Models of Intertemporal Choice: How Humans Discount Delayed Rewards (2009)

Undergraduate: Evgeniya Bakunova

Faculty Advisor: Jonathan Hill
Department: Economics

---

This paper attempts to find a model that best describes intertemporal discounting. The basic question asked is, how do people value a set amount of money at a delay—for example, how much do you value $100 if it is promised in 10 days? If you’d prefer $100 immediately, it means that it is more valuable now. There are a number of discount functions in the literature that predict how much you value today what you’ll get tomorrow. To answer the question of which model is the best, a survey is administered in an online game to collect two indifference curves from each individual in the sample (n=18). The procedure is tested by simulation and is used to collect behavioral data. One-curve-at-a-time analysis shows that the indifference curves are linear, supporting the hyperbolic discounting model from neuroscience. But when two curves from one individual are examined at once, they do not display the behavior predicted by any of the models. Specifically, each formula from the literature suggests that an indifference curve of the next order of magnitude should be an order of magnitude steeper. This is not observed. Instead, the slopes are much closer to each other. This leads to the proposition of new models that can describe the change in slopes better. These models are tested on the pairs of indifference curves. Lastly, all models and additional explanatory variables are tested on the master indifference curve pooled from all indifference curves in the sample.