Updated paper
Standard economic theory often treats preferences as time-invariant, yet little is known about their stability at high (daily) frequency in representative samples. The present study fills this gap by daily eliciting risk and time preferences simultaneously of 2266 individuals using a Convex Time Budgets design, alongside simpler measures. Our design allows us to study day-to-day dynamics within the same individuals and relate them to an exogenous, plausibly salient signal: the daily change in national COVID-19 hospitalizations during 2020. We find that risk aversion, time consistency, and patience move at a daily frequency and co-vary with this signal, while aggregate, wavelevel comparisons imply stability and simpler elicitation methods do not detect any variation. Our findings are consistent with fear- and uncertainty-based mechanisms. Robustness checks address alternative mechanisms, controls, and specifications. We conclude that preferences can vary systematically over short horizons, and that finegrained elicitation is essential to detect this variation.