303, April 2004, Q2 - RI i) the question asks for the "functionality of the model" - its very easy to interpret this as the requirements of a good model (valid, adequately documented .. ) isnt it? the solution on the other hand seems to present the procedure for actually doing the model (start with claims... simulate...) rather than functionality! In any case, the solution states: multiple simulations, different scenarios, and varying RI structures. What would count as a simulation and scenario here? I thought - simulation would be 10,000 possibilities, picking from the distribution - scenario would be different RI structures but that invalidates the need for "vary RI structures" in the solution. What does it mean by scenario here?
I disagree Gijap. The solution gives a pretty clear description of the requirements for this particular model. This sounds fine to me.
thanks Katherine.. But if we are correct, then the solution doesnt need a separate point "vary RI structure" afterwards? The solution is suggesting there are 3 separate things to vary: 1) simulations (10,000 etc) 2) ? 3) ? I only see two - simulations and RI structure. Am I missing something?
I'm really struggling to see what you're looking at here Gijap. The points you mention aren't in the Examiners' Report. Perhaps you looking at the fuller description in our Revision Notes? In any case, the general principle remains the same: Simulate claims and feed them through a trial reinsurance structure. It's important at this stage not to look at just the mean net result, but also the net result for the extreme scenarios, ie the scenarios when reinsurance recoveries will be most critical. Repeat for different trial reinsurance structures, and then choose the optimal one.
