From 1927 through 2015, there has been a very large difference between the returns to the S&P 500 and the returns to risk-free Treasury bills—about 8.5% on an annual average basis and about 6.7% on an annualized basis. This large spread is frequently referred to as the equity premium puzzle, because unless investors possess implausibly high levels of risk aversion, the equity premium’s historical average is too high to be justified by standard economic models.
Shlomo Benartzi and Richard Thaler, authors of the study “Myopic Loss Aversion and the Equity Premium Puzzle,” which was published in the February 1995 edition of The Quarterly Journal of Economics, proposed that the puzzle’s answer lies in a behavior known as myopic loss aversion (MLA). MLA describes the tendency of investors who are loss-averse (the pleasure felt after observing a gain is inferior to the pain experienced after a loss of an equivalent amount) to evaluate their portfolios too frequently, thus causing them to take a short-term view of investing (losses are experienced more frequently at narrow time scales).
That, in turn, leads to a focus on the short-term volatility of the market. Consequently, they invest too little in risky assets. Using simulations, Benartzi and Thaler demonstrate that when investors only evaluate their portfolios annually, the size of the equity premium is consistent with parameters estimated in prospect theory.
Rebalancing & Evaluation
Boram Lee and Yulia Veld-Merkoulova contribute to our understanding of investor behavior with their study, “Myopic Loss Aversion and Stock Investments: An Empirical Study of Private Investors,” which appeared in the September 2016 issue of the Journal of Banking & Finance. Following the evidence from prior studies, they use portfolio evaluation frequency and rebalancing frequency as proxies for myopia. The authors then investigated the link between MLA and investor behavior.
Their data sample comes from a survey by CentERdata, a specialized online research institute at Tilburg University in the Netherlands. It consisted of more than 2,000 households including members over 16 years of age and is representative of the Dutch population in respect to a number of important demographic characteristics.
The survey data covered the period 1997 through 2010, and averaged eight years of data per investor. To disentangle the effects of evaluation frequency from rebalancing frequency, they adopted a two-by-two design in which each frequency is divided into two levels (low and high). For both evaluation and rebalancing, frequency of more than quarterly is considered high.
Following is a summary of the authors’ findings:
