Articles - Mortality Improvement Experience & Projection Techniques

Global Mortality Improvement Experience and Projection Techniques

 By Marianne Purushotham, Emil Valdez and Huijing Wu of Towers Watson
 This report is intended to serve as an aid in the discussion of best practices in the development of a view regarding future expectations for rates of insured and annuitant mortality improvement. To that end, the report first reviews historical rates of improvement for both the general population and the insured population from a global perspective. It then provides an overview of current techniques for modeling and projecting future rates of mortality improvement. In order to provide a broad perspective on the topic, this report reviews the work of both the actuarial profession as well as other disciplines, including demographers, economists, statisticians and members of the medical profession.
 As a first step in developing the report, the Research Team created a detailed database of research papers, industry publications and other sources of information related to mortality improvement experience and modeling approaches. Appendix A provides detailed citations for data sources identified as part of this project work.
 Information is presented in three major areas — global trends in mortality improvement (Section 3), detailed information regarding mortality improvement for both the insured and general populations in the United States (“US”), Canada and the United Kingdom (“UK”) (Section 4), and finally, biometric theories regarding future rates of mortality improvement as well as mortality improvement modeling and projection techniques in common use (Section 5). Section 6 outlines a current view regarding future rates of mortality improvement over the medium term (10-20 years) for individual life and annuity insureds based on information reviewed as part of this project work.
 General Approach for Developing Mortality Improvement Results
 Note that this report was developed with the intent of providing information regarding general trends over time. To that end, the reader should be aware of the following information as he/she reviews the data presented. Because information is drawn from multiple sources with varying formats, the following general approaches were adopted for combining and compiling data.
 1. Mortality Improvement Calculation.
 For purposes of this report, mortality improvement experience is presented as an average annual rate over a specified period of time. Specifically, the mortality improvement for an individual age x between time t and time t+k is calculated as follows:
 Avg Ann Improvxt to t+k = 1- (qxt+k/qxt)^(1/k)
 2. Age Groupings.
 Some sources of information do provide data by individual age while others do not. Therefore, for purposes of making comparisons in experience data across various sources, a common age set of groupings was determined as follows: 0-1, 1-4, 5-9, 10-14, 15-19, 20-24, 25-34, 35-54, 55-64, 65-75, 75-84 and 85+. For data sources where available mortality data included death counts and amounts, exposures by count and amount, and central death rates or mortality rates, data was combined into the various groupings defined above. For other sources, including some of the insured data, only the final mortality rates are available in the level of detail required. In these cases, we have presented results for individual representative ages, typically ages 30-95. Care should be taken in applying insured data mortality improvement results directly. For this reason, in developing a medium term view of future levels of improvement, we have focused on population mortality as the basis for our estimates.
 3. Experience Periods.
 An attempt was made to align the time periods over which mortality improvements are measured for the general and insured population data presented in order to allow for comparison. It should be noted that for making comparisons of insured, annuitant and pension data, the time periods over which improvements in mortality are measured do not precisely overlap and therefore there is some distortion introduced in these comparisons.
 Also, most academic and technical sources measure mortality improvements over long periods of time (40-50 years or more), however consistent data over this length of time is generally not available for the insured populations. For this reason, we have focused on the period from 1965 to the present and, wherever possible, we have broken down results into the following sub-periods: 1965-1980, 1980-2000, and from 2000 to the most recent year available.
 Finally, in cases where an experience study covers multiple years, we have used the midpoint of the period as the reference point for measuring mortality improvement levels.
 4. Estimating Overall Improvement Levels.
 Except in cases where there was insufficient information, we have included a “total” or “all ages” mortality improvement statistic for population and insured data.
 5. Primary Data Sources for Mortality Improvement Measurement.
 In order to facilitate the comparison of population data across both geographic regions and demographic groups, much of the general population information presented utilizes the Human Mortality Database (“HMD”) maintained by the Center for Demographic Studies at the University of California at Berkeley. In cases where alternate data sources within a local country are available, we have noted those sources and any material variations from HMD results.
 For insured data, wherever possible, we have measured mortality improvements by comparison to the basic experience tables developed by various actuarial organizations worldwide from periodic mortality experience studies. Basic experience tables have been supplemented by population sources for age bands where actual experience is less credible and have been smoothed and graduated.
 We note that at this time insured data is difficult to interpret due to “noise” in the results. The two main sources of noise arise from shifts in the mix of companies participating in industry studies over time and from changes in the underwriting and risk classification structure over time. Keeping this in mind, we do present what data is available for purposes of beginning a discussion regarding future assumptions. Finally, detailed information is focused on the US, Canada and the UK where relatively consistent long-term data is available.
 6. Mortality Experience Basis for Insured and Annuitant Data.
 All data for individually underwritten ordinary life insurance is included in the mortality improvement measurements without regard to policy size or other data breaks. Also, in most cases, we present mortality improvement results on an amount basis (rather than a number of policies basis). This decision was made for two reasons. First, in practice, mortality improvement assumptions are applied directly to base insured mortality which is typically developed on an amount basis. Second, in the US and Canada, it is more difficult to obtain historical mortality experience data on a number of policies basis.
 For the UK mortality investigations, mortality experience by amount is currently collected for the immediate annuitant and life office pensioner lines. For all other investigations, data is only available based on the number of policies. The Continuous Mortality Investigation (“CMI”), the Institute of Actuaries group that publishes regular studies of UK insured and pension mortality experience, is currently moving to a seriatim data collection basis and they expect that future reports will include mortality results on both a number of policies and amount basis.
 To view the entire research paper please download the PDF or alternatively visit the SOA website here
 © Society of Actuaries, Schaumburg, Illinois. Posted with permission.

Back to Index

Similar News to this Story

Connecting business volume and volatility
A key challenge faced by capital modellers is how to appropriately connect changes in premium and reserve volume with the volatility parameters used i
Cutting your retirement cloth to fit your pension
Our local tailor has decided to retire. Not something which is usually newsworthy. However, Tarcisio is 88 years old. And he is still not ready to giv
Climate models and investing what is the issue
Join Senior Sustainable Investment Consultants Clare Keeffe and Jordan Griffiths as they discuss climate modelling, and the many myths that surround i

Site Search

Exact   Any  

Latest Actuarial Jobs

Actuarial Login

 Jobseeker    Client
Reminder Logon

APA Sponsors

Actuarial Jobs & News Feeds

Jobs RSS News RSS


Be the first to contribute to our definitive actuarial reference forum. Built by actuaries for actuaries.