Hi, This question deals with re-pricing a term assurance product and considering adding a mortality option - renewing at the date of maturity without extra underwriting. The first question asks for advantages of using a cashflow approach, second asks for a detailed explanation of how to deterministically price the product. The third question asks: 'Discuss whether it would be appropriate to use a stochastic approach to reprice this product.' One of the paragraphs in the answer reads: 'There are no embedded derivatives to consider in this contract. Stochastic projections might be useful to determine the full range of possible mortality rates that might occur at the date of conversion, but it is unlikely these would give greatly different results than scenario testing.' What does this mean? Is it suggesting that when considering different mortality rates, the premiums obtained would be similar whether stochastic simulation was done or using a number of deterministic scenario tests? I would suggest that stochastic simulation would be very useful due to the relationship between % uptake and mortality of people that uptake. Would this not be a crucial assumption that would justify stochastic simulation? Thanks in advance for any responses. Andrew
Hi Andrew I think you're absolutely right that there's a link between take-up rate and mortality assumption that should be reflected somehow. The choice is then whether to do this via a deterministic approach (and specify deterministically how the 2 are related in different scenarios) or do it stochastically (which might be worth it, if felt could specify the correlation with some confidence). I read the paragraph you quote from the examiners' report as saying that if there are embedded derivatives in a contract, this might help tip the balance in favour of a stochastic approach. An "embedded derivative" means that some of the cashflows are the same as under a derivative. This most often happens in life insurance in relation to investment guarantees. The idea is that these are best valued stochastically, to reflect their time value as well as intrinsic value. As there's no embedded derivative here, no need to use a stochastic approach for this reason, so perhaps deterministic scenarios give enough info Best wishes for Monday Lynn
