Time-Varying Time Preferences
Recommended citation: Holloway et al. (WP). "Time-Varying Time Preferences." WP.
Abstract:
Dynamic inconsistency in intertemporal choice has long been considered a hallmark of non-exponential discounting. Recent work has challenged this view from a variety of perspectives, including the view that time variance –shifting preferences between measurement dates– can also explain apparent preference reversals. While a nascent literature identifies time-variance and demonstrates its role in explaining time-inconsistency, we lack both a model that allows time-variance to tractably interact with other properties of time preference, and a longitudinal study of sufficient depth to identify such a model. In this paper, we develop the ``nested exponential’’ discount function which is general with respect to time-invariance, time-consistency, and stationarity. The function nests both exponential discounting and a version of present-biased discounting within its parameter space, enabling transparent model selection at both the aggregate and subject levels. We evaluate time-invariance and the performance of the nested exponential model in a 12-week longitudinal study featuring seven surveys. Our elicitations give us unprecedented precision in estimating dynamic inconsistency, non-stationarity, and time-variance. We find that subjects in our study exhibit significant decreasing patience over the course of the study, and that time-variance explains roughly 72% of time-inconsistent choices in our data. This does not mean our data are best-explained by exponential discounting plus preference drift: hyperbolicity is a key feature of our data, and it is well captured by the nested exponential function.