Loss development using credibility

[teddybear]

All 10, volume 1b: Page 44

Regarding the Negative Binomial-Binomial model:

It says that E(Y)=r(1-p)/p. But as far as I can remember about negative binomial, E(Y) =r/p, did I miss something?


Thanks

[Tex Act]

There are different ways to define the parameters in the neg. bin.

Just going from memory, it seems like it's whether you think of the probability of loss as p or 1/p.

See if that works out. You should see that r(1-p)/p is equivalent to r/p.

[teddybear]

Say p=0.8, 1-p=0.2

P/(1-P)=4
1/p=1.25