Learning and Visceral Temptation in Dynamic Savings Experiments

Abstract

In models of optimal savings with income uncertainty and habit formation, people should save early to create a buffer stock, to cushion bad income draws and limit the negative internality from habit formation. In experiments in this setting, people save too little initially, but learn to save optimally within four repeated lifecycles, or 1-2 lifecycles with "social learning." Using beverage rewards (cola) to create visceral temptation, thirsty subjects who consume immediately overspend compared to subjects who only drink after time delay. The relative overspending of immediate-consumption subjects is consistent with hyperbolic discounting and dual-self models. Estimates of the present-bias choices are β=0.6-0.7, which are consistent with other studies (albeit over different time horizons).

Files

Additional details