Measuring Valuation of Liquidity with Penalized Withdrawals

We propose using penalized withdrawals from retirement savings accounts, identified from U.S. tax records, as a revealed-preference tool to characterize households' valuation of liquidity. A simple dynamic model formalizes the notion that the prevalence of withdrawals can be used to characterize American households' valuation of liquidity over time and space. We find pervasive evidence of high valuation of liquidity, hence that shocks are imperfectly insured. Declines in households' income lead to sudden, large, and persistent jumps in the probability of penalized withdrawals. Both local economic conditions and persistent household characteristics play an important role, with the average valuation of liquidity being higher in financially underdeveloped areas as well as in black communities which are plausibly marginalized from the credit market. Finally, applying our tool to the Great Recession, we find that more affected areas saw larger increases in penalized withdrawals, plausibly driven by tightening of local credit conditions. Our analysis offers a new tool to study the valuation of liquidity and our results point to sizable welfare gains from social insurance policies targeted at both households and locations over time.