Shane Frederick, George Loewenstein, and Ted O'Donoghue, “Time Discounting and Time Preference: A Critical Review.” Journal of Economic Literature 40(2): 351-401, June, 2002.
• Samuelson’s discounted utility model (DU), 1937: utility is additively separable across time, with future utility discounted exponentially at a constant rate.
• Anomalies with respect to DU are not mistakes that people want to correct.
• DU doesn’t permit a preference for increasing utility flows over time, holding total discounted utility constant.
• DU assumes that utility in a given period is independent of utility in other periods. Your preference for Italian or Thai food tonight is independent of what you ate yesterday.
• The discount rate in the DU model is the same when applied to all forms of consumption. The per-period discount rate is fixed, implying time consistency.
• Empirical evidence is not kind to the DU model, on many fronts.
• Discount rates don’t appear to be constant; rather, they decrease over time. Further, they seem to vary widely across different types of consumption.
• Gains are discounted more than losses, and small amounts are discounted at a higher rate than larger amounts. Many people like to incur a loss immediately rather than delay it.
• Preferences for increasing returns over time might reflect a fear of limited self-control that would lead to overconsumption if larger amounts were received initially.
• The beta/delta or quasi-hyperbolic model (page 366) formalizes a present bias and dynamic inconsistency. This model can indicate why people might use illiquid assets as a commitment mechanism, and explain the simultaneous existence of substantial savings and credit card debt. Sophisticated present-biased people might seek out commitment strategies.
• Other models involve habit formation, anticipation utility, visceral factors, projection bias, mental accounts, and temptation utility.
• Perhaps there is no isolated concept of time preference that is consistent across applications.
No comments:
Post a Comment