Hello, I am really struggling to understand what is going on in this question, especially in part (a). The solutions doesn't really explain how it goes from one step to the next.
This is an unusual question on classical credibility theory. It is probably easier to attempt this question by doing parts (a) and (b) at the same time, which the Examiners commented was an equally valid approach. In summary, you need to assume a Poisson distribution for claims frequency, make some suitable assumptions for k and P (eg k=0.05 and P=0.95) and note that you do not have sufficient claims to assign full credibility to your data (which answers (a)) and then use the square root rule to answer (b).
Hi, I understand every part of the solution except the bit where it re-calculates the observed rate of claims per year. Given that the question said there's no more IBNR, there should be no more claims coming through in the next 3 months of the policy term no? Or is it because the policy is not fully earned yet, we assume claims could come through from the unexpired period? (i.e. 11 more claims to be precise: 120 minus 109) Thank you.
IBNR only covers claims which have been incurred, ie where the exposure has been earned, so you need to allow for claims from the unexpired period of exposure.