The impacts of elicitation mechanism and reward size on estimated rates of time preference

• We estimate discount rates from experiments using real monetary rewards.
• Discount rates are significantly higher for smaller rewards.
• These results are consistent with a magnitude effect.
• We compare discount rates from a matching task with those from a choice task.
• Discount rates do not differ between the two tasks when modeling behavioral noise.

We run experiments with real monetary rewards ranging from $10 to $500 to estimate rates of time preference and test for hyperbolic discounting. Individuals become more patient with increasing reward sizes, which is consistent with a magnitude effect. This magnitude effect is robust across specifications including a nonparametric analysis and structural maximum likelihood estimation. Subjects are divided between two different elicitation mechanisms (one a matching task and one a choice task) that should both theoretically provide an incentive for participants to reveal their true time preferences. We find some evidence of differences between the rates from the matching and choice tasks but these differences disappear when appropriately modeling the behavioral noise. We uncover little to no evidence of present-biased time preferences.