Mango

[jkoch]

I've tried to read Mango three times, and never really get anywhere. All I picked up so far is the formulas for the MV and MS methods. Can anyone offer advice on how to approach this reading?

[Da Bears!]

Just learn the formulas for the various methods and understand why each overestimates or underestimates

[Jon Fox]

This is a neat paper that boils down to the question - how do you allocate risk load between risks in a portfolio at policy inception and then at renewal, taking into account the variance each component contributes.

There are two important math calculations you should be able to do:
1. The Renewal risk Load calculation
2. The Build up risk load calculation (new marginal risk)

You should be able to do these calculations using the following methods:
1. Marginal Variance
2. Marginal Surplus
3. Shapely Method

The fourth method presented in the paper - Generalizing Covariance Sharing requires event level data from a simulation as I recall; it would be more difficult to create a mathamatical calculation around this concept that could be answered in 5-10 minutes for an exam. Hence I don't think it as as important to be able to reproduce that calcualtion.

If you can step through the above calculations which are shown in the Appendices and know the main concepts (what is and isn't renewal additive and why), you should be good to go.

[JollyGoodFCAS]

There are some similarities with the Bault paper, so be sure to compare the two.

[cohomology]

1. The Renewal risk Load calculation
There are some similarities with the Bault paper, so be sure to compare the two.

Bault just talks about new policies and complains that the order matters (then he rips Feldblums CAPM method amoung others), but Mango finds a key problem: the fact that once policies are renewed then the total risk loads are more then the total variance of the combined portfolio.

This was actually the first paper I read (because the name "game theory" and if you took operations research Shapley value is well known) but I think its easier to start with Feldblums backround on risk loads.