A behavioral study of Roth versus traditional retirement savings accounts

Motivated by a popular perception that Roth accounts are welfare-improving for most people, this paper compares the effects of mandated Traditional (tax-deferred) or Roth (taxprepaid) retirement policies in a controlled laboratory setting. Selection effects, which complicate analyses of observational data, are avoided by random assignment to policies. Subjects receive exogenous incomes during "working" periods, followed by no-income "retirement" periods. In each period, subjects decide how many lab dollars to convert into "takehome pay," akin to consumption with diminishing returns. Subjects' decisions determine retirement savings and tax payments. Flat income and tax-rate profiles facilitate the analysis of behavioral factors like present-period tax avoidance, while optimal consumption and after-tax savings are identical for both treatments. Our results show that observed savings are suboptimal in both treatments and are influenced by gender, patience, and risk aversion measures. In contrast to conventional wisdom, there are no significant differences between policies; if anything, the Traditional treatment leads to marginally higher post-retirement consumption.