Hyperbolic Discounting

A pattern of time preference in which people discount the near future more steeply than the distant future, unlike constant exponential discounting.

Source: Laibson 1997

Exponential discounting applies a constant rate between any two adjacent periods, so preferences formed today about a future choice remain the preferences held when that moment arrives. Hyperbolic discounting applies a steeper rate over short horizons than long ones, so the drop between now and next week is larger than the drop between one year and one year plus a week. The consequence is that preferences reverse as the near option approaches, without any new information having been received. The reversal is the defining prediction, since exponential discounting at any rate, however high, cannot produce it, so observing preference reversal without new information is evidence of the departure rather than of impatience.

Source: Laibson 1997

The standard formulation modifies exponential discounting by applying one additional constant factor to every period beyond the present, which makes the immediate period distinctively favoured while all future periods are treated consistently relative to one another. This single extra parameter reproduces preference reversal, and it nests the conventional model as the case where the additional factor equals one, so the departure can be tested rather than assumed. Estimates of the additional parameter vary substantially with the elicitation method, and approaches separating impatience from the curvature of utility have produced smaller estimates than the earlier literature reported.

Source: Laibson 1997

Repeated postponement of actions whose costs fall now and whose benefits arrive later, including starting treatment, attending screening, stopping smoking and beginning to exercise, with each delay individually small and collectively consequential. It also predicts the reverse pattern for behaviours with immediate reward and delayed cost. Crucially it predicts that people will hold and express genuine intentions to act while systematically failing to do so, which distinguishes it from simple lack of motivation. It also predicts that people will pay for commitment devices that restrict their own future options, which is difficult to reconcile with any model in which preferences are stable over time.

Source: O'Donoghue & Rabin 1999

Because it determines what interventions can work. Someone who anticipates their own reversal will seek deadlines, restrictions and arrangements removing the later choice, and can be helped by making such devices available. Someone who does not anticipate it will plan on the assumption of following through, decline any device offered, and postpone repeatedly while continuing to intend otherwise. The two produce identical stated intentions and very different behaviour, so voluntary commitment reaches only the first group. Partial awareness is the common case, where the reversal is anticipated but its strength underestimated, which produces commitment arrangements set too loosely to bind when they are needed.

Source: O'Donoghue & Rabin 1999

Generally not. Guidance continues to specify constant discounting, on the grounds that the social rate applied in appraisal expresses a normative judgement about weighting future generations rather than describing individual behaviour, and that a time-inconsistent social rate would produce appraisals reversing without new information. Hyperbolic discounting therefore enters economic evaluation through the modelling of uptake and adherence rather than through the rate applied to costs and outcomes. The distinction matters because it separates a descriptive finding about behaviour from a prescriptive choice about how a society should weigh consequences falling on people in the future.

Source: Drummond et al. 2015