Exponential Growth Bias: Theory and Experiments

Download data and study materials from OSF

Principal investigators:

Joshua Tasoff

Claremont Graduate University

Email: joshua.tasoff@cgu.edu

Homepage: https://sites.google.com/site/joshtasoff/

Matthew Levy

London School of Economics and Political Science

Email: m.r.levy@lse.ac.uk

Homepage: http://www.lse.ac.uk/economics/people/facultyPages/MatthewLevy.aspx


Sample size: 518

Field period: 5/1/2012-5/30/2012

Abstract

Exponential-growth bias (EGB) is the tendency for individuals to partially neglect compounding of exponential growth. We develop a model wherein biased agents misperceive the budget constraint, and derive conditions for overconsumption and dynamic inconsistency. We construct an incentivized measure of EGB in a US-representative population and find substantial bias, with approximately one-third of subjects estimated as the fully-biased type. The magnitude of the bias is negatively associated with asset accumulation, and robust to a simple graphical intervention.

Hypotheses

What is the extent of EGB in the U.S. population?

What are the financial correlates of EGB?

Can EGB be mitigated with a simple interactive graph typical of many investment websites?

Experimental Manipulations

A fraction of the subjects received a simple interactive graph typical of many investment websites.

Outcomes

EGB as measured by subject responses on financial questions that involve compounding interest.

Summary of Results

We find vast EGB. One third of Americans do not compound interest. Furthermore, EGB is highly correlated with assets. Going from full bias to full accuracy is associated with a ceteris paribus 55–90% increase in accumulated assets. The graphical intervention did not have a significant effect on EGB.

Additional Information

The TESS portion of our study is Experiment 2 of a three-experiment paper.

References

Levy, Matthew, and Joshua Tasoff. 2016. "Exponential-Growth Bias and Lifecycle Consumption." Journal of the European Economic Association 14(3): 545-583. DOI: 10.1111/jeea.12149