I always find these questions tricky-I realise they should be straight forward but I feel the the solutions can vary a lot and I feel sometimes I'm wasting my time mentioning things like,consider time period and projection frequency. Sometimes this gets mentioned,other times it doesn't. What I would like is tips on how i should answer these questions. Linked to the above-I would have thought that the solutions for assessing the cost of a maturity g'tee and determining the charges for a maturity g'tee should be the same except for the last few steps but this isn't the case in the modelling revision booklet. Emphasis is put on different things which is why I get confused as to what is required to be said. Also, are two sets of simulations required for assessing the cost or just one? I.e can you project the unit fund to maturity,check what the cost is in one,then simulate lots of times to get a distn of costs? The solution implies two separate sets of simulations which is confusing me. If I had work experience in modelling I think it would help!!! (questions- st2 april 2006 qu1 and 302 April 2003 qu 8).
Hi - I can't really help too much with your first query but rest assured that alot of people will feel this way, not just you! You're right about the cost of the guarantee - there is only one set of simulations. These will drive the unit growth and everything dependent on it like guarantee costs, charges, dynamic policyholder behaviour, expense inflation etc. The average of the present value of the guarantee cost gives a time 0 cost - but the company will most likely want to add this onto the regular amc - therefore further calculations would be needed to turn this into an equivalent fund-related charge.
Hi I agree, there's only one set of simulations in costing the guarantee. Something I find it helpful to look out for in modelling questions is whether the question is essentially "describe the features/requirements of a good model" or else "describe how you would use a model to.....". These two types of question have slightly different answers. (For example, more likely to refer to time period/projection frequency for the first rather than the second, as this is what the Core Reading does.) Also, there can be subtle differences in the info the question provides - for example the April 2003 question you mention doesn't say whether to use a deterministic or a stochastic approach (so the answer covers this choice) whereas the April 2006 question specifies what type of model to use. Not sure there are any absolutely fail-safe rules though Best of luck for the exams Lynn
