Experimental Research on Retirement Decision Making: Evidence from Replications

We adapt the design of five experimental studies on retirement decision making and conduct replications with a larger sample from the broader population. We replicate most of the main effects of the original studies. In particular, we find that consumption decisions are less efficient when subjects need to borrow from the future than when they need to save from the present. When subjects collect retirement benefits as lump sum instead of annuities, they choose to retire later. The duration of retirement affects the saving behavior of the subjects. Savings are higher when they are incentivized with matching contributions than when incentivized with tax rebates. When faced with varying survival risk, subjects make partial adjustments to spending paths. We also propose a further experimental research agenda in related topics and discuss practical issues on subject recruitment, attrition, and redesign of complex tasks.