Time Preference: Experiments and Foundations∗

In order to elicit discount functions, experiments commonly analyze how subjects trade-off money and time. However, discounting reveals itself directly only in behavior obtained by fixing the money dimension and varying the time dimension. This paper presents an experimental procedure that, in a sense made precise, better identifies the discount function. The procedure is based on a general theorem that characterizes the set of discount functions and utility indices compatible with any ‘regular’ preference over dated rewards. The money-time separability and curvature assumptions that are common in the literature are not required. ∗Dept of Economics, Boston University, 270 Bay State Road, Boston MA 02215. Email: [email protected]. The author is grateful to Larry Epstein, Bart Lipman and Faruk Gul for very useful discussions, to Glenn Harrison and Daniel Schunk for detailed comments and help getting to know the experimental literature, and to Qiang Wang for expert research assistance. Helpful comments from seminar participants at Boston University, Caltech, UC San Diego, UC Santa Barbara, UC Irvine, Johns Hopkins and NYU and participants at the Canadian Economic Theory Conference (2010) and European Summer Symposium in Economic Theory (2010) are also gratefully acknowledged. The usual disclaimer applies.