Working Paper: NBER ID: w16868
Authors: John Beshears; James J. Choi; David Laibson; Brigitte C. Madrian
Abstract: Many previous experiments have found that participants invest more in risky assets if they (i) see their returns less frequently, (ii) see portfolio-level returns (rather than individual asset-by-asset returns), or (iii) see long-horizon (rather than one-year) historical asset class return distributions. In contrast, we find that such information aggregation treatments do not increase equity allocations in an experiment where—unlike previous experiments—participants invest in real mutual funds over the course of one year. In a follow-up experiment, we start with the classic Gneezy and Potters (1997) experimental design and modify it step-by-step to move closer to the design of our first experiment. Using this identification strategy, we show that previously documented aggregation effects are not robust to (i) changes in the distribution of the risky asset’s returns and (ii) the introduction of a multi-day delay between the initial portfolio choice and the realization of returns.
Keywords: portfolio risk; information aggregation; behavioral finance
JEL Codes: D14; G11
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| aggregated returns disclosure (G12) | portfolio risktaking (G11) |
| aggregated returns disclosure does not increase equity allocations (G11) | portfolio risktaking (G11) |
| historical return graphs (N22) | equity investments (G12) |
| type of historical returns shown (G12) | equity investments (G12) |
| return distribution alteration (D39) | aggregation effects (E10) |
| delay in return realization (C41) | aggregation effects (E10) |