Skip to main content

Stochastic Mortality and the Dispersion of Subjective Estimates of Survival Probabilities – Evidence from 11 European Countries

Final Report Summary - DISPERSION (Stochastic Mortality and the Dispersion of Subjective Estimates of Survival Probabilities - Evidence from 11 European Countries)

For the past several decades, the industrialized world has experienced rapid improvements in life expectancies and survival rates. The annual rates of these improvements exhibit considerable variation. The erratic paths of the survival rates reflect the underlying complex interaction of mortality determinants such as medical innovation, nutrition habits, or environmental factors (e.g. weather and climate) whose progress and impact over time are non-deterministic. The resulting uncertainty about future survival rate improvements, that is stochastic mortality (also called longevity risk), adds an important systematic component to the life span uncertainty that individuals need to consider when planning their life-cycle consumption and savings.

Theoretical models indeed suggest that stochastic mortality is an important determinant of individual saving behavior as well as of decisions on asset allocation and retirement timing. This project analyzes whether individuals are actually aware of stochastic mortality and, if so, whether this awareness affects their actual saving behavior. The projects analyzes survey data on subjective survival expectations and savings indicators elicited from more than 26,000 individuals in the survey of health, ageing and retirement in Europe (SHARE) and corresponding life table data from the Human Mortality Database.

A positive relationship between stochastic mortality and the dispersion of survey responses is found. Such link between the objective dispersion of a variable (here: mortality) and forecaster dispersion (here: subjective estimates) has been extensively tested in other fields. Based on that literature and numerous statistical tests and econometric analyses we conclude that there is evidence for that individuals are to some extent aware of stochastic mortality. A comparison of SHARE savings indicators against the predictions a life-cycle model with uncertain income and stochastic mortality shows that individuals do not save more on average when faced with stochastic mortality- although the model suggests they should. That means, that the positive relationship between stochastic mortality and the dispersion of survey responses found is rather resulting from disagreement effects than from true awareness of uncertainty.

These findings have implications for public policy and regulation. In most European countries pension systems are financed by pay-as-you-go mechanisms. Currently these countries are facing a major demographic transition, characterized by decreasing fertility rates and increasing longevity. The resulting shrinkage of the working-age population in relation to pension benefits recipients puts serious financial strain on pay-as-you-go pension systems and imposes enormous challenges in many other areas for almost all developed countries. Many developed countries have undergone a shift from pay-as-you-go to individually managed defined contribution (DC) pension plans or consider to doing so. The success of such plans depends on individuals making informed saving decisions based on a correct assessment of the involved risks, including stochastic mortality. The project's findings on saving behavior highlights that communication and education regarding stochastic mortality should be improved.