Kevin Armstrong, chair of the Continuous Mortality Investigations Life Office Mortality Committee, explains how the investigation is adapting to the challenges presented by the evolving annuity market and considers how it can best serve its users in this area
One of the CMI's most important roles is to provide users with reliable, relevant and useable mortality tables that accurately reflect the experience of a specific subset of the population. For many years the CMI has collected, analysed and reported the experience of life insurers and produced tables that are benchmarks for financial reporting. These cover life insurance, annuities, critical illness insurance and income protection and each area presents technical and practical challenges as markets evolve.
More recently, pensions legislation and developments in the annuity market have meant that the factors influencing mortality rates are less easily attributable to fixed categories of business. Consequently, the ability to split data into categories is less important than having more granular data that can be analysed by a variety of risk factors.
In the past, the CMI relied on 'Scheduled Data' for its analyses of life office mortality data. Contributing offices summarised the data before submitting it and the analysis that the CMI could undertake was therefore restricted. Now that the technical resources available to perform the analysis are so much greater, the CMI is keen to extend the analysis that it can offer. So we have developed a new 'Per Policy' approach to data submission that aims to improve the granularity of data, which would in turn allow better validation of data and more detailed analysis. But one of the strengths of the CMI - its broad market coverage - is also a potential weakness: the wide range of contributing offices used different systems, and providing data in detail and to a fixed format required considerable effort. Data contribution slowed down considerably.
This led the CMI to a special data initiative for 2007 onward. We would request only certain essential data fields so contributors could submit data in a form that suited them but could nevertheless be used to provide reliable and meaningful results. The full 'Per Policy' data format remains an aspiration but this interim approach is more realistic in the short term.
The response from data contributors has been very encouraging indeed, and we are grateful for their engagement and commitment to providing accurate, useable data. Chart 1 shows how the latest exercise has already boosted the data volumes for annuity business, with some offices still intending to submit data. Unfortunately one downside of the greater flexibility in data formats is that the CMI secretariat is not always able to segregate the new data; hence the chart shows a single bar for the 2007-2010 data. As results are being sent back to offices, the secretariat is trying to obtain greater clarity on the types of business covered, where possible.
Producing a sufficiently homogeneous data set is not straightforward when data is provided by a number of contributors (and potentially different contributors over time). This is not a new challenge but we are looking at methodology that allows better for the inclusion of different offices' data in the analysis from time to time, and techniques for analysing the data such as Generalised Linear Models or survival modelling. Even so, the development of the enhanced annuity market presents the CMI with a particular challenge.
Only a few years ago, the distinction between standard annuities and enhanced annuities was relatively straightforward. Indeed, annuities for customers in ill-health were known as 'impaired life' annuities and were only available to a small proportion of lives deemed to have a materially reduced life expectancy (albeit even then according to individual offices' definitions). Now, enhanced annuities are available to people with relatively mild medical conditions and to smokers. And when the majority of annuities sold are postcode-rated, does it even make sense to talk about 'standard' annuities?
Further, it is not straightforward to analyse the impact of the development of the enhanced annuity market on emerging experience. Enhanced annuitants are expected to have higher than average mortality but the majority of deaths in current experience will be in respect of business that originated when the enhanced annuity market was nascent or non-existent. The improved healthiness of lives in the 'standard annuity' category will only gradually become apparent in experience data.
This picture may also be blurred by differences in purchasing behaviours. There may be a greater tendency for those in ill-health to purchase an annuity via an open market option, so CMI 'standard annuity' experience may be influenced by the mix of contributing offices. The extent to which an insurer brings this option to customers' attention may also influence their experience.
Ideally, the CMI might seek to produce analyses and tables on three bases:
> 'Standard annuity' business only, to illustrate how the development of the enhanced annuity market is affecting the pool of lives that do not qualify for any enhancement;
> Enhanced annuity business only, which would illustrate the differential in mortality for those who do qualify; and
> A combined analysis, which would illustrate the experience for the overall market and might better illustrate trends in mortality.
However, analysis covering only enhanced annuity data seems some way off, due to the concentration of in-force annuities amongst a small number of insurers.
For 2007-2010 annuity results, the more immediate question has been whether to exclude the enhanced annuity data that we have received or to include it in an aggregate data set, covering the spectrum from the super-healthy to the severely impaired. We suspect that the proportion of enhanced annuity business in the data received to date may not be representative of the market and our tentative conclusion excluding enhanced annuities may be most useful to our users. These results would offer a broad overview of the mortality experience of the non-enhanced annuity market and hence provide the clearest benchmark for users seeking to compare the experience of their annuity portfolios against the market.
There are considerable challenges for the CMI in this area, but they are challenges that the industry must face and the CMI is well placed to provide the thought leadership needed to ensure that risks are clearly understood. However, we need support from the industry to do this, through data contributions that give the analysis maximum credibility and from wide engagement that informs how we can best provide the analysis which users need. We intend to contact key stakeholders soon to obtain their views, but we would welcome views from any current or potential users.
