coefficient of variation in
> both question is zero. For this to be true then P(Xi=d)=1 has to be
> true, but I don't think it is. Could it be a mistake? I just want to
> make sure.
>
> Thanks
> Dimitra
> dr@g-g-a.com