Introduction
Longevity is today one of the main risks assumed by insurance companies and one of the more complex to predict. In fact, all past predictions have underestimated longevity risk. Predicting how long an individual is going to live depends on several factors, from lifestyle and genetics to environment and medical advances, as even simple ones can significantly influence life expectancies.Some are even saying that in the future, aging might be treated as a disease condition rather than an inevitability.
Have humans reached the upper lifespan limit? Is 122 years, 164 days – attained two decades ago by Jeanne Louise Calment, the world’s oldest known individual to date – the upper age limit for humans? Or, perhaps, has the person already been born who might live to age 150 or more?
These questions will surely be answered in the future. Today, however, we can confirm the alteration in the shape of the survival curve, called ”squaring” (or “rectangularizing”), which means that no one has yet surpassed Mme. Calment, but more are reaching age 100 and the number of the world’s supercentenarians – people living past their 110th birthdays – is rising fast.
Figure 1 shows the survival curve squaring in Spain from 1977 to 2017, and shows the percentage of survivors for each age band, using Instituto Nacional de Estadistica (Spanish Statistical Office) data.
Figure 1 – Survivors by age band in Spain
Source: Spanish Statistical Office
Main causes of death in Spain
Figure 2 depicts the main causes of death in Spain in 2015. The three most frequent causes –circulatory system conditions, cancerous tumors, and respiratory diseases – caused approximately 70% of mortality that year.Figure 2 – Main mortality drivers in Spain
Source: Spanish Statistical Office
If a cure for cancer were to be found, mortality rates due to cancer would surely decline, yielding a modified distribution of causes of death. Life expectancies of newborns would also increase.(Life expectancy in Spain is already age 83 as of 2016, up from age 79 in 1996.) Might a cure also make it possible for people to age beyond 122 years 164 days? Perhaps, but right now any increase in longevity is unlikely to be significant, which would continue to favor the survival curve squaring.
Can longevity risk be estimated?
For a better understanding of longevity risk, it may be broken down into three sub-risks:
- Base risk: the probability there will be an increase or decrease in the death or disability rates because average mortality or morbidity risk assessments are incorrect.
- Trend risk: the uncertainty about long-term trends (generally calculated using past trends).
- Volatility risk: a measure of the randomness in the mortality or morbidity claim amounts in the near future.
Risk mitigation and data processing
Actuaries have a range of tools and techniques to mitigate these risks.
- For base risk, for example, more data can improve results.
- For trend risk, assess the impact of qualitative changes such as medical advances or lifestyle alterations.
- For volatility risk, ensure adequate portfolio diversification and underwriting risk selection, and use advanced statistical modeling and analysis techniques for valuation.
From a statistical and modeling point of view, it is imperative to collect good historical data to calculate mortality improvement factors. Mortality is not static: it is dynamic and behaves differently in different generations. Using a static mortality table (broken down by age) as an assessment tool can be improved materially by using generational tables, which have a base death rate table and incorporate improvement factors that differ by age, generation and calendar year.
Finally, the data must clearly indicate whether it comes from the general population or the insured population. Each has different behaviors, and population data can be risky to use to project estimates in insured populations. Therefore, selection adjustments are made.
Measurement and valuation
Figure 3 shows three factors involved in the probability of death at different ages:
- Child mortality;
- Accidents at young ages; and,
- Mortality due to disease in more mature ages
Figure 3 – factors involved in the probability of death at different ages.
Source: RGASource: RGA, Human Mortality Database
Today there are advanced models that allow us to estimate longevity risk with more accuracy. In particular, among others we can highlight the best known: Lee-Carter, P-splines, Whittaker-Henderson, Renshaw-Haberman, and even specific models for advanced ages where there is very little data, such as Kannistö. As an example, the Lee Carter model fits the Spanish reality for ages over 45, but for older ages, it is no longer reliable.
Which one to use? It depends on the portfolio that one wants to project into the future. Each model suits a particular need; however, the model to use should be simple model to apply.
How to manage longevity risk from a risk and economic capital point of view?
Risk management is not the same as the economic capital management.
- Risk management is a structured approach to identifying, analyzing, measuring and responding to all possible risk factors, in such a way that the fluctuation relative to a threat can be managed.
- Economic capital is the unexpected maximum loss in a determined future period (in Solvency II, 12 months) and a given confidence level (in Solvency II, 99.5%). This means that the greater the risk assumed, the more economic capital is needed. Economic capital measures risk using economic realities rather than accounting and regulatory rules, and is thought to give a more realistic representation of a firm's solvency.
Risk itself is not a negative, but it must be managed. One way to do so is by managing economic capital, which measures what would happen in an adverse scenario of remote risk. The European solvency framework (Solvency II) establishes a volume of minimum own funds to cover this remote scenario.
Solvency II’s longevity shock figure of 20% for the risk is high. Indeed, this onerous economic capital requirement may be harming annuity sales. The shock requires insurers to allocate more assets to Own Funds to mitigate the impact of having policies in their portfolios for individuals with longer life expectancies than initially expected. The required valuation of the longevity shock, as can be seen in Figures 4 and 5, is a single, immediate and permanent reduction of 20% of expected mortality. This would be approximately equivalent to eradicating more than 65% of deaths due to circulatory causes, or reducing cancer mortality by 70% in Spain.
Source: GRM95
Solutions for managing risk and economic capital are not the same.
The question insurers need to ask in relation to their annuity portfolios and the decision on the solution is: what is the source of the concerns? Should longevity-driven capital requirements be reduced, or should 100% of future longevity exposure be covered? It is not the same to cover all longevity risk or cover only a remote risk. The optimal solution may differ.
