Abstract
The classic normative model for intertemporal preferences uses a constant discount rate, but behavioral experiments have shown that people do not tend to make choices consistent with a constant discount rate. We first present normative models for time preferences, then discuss descriptive results for choices between single outcomes occurring at different times. People often have different implicit discount rates for different types of scenarios, and several anomalies arise consistently. We contrast those results with findings for preferences for sequences of multiple outcomes over time. People tend to prefer increasing or constant sequences of outcomes over time, especially when the outcomes are non-monetary in nature. This suggests a willingness to wait for improvement, but not in the way that classical discounting would prescribe. We end with prescriptive nudges to improve dynamic consistency and pose questions still to be resolved about how decisions involving outcomes over time can be improved prescriptively.