Measuring intertemporal preferences using response times

We use two different approaches to measure intertemporal preferences. First we employ the classical method of inferring preferences from a series of choices (subjects choose between $X now or $Y in D days). Second we adopt the novel approach of inferring preferences using only response time data from the same choices (how long it takes subjects to choose between $X now or $Y in D days). In principle, the inference from response times should work, since choices between items of nearly equivalent value should take longer than choices between items with substantially different values. We find that choice-based analysis and response-time-based analysis yield nearly identical discount rate estimates. We conclude that response time data sheds light on both our revealed (choice-based) preferences and on the cognitive processes that implement those preferences.