Accurate assessment of the risk that arises from further increases in life expectancy is crucial for the financial sector, in particular for pension funds and life insurance companies. The Dutch Actuarial Association presented a revised projection model in 2010, while in the same year two fundamentally different approaches were published by other institutions. This situation invites study of the consequences that the choice of projection model has on estimates of future life expectancy, which is the purpose of this paper. We firstly compare the three approaches against theoretical findings in the international literature. Secondly, we compare their outcomes in terms of period and cohort survival. In addition, we estimate the impact of each model on the present value of future pension payments. Our results indicate that, even in the short term, remarkable differences in life expectancy occur that also translate into different pension values. The literature review suggests that there is currently no blueprint for mortality projections; that calls for the application of various approaches to discount the uncertainty of the individual models. Instead of relying on extrapolation methods only, the pension sector should also take expert-driven forecasts into account as well as approaches that model causal influences on mortality. The model of the Actuarial Association could be improved by taking cohort influences into account as well as the estimate of uncertainty bounds around the outcome measure. Also, the consistency of the projection in terms of the age and gender dimensions but also other countries should be enhanced.

Netspar - Network for Studies on Pensions, Aging and Retirement, Tilburg
hdl.handle.net/1765/79740
Netspar design papers

Peters, F., Nusselder, W., & Mackenbach, J. (2012). The longevity risk of the Dutch Actuarial Association’s projection model (No. 11). Netspar design papers. Retrieved from http://hdl.handle.net/1765/79740