Myopic loss aversion, bond returns and the equity premium puzzle

Abstract

In an influential paper Bernatzi and Thaler (1995) (B&T) show that Myopic Loss Aversion (MLA) can explain the equity premium in the US over the period 1926 to 1990. However, bond returns, in their simulations, are based on coupons only. Allowing for capital gains on bonds in the simulations yields results that are somewhat different from those obtained by B&T. Furthermore, the simulations reveal another asset market puzzle related to the demand for bonds of long duration.

Acknowledgements

Helpful comments and suggestions from a referee are gratefully acknowledged. Keeli Hennessey provided excellent research assistance.

Notes

¹ In the equity premium literature the long bond rate is often used as the risk-free rate because the covariance between long rates and consumption growth is close to zero and that is what defines riskiness within the neoclassical framework. McGrattan and Prescott (2003) advocate using long bonds to address the equity premium puzzle. They argue that individuals do not hold short bonds for their retirement.

² We follow the parameter values used by B&T, which are in turn based on experiments among students. Haigh and List (2005) find that MLA is more pronounced among professional traders than students.

³ The software can be downloaded from http://cran.r-project.org/ along with tutorials and programming examples.

⁴ From http://www.nber.org/databases/macrohistory/contents/chapter13.html series m13002.

⁵ From www.GlobalFinancialData.com series IGUSA10M.

⁶ B&T estimate an evaluation horizon of 13 months.

⁷ Although not mentioned by McGrattan and Prescott (2003) municipal bonds are still tax exempt.

⁸ However, Durand et al. (2004) find that evaluation horizon at which the investor is indifferent between short bonds and equity, is likely to be shorter than the one estimated by B&T. They show that the null hypothesis of shorter intersection horizons cannot be rejected at conventional significance levels, while the null hypothesis is rejected for longer horizons.

⁹ A fraction that typically depends on the expected sharp ratio for equities relative to bonds and investors risk tolerance.