Inferring discount rates from time-preference experiments

We observe that identification of the discount rate from experimental data requires an assumption about the consumption period, the length of time over which a payment will be turned into utility-providing consumption. We show that the optimal consumption period is substantially longer than assumed in previous studies. When the consumption period is allowed to take on more reasonable values, the discount rates implied by experimental choices are unreasonably large and relatively insensitive to assumptions about utility curvature.