Problems

 

When we talk about a maths tool kit, we need a tool to solve a problem. I thought it might be a good idea to try and get some clarity about what sort of problems we are talking about as it might provide some ideas as to what tools are required.

 

This is good point! Here are some examples:

 

• What methods are available to draw a smooth line through some data?

 

Hastie and Tibshirani (ISBN 0-412-34390-8) have a couple of good chapters on smoothing - Im guessing these have all been implemented in R - and would be interested in someone attaching a link to the relevant code.

 

• How do I allocate ICA values down to various classes of business?

 

Capital allocation is probably worth a page on its own

 

• What methods are available to determine the reserving uncertainty for XL claims, where no stable pattern exists (because of only a few claims each year)?

 

This paper by England & Verrall provides a good summary of methods. But these seem geared towards larger accounts with reasonably stable development patterns. Are there any methods for dealing with a handful of claims each year?

 

• What methods can actuaries use to generate random numbers that are suitable for actuarial applications?

 

This page starts the discussion.

 

• Others?

 

What about something on using weighted samples to make Stochastic models run quicker. I know the life industry use smaller numbers of simulations (e.g. 5000) but the input parameters are chosen carefully so that it is representative. How would you go about designing such a set?