I compare two different ways of integrating mortality into life-cycle models: the standard additive model with time preferences, on the one hand, and a formulation that rules out the existence of time preference, but allows for risk aversion with respect to the length of life, on the other hand. These models are of similar complexity, but rather different in their fundamental assumptions. I show, however, that the latter formulation can reproduce all the predictions of the additive models with non-negative rates of time preference, as far as life-cycle behaviors under a single exogenous non-degenerate mortality pattern are considered. It leads, nonetheless, to radically different predictions for the effects of mortality changes.
D91 - Intertemporal Consumer Choice; Life Cycle Models and Saving ; D81 - Criteria for Decision-Making under Risk and Uncertainty ; J17 - Value of Life; Foregone Income