Can an Experiment Shed Light on the Decision to Annuitize?
Lifetime annuities or simply life annuities—either immediate or deferred—sold by life insurance companies have never been popular with Americans. There is now a huge literature on the subject of this lack of popularity, which economists often refer to as the annuity puzzle. Some research has focused on possible behavioral biases that may tip the choice away from annuities. One source of bias could be the reaction by would-be annuitants to the way the decision to annuitize is presented or framed. For example, when the decision is presented as an investment decision, annuitization is less popular than when it is presented as a way of insuring against the financial consequences of an unexpectedly long life. Another important source of bias, one that arises when participants in a retirement plan have a choice between taking their savings out as an annuity or as a lump sum, is the plan’s default setting. Thus, if annuitization requires the conscious decision to reject the plan’s standard of a lump-sum distribution, fewer participants may end up with an annuity than would be the case if participants had to opt in favor of a lump sum when an annuity was the default setting.
The 2012 article by Robert Gazzale, Sandy Mackenzie (the author of this Insight), and Lina Walker reports on an experiment that examines the potential role of the default setting, and also addresses the potential attractiveness of a deferred-income annuity or longevity annuity. Experiments are a relatively new development in empirical economics. The experiment described in this article involved about 220 participants aged 22 to 70 who were working either full time or part time. The experiment was designed to have a working-life phase and a retirement phase. In the working-life phase, participants earned tokens by performing various tasks on their computers. The total number of tokens each earned was his or her retirement nest egg. The retirement period was divided up into four subperiods, with “survival” from one period to the next being determined randomly.