Savage for Dummies and Experts

Overview & Synopsis

We present a novel method-called risk equivalents-that uses a single measure to elicit discount rates while avoiding concerns about the shape of the utility function. The method is valid under discounted expected utility (DEU), and also under several of its behavioral extensions including more general models that account for a biased perception of time and risk (such as time-or likelihood-insensitivity). We implement the method in a field experiment in India measuring inter-temporal discount rates for money and the consumption of tea.We empirically observe that discount rates elicited by traditional methods are related to utility curvature, whereas discount rates elicited by risk equivalents are not. Risk equivalents also mitigate differences in discount rates measured for money and for tea, suggesting that unaddressed utility curvature may play a role in results that demonstrate good-specific discounting. Risk equivalents are general, fast and tractable, three features that are particularly useful in field studies.