[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries

Making sense of the past

Hai Luc and Grigory Spivak presented their paper, subtitled ‘Issues to consider when analysing mortality experience’ to the SIAS meeting on 5 July 2005. Over 50 members from across the age range heard chairman Charles Woodd praise Hai and Grigory for writing a paper while students, and encouraged other students to follow their example.
Hai introduced the paper by saying that their aim was to improve the standards of experience analyses undertaken. The paper includes many reasons why direct offices should conduct better analyses. An additional reason was provided from the Morris report: improving the quality of information and advice provided by actuaries to our non-actuarial colleagues.
Hai and Grigory ran through a number of the practical issues of undertaking and interpreting experience analyses. Grigory repeated the point made in the paper that the select shapes of the 92 and proposed 00 tables were wrong, and that this may causes errors in use. In particular, the limit of five years for the select period was highlighted as an issue, especially when compared with the 25-year select periods used in North America.
Speakers welcomed the paper as a valuable guide and reference manual. Debate focused around a number of areas.
The possible use of generalised linear modelling was raised by some speakers as a simple and powerful additional tool. Extensively used in general insurance, impressive examples of its value were quoted. However, other speakers queried whether there were sufficient claims to produce credible results.
The use of multiple decrement tables for combined mortality and morbidity analysis was discussed. The CMI currently uses a simple approach for its critical illness study, but hopes to extend this to cover different impairments as volumes of data increase. An Australian actuary outlined the problems in his home country where the extensive sale of ancillary rider benefits on mortality plans was forcing actuaries to work out how they could do multiple decrement analyses to identify the underlying mortality experience.
Postcode rating was considered, but rejected as impractical and potentially racially discriminatory and therefore illegal. The postcode where employed was, however, considered legal and of great value on identifying aggregation risks.
There was extensive debate on the appropriate length of the select period. The role of anti-selective lapsing was highlighted as a potential major driver of extending the select period. One speaker quoted reasonable models producing 10% higher mortality rates after ten years. Rebroking in recent years will have a significant impact on current and future reported experience. The question about the CMI tables may not be ‘are the select shapes right?’, but ‘is the ultimate experience at the right level?’.
It was pointed out that US tables with 25-year select rates were quoting ultimate experienced based on business written in the late 1970s. If we did this in the UK then we would be pricing using the experience of business on different types of policies, written in totally different ways, through different distribution to different standards.
The authors have helped actuaries make better sense of the past the next stage is actuaries making sense of the future.