The Danger of Longevity Looms? Bonds Can Come to the Rescue

Recent developments in pension reform are oriented toward contribution plans having a definite duration. In such a context, pensions take the form of a lifetime annuity based on payments made by the insured before pension age, which is paid until the individual dies. If the lifetime of the individual is longer than that what was expected at the moment when contributions were paid, insurance companies and government institutions who offer pension plans face a particular type of risk: the risk of longevity. Longevity risk has an individual and an aggregate component. The individual component depends on the fact that anyone can live longer than the average of individuals in his/her age group, while the aggregate component depends on the evolution of life expectancy across the whole population. The individual component can be diversified by offering varied pension plans to wide groups of people having different mortality risks. But the aggregate component cannot be diversified by a wide portfolio, precisely because it affects all individuals. The existence of an aggregate component of longevity risk poses three questions. How much weight does this component have? How do you measure the aggregate risk of longevity? Data tell us the aggregate component of longevity risk is important. The probability of dying within one year after 65 years of age has drastically dropped for the overall population over the last forty years. But such reduction has not been uniform. UK data say that improvements in the postponement of mortality have been more drastic for 70-year olds. In 1971 a 65-year old individual had a 81% probability of being alive at 69, a 34% probability of being alive at 79, and a 5% probability of being alive at 89. These probabilities have changed to 92%, 64 %, and 20 %, respectively, in 2009. It thus important to investigate how mortality will evolve in the future. Stochastic mortality models enable us to measure longevity risk by forecasting future mortality for a given level of statistical confidence. The width of confidence intervals defines mortality risk. The forecasting simulation of a factorial model explaining mortality says that in 2020 the probability that a live individual of 65 years of age will reach the age of 90 has an expected value of 25%, but the 95% confidence interval of such an expected value is comprised between 17% and 33%. Now, the fact that 33% rather than 25% of 65-year olds could be alive at 90 exposes social security institutions and insurance companies paying lifetime annuities to considerable risk. The issue is how to diversify such risk. A proposal emerged in the economic literature, which has occasionally been implemented, is to create a market for "longevity bonds", i.e. assets indexed to longevity. Longevity bonds are financial products paying a maturity indexed to actual mortality for a given group of individuals. A longevity bond issued in 2012 indexed to the mortality of 65-year olds in 2012, pays a maturity proportional to the actually observed mortality from 2012 onwards of those people having 65 years of age in the starting year. If 65-year olds have a survival rate above expectations in 2012, the maturity will be correspondingly higher in the same year. This mechanism allows the buyer of longevity bonds to have in its portfolio an instrument to diversify longevity risk. The thorny issue linked to longevity bonds concerns what institution should ideally issue them. One school of thought remarks that such bonds are born to solve a market failure, generated by the fact that future generations are excluded from insurance markets. In such a context, collective welfare improves with government intervention, because through the power of taxation governments can redistribute across generations. This argument points to governments being the natural issuers of longevity bonds. But the problem is being still debated for two reasons: governments may not perform optimal intergenerational redistribution because present generations vote and future generations do not, and because the issuance of longevity bonds would increase the exposure of governments to longevity risk.

The risk of longevity thus poses interesting questions at the crossroads of demography, forecasting, finance, and public economics.