ASFEE 6 in Paris

Ambiguous Survival Probabilities and the Demand for Annuities: An Experimental Test through Charitable Giving
Giuseppe Attanasi  1@  , Hippolyte D'albis  2@  , Emmanuel Thibault@
1 : Bureau d'économie théorique et appliquée  (BETA)  -  Website
université de Strasbourg, CNRS : UMR7522, Université Nancy II
P E G E 61 avenue de la Forêt Noire 67085 STRASBOURG CEDEX -  France
2 : Université Paris 1
Université Paris 1

In this paper, ambiguity aversion to uncertain survival probabilities is introduced in a life-cycle
model with a bequest motive to study the optimal demand for annuities. Provided that annuities
return is sufficiently large, and notably when it is fair, positive annuitization is optimal in the
ambiguity neutrality limit case. Conversely, the optimal strategy is to sell annuities in case of
infinite ambiguity aversion. Then, in a model with smooth ambiguity preferences, there exists a
finite degree of ambiguity aversion above which the demand for annuities is non-positive. To
conclude, ambiguity aversion appears as a relevant candidate for explaining the annuity puzzle.
We tested our theoretical results through a laboratory experiment. First, a subject's coherentambiguity
attitude has been elicited in a simple experimental setting able to make the smooth
ambiguity model operational. Then, in a bond-annuity two-period decision problem, the subject's
bequest in the second period has been presented as a contingent donation – contingent to surviving
after the first period – to a previously chosen charity. We found that coherent-ambiguity-averse
subjects invest less in annuities than coherent-ambiguity-neutral ones, and that the donation to the
chosen charity is increasing in the investment in annuities. These findings confirm our theoretical
predictions.


Online user: 2