The Disposition Effect on Optimal Stopping Decisions: a Direct Test∗

This paper develops a laboratory test of a distinctive prediction of reference-dependent preferences, testable only in dynamic settings: decision makers suboptimally delay realizing disappointing outcomes (procrastination) but suboptimally rush to realize outcomes that are better-than-expected (rushing). In the experiment, subjects invest in a risky asset, whose price evolves in near-continuous time, and they are provided with the option to liquidate it at a fixed salvage value. Optimal behavior is characterized by an upper and a lower stopping thresholds in the asset price space, thus producing a clear rational benchmark and eliminating known confounds. Most subjects indeed tend to delay liquidating losing assets beyond the optimal point and to sell winning assets before reaching the optimal stopping time. Among subjects who show the effect, the median stopping points imply the probability of realizing a winner conditional on stopping is 70% larger than optimal. Such behavior is shown to be consistent with a model of a decision maker who evaluates payoffs relative to an expectation-based reference point, is risk-averse over gains and risk-seeking over losses.