| Retrieval Failures and Consumption Smoothing: A Field Experiment on Seasonal Poverty was authored by Ned Augenblick, B Kelsey Jack, Supreet Kaur, Felix Masiye and Nicholas Swanson. It was published in Q.J. Econ. in 2026. |
Do failures to recall known future expenses contribute to undersaving and seasonal poverty? This study tests whether prompting people to think through upcoming expenses—without giving new information—changes remembered spending and real financial behavior. Using a randomized field experiment among farmers in Zambia and a replication of belief effects among low-income Americans, the authors examine whether associative prompts reduce retrieval failures, raise savings, and smooth consumption over the hungry season.
Field randomized intervention Measured remembered expenses and actual savings Replication among low‑income Americans
The primary evidence comes from a randomized field experiment conducted with Zambian households whose income derives from a single annual harvest. The intervention asked treated participants to list and think through future expenses in associative categories; it did not provide external budgeting guidance or new information. The paper reports changes in self‑reported ‘‘remembered’’ expenses, short‑term spending, measured household savings six weeks after treatment, and subsequent crop revenue in the following year. Belief effects were also tested in a separate sample of low‑income Americans.
Treated individuals reported substantially higher remembered future expenses (reported increases of 36%–60% in the abstract), with larger effects for small, irregular, and stochastic items. Immediate spending fell after the prompt and treated households held higher measured savings six weeks later (reported as 15% higher). Those households entered the hungry season with roughly one additional month of savings and a flatter spending profile across the year. The increased savings were followed by greater self‑financed farm investment and a reported 9% increase in next year’s crop revenue. The authors also report that the intervention reproduced its effect on beliefs in a low‑income American sample, which the abstract presents as evidence of generalizability.
