O’Donoghue and Rabin (2015) on Present Bias

Ted O’Donoghue and Matthew Rabin, “Present Bias: Lessons Learned and To Be Learned.” American Economic Review 105(5): 273–279, 2015 [pdf].

• In extending the standard exponential discounting model to incorporate present bias, the β, δ functional form has proven to be “useful, tractable, and (importantly) disciplined [p. 273].” Further, the β, δ approach seems to correlate well with the psychological findings, in that most of the action in terms of changing discount rates over time concerns right now versus the future. [For more on the quasi-hyperbolic, β, δ approach, see this Behavioral Economics Outlines post.] 

• If β is less than 1, and hence the individual displays present bias, we still need to inquire as to whether the person comprehends that she is present biased. A person who fully understands her taste for instantaneous gratification is termed “sophisticated,” whereas someone who fails to understand her present bias – she repeatedly says, and believes, that she will quit smoking tomorrow – is said to be “naïve.” People who recognize that they are present biased but underestimate the extent of their bias are partially sophisticated or partially naïve. 

• Without uncertainty or liquidity constraints, choices among monetary streams should be made by maximizing present value at market interest rates: individual preferences and discount rates are irrelevant. In low-stakes experiments, people are unlikely to be liquidity constrained; therefore, those choices should not depend on discounting. For this reason, recent experiments exploring present bias try to use real effort flows, not monetary streams. 

• Pairs of decisions, such as those involving credit card borrowing along with those involving retirement savings, can inform the calibration of present bias. The idea is that credit card purchases are influenced by present bias, whereas retirement decisions reflect long-run (β=1) preferences. 

• Welfare assessments are possible despite the non-unitary (over time) actors that present bias reflects; long run (that is, β=1) preferences have much to recommend them as the welfare criterion [example here]. 

• Don’t rush to “explain” heterogeneous behavior by different degrees of present bias. Habit persistence or tastes for tobacco probably explain more variance in smoking than does different degrees of present bias, for instance.