Calculating Pension Annuity Factors - Medium Cohort adjustments

"sylvian stone" wrote

Are you after the full double-entry table, or just one of the many single-entry tables which it is possible to construct?

"sylvian stone" wrote

I'm not aware of any - I think people just construct the particular projected table which they wish to use each time...

"sylvian stone" wrote

Can it be only me? ;-)

Hi Tim, it is quite possible that you are the only actuary who reads this newsgroup - we'll possibly never know.....

What I am able to do is to write a programming script to create a new single-entry table from the base tables adjusted to the year of birth, as you advised previously.

This can be fed into another script which calculates the annuity factor based on the qx data fed into it.

I probably did'nt word my post above clearly enough. I should have said I understand the CONCEPT of why the cohort basis is applied to mortality - however, I don't fully understand exactly what adjustments should be made to year of birth tables once they have been generated.

Does this make any more sense ?

"sylvian stone" wrote

OK so far...

"sylvian stone" wrote

Not really! Using the medium cohort is just one way of producing a projected "year of birth" table. So, you don't really need to make any adjustments to the 'year of birth' table once it has been generated....???!

OK, I'll try and explain another way.

In order to create the new table, I:

• Import the base table
• Calculate the values of f(x) and A(x) and use these to calculate RF(x,t), the reduction factor as specified in CMIR17. From this a new 'Year of Birth' table is generated.
• Use the qx values generated in the new table to calculate an annuity factor.

At exactly what step does the 'medium cohort' come into play ?

While I can read in plenty of places that the medium cohort basis should be used, I can't figure out at what step it is applied......

Does this make a little more sense now ? (or not.....)

"sylvian stone" wrote

The medium cohort *replaces* the RF(x,t). RF(x,t) is the "old" method, 'short/medium/long cohort' are the "more recent" methods (and the "new" methods are still being discussed!).

Does that help?

It does to a point.

But I can't find any resources that tell me how to apply it - everything I have looked up just states 'use the medium cohort' basis.