Fall 2013 (Q22 part a. sample answer1) vs Fall 2017 (Q25 sample answer 1)

keshav1996

Hi,

In the Fall 2013 question part a., irrespective of the fact that it's a data reconciliation ques, I see that they apply stop loss limit to the Net of XoL triangle cells only.

So, while coming from 2013 questions to the Fall 2017 questions during my practice, I follow the same approach (i.e applying stop loss limit in the relevant years) to the net of XoL triangle cells and then develop each year to ultimate, I see that my answer is wrong.

I am not able to imagine what's the correct approach. I mean if I tweak the numbers in Fall 2013 question and come up with very big ultimates by developing the net of xol triangle and then apply stop loss limit (like its done in 2017 question), then, unlike the sample answer conclusion there, I can conclude that at least the stop loss treaty is working fine.

Side note: I know that my 2016 estimate in Fall 2017 ques would be wrong due to no stop loss limit given, but the other years have the same stop loss limit of 5000 like in Fall 2013 question.

Graham

Strictly speaking, this problem cannot be solved with the information given. Part of the reason is as follows:

• Suppose we have a claim reserved at 90,000 but the excess-of-loss treaty caps individual claims at 100,000. (Note that we are not given the excess-of-loss attachment point.)
• If we develop this claim for 90,000, it could end up having an estimated ultimate of 110,000 but we have no way of knowing that because we don't have information on individual claims. This claim should be capped at 100,000 so we would be overestimating.

In general, when reinsurance treaties are involved, we won't be able to get a totally accurate estimate of net ultimate before all claims are settled. Unless of course we're given information on individual claims and details of all reinsurance treaties.

Note that the 2017 problem accepted 2 different answers, neither of which are absolutely correct, but that's the best you can do with the information you're given. Your method may be just as valid as the sample answers in the examiner's report.