February 2023
Steven Hamilton (George Washington University)
Geoffrey Liu (Harvard University)
Tristram Sainsbury (Australian National University)
Abstract: During the COVID-19 pandemic, the Australian government allowed eligible individuals to withdraw up to A$20,000 (around half median annual wage income) across two tranches from their retirement accounts, ordinarily inaccessible until retirement. Based on historical returns, the modal withdrawal by the modal-aged withdrawer can be expected to reduce their balance at retirement by more than $120,000 in today’s dollars. One in six working-age people withdrew a total of $38 billion (on average, 51% of their balances). These transfers represented a liquidity shock and were much larger than those considered in the literature to date. Using administrative and weekly bank transactions data, we find a high marginal propensity to spend (MPX) given the size of the transfers of at least 0.43 within eight weeks, spread broadly across categories (including around half or more on non-durables) and across withdrawers. The response to the second withdrawal, which two-thirds returned for and which occurred after activity had recovered, was even larger at 0.48. Withdrawal and spending are predicted strongly by numerous measures of poor financial health, high pre-withdrawal rates of cash withdrawal and gambling, and younger age. The MPX of rational, forward-looking but liquidity constrained consumers can be expected to asymptote to zero as the transfer size rises, while that of present-biased consumers can be expected to remain high. Our findings overwhelmingly are consistent with the latter, suggesting roughly 80% of withdrawers were present-biased. In selecting strongly on the present-biased, the program presents a sharp trade-off between effective macroeconomic stimulus and suboptimal retirement saving policy.
JEL Codes: E21, E63, E71, H31, H55, J32
Key Words: Stimulus, retirement saving, marginal propensity to consume, present bias
