CIA VTP2 - Selective Lapse Rates

[sophie]

Hi,

I'm doing some reserve auditing for a Canadian client. I work outside Canada and I'm still not familiar with the Canadian Standards of Practice.

Anyways, currently I am validating the reserve model that is done in AXIS. I am trying to replicate the computation of the valuation mortality rate with mortality deterioration due to selective lapse. I have found a worksheet in the GGY website that computes the final mortality rate with adjustments to mortality due to selective lapse. I tried to follow how it was done in the worksheet but the valuation mortality rate that I have computed differs slightly than the results in AXIS. Before this I also read the CIA VTP 2 paper to get a background on selective lapse and mortality deterioration.

The model has an underlying lapse table, selective lapse table, and total lapse table that was defined by the client. The reference mortality table for the underlying lapse rates is CIA 6975. From the way it was modeled, the total lapse rates = underlying lapse rates + selective lapse rates. It was also mentioned in the client's documentation that all additional lapses are selective.

I emailed GGY and asked some queries regarding the calculations, and they mentioned something that confused me: "The underlying lapse rate selected for the reference mortality table (and the only ones that should be used) should be the lapse experience of the data from which the reference mortality table was derived. The VTP2 paper describes the underlying lapse rates for the CIA 6975 table which are 15%, 12%, 9%, and 7% respectively during the first four policy years. Starting with policy year 5 the lapse rate is a level 5% every year." From the way I interpret it, that means that if I use the CIA 6975 reference mortality table, I should use 15, 12, 9, 7, 5 as underlying lapse rates. I saw those underlying lapse rates in the CIA VTP 2 paper (Appendix A), although I cannot find anywhere in the paper that says these rates are the required underlying lapse rates to be used if you use the CIA 6975 table. And since the client has modeled a different underlying lapse rate table, I am not sure if I should consider this as a model error or there is a leeway for the client to assign their own underlying lapse rates. If this is a model error, then maybe that is the reason why I am arriving at a different valuation mortality rate even if I follow GGY's worksheet.

Can anyone enligten me on this? Any help would be greatly appreciated!

Thanks so much!

[asdfasdf]

Are they using the CIA 6975 mortality table directly, or adjusting for their experience?

In general I would expect the lapse rates used to be best estimates (not fixed based on an industry table). I don't have a great understanding of the anti-selective lapse rules thought.

Sounds reasonable to get your client to explain more clearly what they are doing and why, and then with that you can assess for reasonableness.