The Increasing Importance of Mortality and Morbidity Risks
The reduction in long term interest rates over the last decade, combined with unexpected and rapid recent increases in longevity, has meant that the relative significance of mortality risks, especially for longevity products, has increased dramatically. Rapid increases in longevity, particularly for the current generation of pensioners as shown by work on cohorts by Willets, R (1999), have been driven by a number of factors that include medical advances, increased availability of healthcare, changes in diet and reductions in the prevalence of smoking amongst males.
Medical advances in recent years have also enabled earlier detection of critical illnesses as well as better treatments and higher recovery rates. Further, the improved diagnostics have also driven changes in medical definitions of what constitute critical illnesses. These have resulted in higher critical illness claims costs as claims are being paid for conditions that were thought to have been excluded, and because claims are happening earlier than expected in the policy term when fewer premiums have been collected.
At the same time, demographic and economic changes have resulted in insurers writing an increasing proportion of risk business. In particular, sales of longevity insurance products which have seen rapid recent growth seem set for long and sustained periods of further growth. Many companies also have leveraged exposure to longevity risk via Guaranteed Annuity Options (GAOs) on their existing books of business. The combination of increased risk and exposure has resulted in increasing focus on mortality and morbidity risks within the insurance industry and by the regulators. Indeed, regulators have included firms with "exposure to areas with greater underwriting uncertainty such as longevity and morbidity risks" amongst those that can expect more critical reviews for their Individual Capital Guidance (ICG).
The increasing appreciation of the uncertainty inherent in mortality and morbidity risk business has led to regulators and the actuarial profession highlighting the potential and impact of the uncertainty and providing additional guidance. Reserving and capital requirements now need to make fuller and more explicit allowances for the different sources of uncertainty.
- The risk of year-on-year volatility of claims experience and the uncertainty relating to the estimation of the current level of mortality is probably the best understood of the different sources of uncertainty. These risks are mitigated to a large extent for substantial portfolios (i.e. the bigger the portfolio, the lower the uncertainty).
- The uncertainty, however, of the relevance of historic data to future experience cannot be mitigated by the size of portfolios. Market segmentation, such as the increasing proportion of impaired life annuities, and changing critical illness claims definitions reduce the relevance of historic experience.
- The uncertainty relating to the choice of mortality projection models and the estimation of projection parameters can be difficult to quantify. Judgement is critical when choosing the model and estimating the parameters, as small differences can give very different pictures of the potential risk.
- The heterogeneity in historic data results in uncertainty in the estimation of the current level of mortality and therefore any projection model and parameters. For annuity portfolios the impact of heterogeneity can be severe with, for example, the mortality of those with small pensions being up to twice as high as for those with the highest pensions. Careful analysis and judgement are necessary to reduce the uncertainty arising for heterogeneous data.
- It is much harder, however, to estimate the potential and impact of external events resulting in adverse longterm trends (e.g. lifestyle changes and medical advances) and therefore these are much harder to mitigate.
The FSA requirement for firms to calculate mathematical reserves using prudent rates of mortality and morbidity, based on current experience and containing margins for adverse deviations, remains unchanged with the introduction of the Integrated Prudential Sourcebook (PSB). These rates are required to take account of anticipated improvements in longevity, changes arising from improved detection of morbidity and changes in market segmentation. Allowance must also be made for the incidence of diseases whose impact may not yet be fully reflected in current experience.
The additional PSB requirements relating to Individual Capital Assessment (ICA) calculations require firms to analyse the potential for reserves to subsequently prove inadequate. Firms are also required to determine the effect of adverse claims experience due to unexpected trends as well as volatility. The correlation between persistency and mortality and morbidity assumptions must allow for the risk of selective lapses in conjunction with adverse claims experience. In effect, firms are being pushed to develop stochastic models of mortality risks.
Guidance from the Actuarial Profession
The actuarial profession's proposed revisions to actuarial guidance for the changes arising from the introduction of the PSB require that the margins for adverse deviations account for the degree of uncertainty regarding the causes of observed changes in experience and the consequent range and likelihood of possible future changes. For annuitant longevity assumptions, the guidance recommends using a cohort approach.
The guidance on the calculation of ICAs also steers firms towards the use of scenario testing, if not full stochastic mortality models. These require the ICA to cover long-term adverse trends and large-scale events, such as significant advances in the treatment of a critical illness or a commonly available treatment to significantly delay the normal ageing process, as well as the year-on-year volatility of the claims experience. The ICA should extrapolate, as a starting point, historically observed trends, whether adverse or favourable, unless there is evidence against such extrapolation. The extreme adverse event should then be a reasonably foreseeable worsening of this trend or its rate of acceleration or deceleration. However, independence can be assumed between mortality and economic assumptions where a combined stochastic model is used.
Implications for life insurers
Stochastic mortality models
Given the level of uncertainty inherent in mortality risk, insurers with any significant exposure to this risk will need to use stochastic techniques to meet the new requirements for reserving and capital. Even if stochastic methods are not used directly to set reserves and capital requirements, such models will now be increasingly necessary to provide evidence that sufficiently adverse scenarios have been covered. However, stochastic mortality models are at a very early stage of development, particularly where the projection element is also stochastic, and so significant judgement is needed in choosing and interpreting these models. Except for the very largest, most insurers will struggle to find the necessary expertise and resources in-house.
Peer review of assumptions and models for mortality and morbidity
"House" views of mortality and morbidity can have a huge impact on the level of reserves set up and capital set aside, making it much more important for independent peer review of the assumptions and models relating to mortality and morbidity risks. Where consultants are used to set the assumptions and models relating to mortality and morbidity risks, the consultants' "house" view is substituted and independent peer review may still be necessary. This is particularly true for the case where the consultants were commissioned to do this work due to lack of specialist expertise within the insurer.
The actuarial profession is now set to require general peer review of the reserving and ICA calculations. However, the unique nature of mortality and morbidity risks means that specialist skills are required and so insurers may wish to commission separate independent reviews of their mortality and morbidity risk management processes as well as the assumptions and models used to set reserves and capital requirements.
Please speak to your usual Barnett Waddingham LLP contact to discuss any of the above.
Alternatively contact Tony Leandro, Rajeev Shah or Dave Grimshaw on 020 7776 2200.
Barnett Waddingham LLP, November 2004.