Behavioral Biases in Annuity Choice: An Experiment

Abstract

We conduct a neutral-context laboratory experiment to systematically investigate the role of the hit-by-bus concern in explaining the annuitization puzzle: the low rate of retirement-asset annuitization relative to the predictions of standard models. We vary endowed asset (annuity vs. stock of wealth vs. no explicit endowment), and find a strong endowment effect. Furthermore, we find that the ordering of survival risks matters. Compared to a frame in which a single draw from a known distribution determines survival outcome, annuity choice is lower when subjects must sequentially survive early periods to reach periods in which the annuity dominates. We conclude with policy implications.