The authors have been compensated to present academic research at events hosted by financial institutions that administer retirement savings plans. See the authors ’ websites for a list of outside activities. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. At least one co-author has disclosed a financial relationship of potential relevance for this research. Further information is available online at

The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research. At least one co-author has disclosed a financial relationship of potential relevance for this research. Further information is available online at

We would like to thank Liran Einav for his helpful comments and Alex Christodoulou for helping us assemble the data for this project. This work was supported by several grants including a supplement

havior and NIA grant awards R01AG021650, P01AG005842, and P30AG034532. The content is the sole responsibility of the authors and does not represent the official views of NIA, NIH, or the NBER. The authors have been compensated to present academic research at events hosted by financial institutions that administer retirement savings plans. See the authors ’ websites for a list of outside activities. What is the socially optimal level of liquidi-ty in a retirement savings system? Liquid re-tirement savings are desirable because liquidi-ty enables agents to flexibly respond to pre-retirement events that raise the marginal utility of consumption, like medical emergencies or income shocks.1 On the other hand, pre-retirement liquidity is undesirable when it leads to under-saving arising from, for exam-1

Personal retirement accounts (PRAs), especially 401(k) plans, have become an increasingly important mode of retirement saving. This paper provides new evidence on the effect of the recent financial crisis, and the associated decline in employment, on PRA saving. We particularly examine how these effects vary across demographic groups. We explore how crisis-related changes in employment and earnings affected PRA balances. We do this by estimating the effect of the crisis on these outcomes and then by considering how PRA ownership and balances depend on employment and earnings as well as other covariates. To assess the effect of the crisis we estimate the relationship between age (and other covariates) and the labor market and PRA outcomes in years prior to the crisis (2004-2006) and then estimate how these relationships change during the crisis period (2008-2010). We find very few statistically significant differences in the parameter estimates for the pre-crisis and the crisis periods. We use the model to predict age profiles of employment rates, earnings given employment, PRA ownership, and PRA balances given ownership in the pre-crisis and crisis periods. We give special attention to the relationship between education and PRA ownership and balances.