[Skip to content]

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

Save millions on annuities!

There are currently 10 million people in the UK over retirement age. Around half of these people are receiving a pension in payment from one of the top 20 insurance companies. Analysis of a selection of these companies reveals some interesting facts:
– The average annuity in payment is approximately £1,750 pa.
– Nearly 70% of deaths are notified to the insurance company within three months of the date of death, with around 90% notified within one year.
– Of the 10% deaths left unreported 12 months after the date of death, the average time before identification is almost four years.
– Assuming an average mortality rate of 2.5% for an annuity portfolio implies there are some 125,000 deaths among pensioners every year.
Of the 12,500 deaths that are unreported after a year, around 2,500 may be single life annuitants and a further 2,500 the first life on a joint life annuity. The remaining 7,500 will be second-life deaths that may lead to a reduction in reserves, but no overpayments. Thus, at any time there could be up to 50,000 very late reported deaths for which insurance companies continue to hold full reserves and moreover continue to pay claims. An estimated £400m in reserves could be released by identifying these 50,000 unreported deaths. Overpayments made over long periods are likely to irrecoverable by the payer.
In reality, the situation will not be quite this bleak, because many pensions are within group schemes and the overpaid annuities can be reclaimed through the trustees, although this is small comfort for the trustees. Pension schemes can also benefit from identifying unreported deaths.
At a time when annuity and pension providers have had to bolster reserves substantially and may have to do even more, limiting the potential for overpayment and over-reserving on spurious liabilities seems a good housekeeping measure. How can you mitigate your exposure to this overpayment? There are two options.

The traditional way
Historically, almost all companies have at some point used certificates of existence to identify unreported deaths. Certificates of existence are advantageous in that the procedures are tried and tested and are often already in place. They can also be applied as often or seldom as deemed required and to any proportion of the whole portfolio.
However, the costs of writing to each and every annuitant can be substantial around £3 per life. Ideally certification exercises should be carried out annually on the entire portfolio, but to reduce costs, companies often limit this exercise to older lives and will only write to annuitants every two to three years. Chasing up non-respondents can add to the costs and even once people have responded, the insurance company has few means of verifying the information provided.
The certification process is generally seen by customers as a nuisance (and perhaps a distressing reminder of impending mortality!). The customer may also incur expense if the certificate has to be witnessed by a third party. It is estimated that the whole process of issuing and processing certificates of existence is of the order of £5 per life.

Into the 21st century
In contrast to the laborious process of writing to each annuitant to make sure they are still alive, there are now more sophisticated ways to check that annuities are still valid. During the past two or three years, electronic databases of deaths registered in the UK have become easier to purchase and the quality has improved significantly.
A key advantage of using a database is that the whole portfolio can be analysed at a fraction of the cost of issuing certificates of existence. As the databases of deaths are regularly updated the portfolio can be analysed as often as required, without the need to contact annuitants. For territories not covered by the databases, certificates of existence can be used, but this would entail a much smaller exercise than writing to an entire portfolio. In addition, a manual process of verification may be required before terminating a annuity, but this again is a more focused and less costly project than contacting each and every annuitant.
The process to match the deaths in the database of deaths to the annuity portfolio is a relatively complex IT task, owing to the different formats and inconsistencies in the stored data. When performing this exercise you need to decide the purpose:
– Do you want to identify as many unreported deaths as possible, in which case a complex matching system will be required? or
– Will you be satisfied with only a portion of the unreported deaths, in which case a simpler system may be more appropriate?
The data will never be perfect, and so some deaths will inevitably be missed because of typographical errors and missing errors. Names can be misspelt, for example Smith with Smyth and people are often known by a middle name or nickname, rather than their given forename. Significant IT resources will be required to establish systems to identify the unreported deaths and to clean the annuitant and deaths data to maximise the number of unreported deaths found. Once a system has been established, it will need to be tested and regularly updated as new and better information becomes available.
Dead or alive?
No single database of deaths contains enough information to be able to identify accurately all deaths within a portfolio and some of the more useful databases can be prohibitively expensive, particularly for smaller portfolios. Using a third-party provider or consultancy can often work out cheaper and more effective than trying to do the exercise in-house. However, many third-party providers will not only miss a significant proportion of deaths, but may also erroneously report that annuitants who are still alive have died. The quality of the matching service provided varies considerably between providers, and quality this will directly influence the accuracy of the exercise.
So, how can you maximise the unreported deaths found, but minimise the number of false death matches? One solution is to use more than one data source and a variety of techniques to identify the ‘best’ data fields. Having chosen the most appropriate and reliable data fields, the items need to be checked for accuracy and completeness. Again different data items may warrant different treatment. For a date of birth, for example, it is a simple enough process to check the range of dates so that unlikely dates or ‘pockets’ of dates which would indicate false data can be flagged as being potentially unreliable. Dates of birth in the future are obvious errors and if significantly more than 1/365 of dates of birth occur on a particular date then some pocketing may be present.
If a company buys databases itself, robust validation processes should be adopted for those deaths databases which are known to contain errors and omissions. For example, if three of the databases are purchased and a death is notified in one database only, it is unlikely to be genuine. A crucial check of genuine deaths is to verify their presence in more than one database, but this obviously has cost implications for individual companies who may not want to buy several expensive databases,
As always, keeping data clean and up-to-date is important to get the most out of IT-dependent solutions. However, unreported deaths may be costing you more than you think and you should be able to generate a positive return on investment easily from this exercise. o