When reviewing mortality study experience, it is important to assess not only the overall level of mortality but underlying mortality trends as well. In this article I discuss how actuaries can massage the experience to reveal underlying patterns. The most common of these methods involves grouping data into meaningful units which permit the actuary to analyze one aspect of mortality at a time.
In order to perform a trend analysis, the experience study should use a standardized table for the calculation of expected claims. Changes in actual-to-expected (A/E) ratios can then be attributed solely to underlying mortality patterns. Contrast this to a study where expected mortality is set equal to an evermoving pricing assumption, such that constant A/E ratios of 100 percent are a good thing.
Once expected claims have been determined, then A/E ratios should be derived across various views, including issue year, calendar year, policy year and underwriting class.
Grouping the data by year of issue can reveal the effects of evolving underwriting practices. For example, during the 1990s the introduction of preferred underwriting and the proliferation of multiple preferred risk classes reduced overall A/E ratios. This was because companies issued a higher proportion of preferred business due to higher placement rates in the preferred classes. The issue year trend of A/E ratios observed during the 1990s would reveal the effect of this change.
Similarly, a company that implemented significant changes in underwriting practices over the years should have a pattern of A/E ratios that correlates with their changes. More importantly, however, the actuary will want to segment the issue year experience by these “underwriting eras” to analyze additional trends unrelated to underwriting. The need for this segmentation will become more obvious as we look at other groupings.
Calendar Year of Claim
By grouping experience by year of claim, the actuary should theoretically be able to observe a “secular” mortality trend. By this term I mean the long-term changes in mortality related to environmental factors that affect populations as a whole – such as access to medical care, lifestyles, nutritional habits, etc.
Unfortunately, insurance companies find it difficult to effectively analyze secular mortality trends from their own data. Since each calendar year contains claims on policies from many different issue year eras, too much variation is therefore introduced into the data by: changing underwriting practices, changes in number and characteristics of preferred classes and changes in product design and the sales market. The ability to analyze claim year trends independently by issue year era helps mitigate the impact of these variations.
Policy Year (Duration)
As I discussed in a previous Messenger article, grouping the A/E ratios by duration typically reveals an upward mortality “blip” in the third policy year, when the contestable period ends. In addition to this impact, durational analysis could also be used to show how well the slope of the expected mortality table tracks actual experience.
However, changes in historic underwriting practices may affect results. The later the duration, the farther back in time the original underwriting occurred. An upward trend in A/E ratios by duration may be more of an indication of earlier preferred class designs producing higher aggregate mortality rather than a true slope differential. Once again, it is important to segregate the experience data into underwriting eras before performing this type of trend analysis.
Preferred Underwriting Class
One of the tenets of modern mortality theory is that insureds deemed “preferred” will exhibit overall lower mortality than their non-preferred counterparts. There is still much debate, however, about the magnitude of this mortality improvement and how long it will persist beyond initial underwriting. Hence, a review of A/E ratios by preferred class should be a significant part of any trend analysis.
Once the data has been segregated into appropriate issue year eras, the actuary should review the relative mortality differentials by class and compare them to pricing assumptions. If data credibility permits, further analysis by policy year may indicate the stability of these differentials as other selection factors begin to wear off. Regrettably, many companies lack sufficiently consistent data to measure this effect beyond the early durations.
Our Mortality Experience System
We have created a mortality experience database culled from information on reinsured policies. Using a convenient pivot table user-interface, our pricing actuaries can view aggregate or client-specific data for a variety of different subgroups. We can view A/E ratios by any combination of issue year, claim year, gender, underwriting class, issue age, company, product type, face band and other characteristics. This provides a powerful tool in understanding not only specific mortality results, but also in analyzing mortality trends.
In Part II of this article I will discuss other data groupings that actuaries use to reveal interesting patterns within mortality experience and provide some concrete examples. Stay tuned...