Pensions and differential mortality in France

In France, there is a long tradition of studies that focus on the relation between occupational groups and mortality while the relation between income and mortality has, to our knowledge, not yet been measured. In this paper we use the "Echantillon Inter Régimes des Retraités", a sample of retired people, to estimate how mortality rate is related to the amount of retirement pension. These data come from administrative registers on pensions and do not suffer from any measurement error nor from any problem of attrition. They make it possible to obtain a robust measure of differential mortality. Our results indicate that male mortality rates are significantly and negatively linked to the level of retirement pension. We estimate that the elasticity of the mortality rate to the pension is between -0.62 and -0.14 for workers between 67 and 91 year old and that the elasticity of life expectancy at age 60 to pensions is about 0.18. Results for women are found to be not significant. However our sample, which includes only people who received a pension, is not representative of the female population. In order to illustrate our results we discuss how estimates of differential mortality according to pension levels could be used for economic studies about pensions. We estimate the redistribution that is generated by differential mortality. We find that, due to differential mortality, the redistribution effect of the French pension system is reduced from a quarter to a half. We also compute how to determine pensions so that lifetime benefits are unaffected by the age at which people start receiving their pension, and show how it should be modified to account for differential mortality.