Duncan Heald and Gerard Kennedy analyse the latest trends in income protection claims, and some of the implications for insurers
Understanding the effect of different causes of sickness is essential to the proper actuarial modelling and management of income protection (IP) business. Knowledge of the distribution of claim inceptions by cause and their subsequent cause-specific patterns of termination, and therefore cost, enables accurate pricing of IP products, efficient reserving and improved management of claims in payment, and better informed product development.
In the recently published CMI Working Paper 72, the Continuous Mortality Investigation (CMI) IP Committee presents an analysis of IP claim experience by cause of sickness. The underlying claims data covers the period 1991-2009 and includes over 56,000 claim inceptions and over 43,000 claim terminations (recoveries and deaths), classified according to 72 individual causes of sickness. Results are stated on a 'lives' basis, with IPM 1991-98 serving as the expected basis for comparison. This article highlights some of the more interesting features of that analysis.
In order to facilitate the interpretation of results, and for statistical robustness, claims are aggregated into 15 cause-of-sickness groups, primarily on the basis of medical similarity. In the figures below, we present results for the seven most significant of these cause groups.
Figure 1 (right) highlights the key contributing cause of sickness groups for each sex. The much higher percentages for females reflect the fact that IPM 1991-98 is based on male experience only.
Cancer (neoplasms) and mental illness are relatively more prevalent for females, whereas injuries and circulatory are relatively more prevalent for males, and musculoskeletal claims are a significant contributor for both sexes.
The figure also illustrates the observed significant improvement in inception rates over time. There are several possible
reasons for this:
? Standards of underwriting have strengthened significantly over time
? There has been a reduction in the number of occupations where an 'own occupation' definition of disability is offered and a more restrictive activities-based definition is now more common, though this trend is
? Exclusions for particular medical conditions are also now more readily applied than in the past. Such changes may have led to the general observed reduction in claims from injuries and musculoskeletal causes
? Restrictions on cover for certain occupations, such as teachers, may have contributed to the observed reduction in mental illness claims
? Significant improvements to health and safety over the investigation period are likely
to have led to reduced inception rates for certain causes
In common with other protection products, IP inceptions due to cancer have remained fairly stable over time, reflecting the fact that this cause is largely impervious to underwriting.
IP claim terminations consist of claimant recoveries and claimant deaths. It is the recovery rates that are of most interest
Figure 2 illustrates the variation in the male recovery rate with duration sick for each of the main cause-of-sickness groups, and for all causes combined, expressed as a percentage of the all-causes basis, IPM 1991-98. The pattern for females is broadly similar. The figure clearly demonstrates which causes lead to short-tail claims (acute respiratory and infections) and those that have a much longer recovery time (mental illness).
The all-causes recovery rate is quite close to expectation for short durations but higher than expected at later durations. Since the expected basis was set using the 1991-98 data, this indicates that there has been an improvement in the more recent (1999-2009) experience and that recovery rates are now higher than before. This could be due to a number of factors, including improvements in medical interventions (for example, improved cancer screening and treatment) and greater emphasis by insurers in helping claimants return to work (including much more awareness and understanding of mental health issues).
A similar analysis of claimant deaths shows that about half of all deaths of claimants in payment are in respect of cancer claims.
Since claims by certain causes tend to be more long-tailed than others, the distribution of claim incidence by cause group often differs markedly from that of claim cost. This is illustrated in Table 1, which, as an example, shows both distributions, for both males and females aged 40 at the onset of sickness, for a policy with a four-week deferred period. This table also shows the average cost per claim, on the basis of a £1,000 a year claim annuity payable from the end of the deferred period until attainment of age 65 or earlier termination and valued at 4% a year effective. It should be noted that these distributions and costs can vary significantly with age and deferred period.
For both sexes, the most prominent cause groups contributing to both the claim incidence distribution and the claim cost distribution are mental illness and musculoskeletal. However, whereas musculoskeletal claims contribute the same weight to both distributions, mental illness claims, due to their long-tailed nature, contribute rather more weight to the claim cost distribution than to the claim incidence distribution, especially so for females. The table includes all known cause groups,
which allows greater visibility of the less significant causes.
It can be seen from these average claim cost figures, assumed to apply in 2002, that the cost of claims are greater for females for nearly all cause groups. Further investigation shows that, since then, female improvements have been greater than males' for both inceptions and terminations, which is helping to reduce this disparity - welcome news to actuaries pricing in the new gender-neutral world.
As the above trends and cost differentials indicate, it is vitally important that pricing and reserving actuaries keep abreast of both evolving market practice and medical advances. Changes in underwriting philosophy, workplace regulations, and claims management processes, as well as in the prevalence and treatment of different medical conditions, all influence the assumptions that underlie both the design and pricing of products and the reserving and management of claims in payment. Since the effect of such changes often varies by cause of sickness, it is likely that actuaries will increasingly need to focus on cause and its effect.