In intertemporal choice research, discount rates measured from participants are not always correlated with real behavior. One possible explanation for the lack of correlation could be use of poor discount rate measurements. We used a simulation with an ideal decision maker to examine the properties of three popular discount rate measurement procedures. The simulation results revealed a narrow measurable range and a crude resolution for all three measurements. As an alternative, we propose a new discount rate procedure using a three-alternative-forced-choice (3AFC) design with a heuristic algorithm. A simulation showed the new procedure can measure a wide range of discount rates (approx. 0.035% APR to 350,000% APR) with high accuracy under 10 questions. An experiment with human participants confirmed the reliability and validity of the measurement, and showed that less than a minute was needed for a completion of the task. The new method can provide fast and accurate discount rate estimation over a wide range.
Haewon Yoon 