New SAPS tables
On 31st October 2008 the CMIB published the new SAPS S1 mortality tables based on the mortality of defined-benefit pension schemes. These tables follow the previous release of the '00 Series' mortality tables based on the mortality of life-office pensioners. It is instructive to compare the life expectancies under the two series.
One drawback is that there are rather a lot of tables in the S1 series: ten in total. We will use the so-called amounts tables, i.e. mortality as weighted by the size of the annual pension, as this gives a better indication of the financial importance. The life expectancies below are calculated using the S1PMA and S1PFA tables from SAPS S1, and the slightly older PCA00 table.
| PCA00 | S1PA | Change | ||||
|---|---|---|---|---|---|---|
| Age | Males | Females | Males | Females | Males | Females |
| 65 | 18.4 | 20.9 | 18.1 | 20.6 | -2% | -1% |
| 70 | 14.5 | 16.7 | 14.2 | 16.5 | -2% | -1% |
| 75 | 11.0 | 12.9 | 10.8 | 12.8 | -2% | -1% |
| 80 | 8.1 | 9.7 | 7.9 | 9.5 | -3% | -2% |
| 85 | 5.9 | 7.0 | 5.6 | 6.8 | -4% | -3% |
So the newer SAPS tables are producing lower life expectancies than the 00 Series tables. How can this be if mortality is continuously improving? The answer most likely lies in the SAPS data having a higher proportion of lower-status lives. In practice it will make little difference if an actuary uses a life-office table or a SAPS table, as demonstrated here. Indeed, we earlier posed the question whether actuaries need standard tables at all. Most life offices and large consultancies have data sets which can be used to create tables which are are much richer in risk factors.
Previous posts
Do we need standard tables any more?
Actuaries are long used to using standard tables. In the UK these are created by the Continuous Mortality Investigation Bureau (CMIB), and the use of certain tables is often prescribed in legislation. As actuaries increasingly move to using statistical models for mortality, it is perhaps natural that they should first consider incorporating standard tables into these models. But are standard tables necessary, or even useful, in such a context?
Add new comment