Try R, which is free, robust and reliable. If you want to step up to something more sophisticated later, then Longevitas will fit a wider variety of models. If you are worried about back-testing, Longevitas also generates files suitable for reading into R and will also generate R command scripts.

If you have a web browser and an Internet connection, you can run Longevitas.

Technically the minimum amount of data is one event. In practice, the smallest data set we have seen used was a pension scheme with 6,000 pensioners.

Try mortalityrating.com, which uses a model of mortality by age, gender and postcode-driven lifestyle to rate small portfolios with little or no experience data. All you need are the following five data items for your pensioners: date of birth, gender, commencement date, annual pension and the full U.K. postcode.

1. The latest mortality models at the click of a mouse-button.
2. Integrated support from a leading industry expert on modelling mortality and other demographic risks.
3. No software to install!
4. Use anywhere in the world.
5. Intuitive, menu-driven approach - no programming language to learn!
6. Online library of what works, what doesn't work (and why!)
7. Unique expert system suggests improvements to your models based on fifteen years' experience of modelling mortality.
8. Automatic report generation to document models.
9. Automatic rate-table generation for use in pricing and reserving.

Yes. Richards and Jones (2004) showed the impact on pension and annuity reserves of five rating factors besides age. Each one of them could cause a change which could wipe out a typical annuity pricing margin.

Standard tables are fine as they go, but they usually only use two rating factors: age and gender. Richards and Jones (2004) found four further rating factors beyond age and gender, all of which were material for pricing and reserving.

Incidentally, Longevitas has a charting tool which enables you to visually modify standard tables to better match the modelled mortality patterns observed in your portfolio. We also have a blog entry questioning the usefulness of standard tables in modelling work.

See our Institute of Actuaries paper, which was peer-reviewed and details how and why postcodes work well for explaining mortality variation. For a generally lighter read without the maths, try our article in Life and Pensions magazine.

First, there are some decrements which many of the same underlying features as mortality, such as critical illness. Mortality-orientated models tend to work well here.

Second, transfer and persistency risks do indeed have different behaviour, which is why we have a separate set of variables for application there.

Incidentally, we have a blog entry on the importance of using survival models for competing risks.

Versions of Longevitas exist for the U.S.A., Canada and the Netherlands, each of which has a hierarchical postcode-like structure, e.g. the nine-digit zip code in the United States. Other countries, such as France and Germany, can use address-based profiling.

See our Institute of Actuaries paper, which was peer-reviewed and details how and why postcodes work well for explaining mortality variation. For a generally lighter read without the maths, try our article in Life and Pensions magazine.