- 25 Apr 2017
It will be worth it, in the end
25 Apr 2017 - 12:00 pm - 1:00 pm
Most descriptive intertemporal choice models proceed by altering the exponential discount function, to a hyperbolic function or other forms. We proceed differently, by assuming that intertemporal choice errors are driven by exponential discounting until a given future date, S, the, “Subjective time horizon.” Utilities beyond S are incorrectly given zero weight (when S < T, the agent’s “Objective time horizon”). We explore this model’s correspondence with the empirical literature on intertemporal choice errors. Crucially, all errors are moderated by S, with errors disappearing as S approaches T. First we demonstrate that only a finite number of intertemporal choice errors are possible in this model. Two errors involve only a single intertemporal prospect (which is either incorrectly favoured or avoided compared versus inaction). Six further errors involve choosing a smaller sooner over a larger later reward. A simple consumption model demonstrates that setting S < T never benefits the agent. We further show that temporal inconsistency can be explained via a diminishing cost of increasing S at further delays. The cost of increasing S further explains within-agent variation in intertemporal choice, via size, loss, and length effects. Debiasing is moderated via S, providing a mechanistic explanation for previous experimental findings.
Technical University of Munich
TUM School of Management