Heterogeneous Default Effects on Retirement Saving: Sledgehammers or Precision Instruments

This paper uses a randomized experiment in a representative sample from the Dutch population to investigate the effect of various defaults on retirement saving. In light of the Dutch pension system, with high contributions that afford generous income replace- ment but no flexibility other than the timing of retirement, we consider defaults that encourage less saving as well as more. The aggregate effects of defaults at deviations from the status quo are symmetric and large: they increase the fraction that either suspends or doubles premium payments for three years by 22 percentage points (pp) on a base of less than 10%. The status quo default is less powerful, raising the fraction by 6–13pp from a baseline around 60%. Rich survey data on demographics and prefer- ences matched with administrative records of wealth and forecasted pensions indicate that the different default effects are driven by different groups. A default of increased saving disproportionately affects those with self-control issues, but has a smaller effect on individuals prone to procrastination. Moreover, this is the only default that inter- acts with variables related to preparedness for retirement, with a larger effect on those with high housing wealth and high adequate expenditure goals. The saving reduction default affects women more strongly and has a weaker effect on university graduates.