In several OECD countries, public pay-as-you-go financed pension systems have undergone major reforms in which future retirement benefit promises have been scaled down. A consequence of these reforms is that especially in countries with a tight tax-benefit linkage, the retirement benefit claims of low-income workers might not even exceed the minimum income guarantee which the government provides the aged. Recently, some German politicians have criticized this likely development because it was unjust that persons who have paid contributions over a long working life end up with no higher benefits than people who have never worked or paid any contributions. However, the government defended the current retirement benefit formula with the argument that every Euro paid as contributions had exactly the same value in generating future retirement benefits. But this logic has been questioned recently, e.g. by Breyer and Hupfeld (2009), since the value of a contributed Euro depends on the life expectancy of the individual, which is positively correlated with annual income. In that earlier paper, we introduced the concept of "distributive neutrality", which takes income-group-specific differences in life expectancy into account. The present paper estimates the relationship between annual earnings and life expectancy of German retirees empirically and shows how the formula that links benefits to contributions would have to be modified to achieve distributive neutrality. We compare the new formula to the benefit formulas in other OECD countries and analyze a data set provided by the German Pension Insurance Office on a large cohort of pensioners to find out how the old-age poverty rate would be affected by the proposed change of the benefit formula. Finally, we discuss other possible effects of a change in the benefit formula, especially on the labour supply of different earnings groups.