Ibbotson’s Default Premium: Risky Data

Posted: 29 Jul 2011 Last revised: 23 Aug 2014

See all articles by Winfried G. Hallerbach

Winfried G. Hallerbach

Fintelligence CCT; EDHEC Business School - Department of Economics & Finance

Patrick Houweling

Robeco Asset Management

Date Written: December 9, 2011

Abstract

Ibbotson’s “Stocks, Bonds, Bills and Inflation” data set is widely used because it provides monthly US financial data series going back to as early as 1926. In this data set, the “default premium” is calculated as the difference between the total returns on long-term corporate bonds and long-term government bonds. This excess return is used in empirical research to represent the compensation for default risk exposure. In this paper we show that this default premium is seriously flawed in two ways: (1) it is not based on subtracting maturity-matched government bonds from corporate bonds, and therefore contaminated with a considerable interest rate component, and (2) it is based on very high quality corporate bonds and hence rather insensitive to market-wide changes in default risk. These maturity and quality biases seriously limit the use of the Ibbotson default premium series in empirical research, because instead of reflecting pure default risk, it also (negatively) reflects interest rate risk.

Keywords: Ibbotson, default premium, corporate bonds, long-term data series

JEL Classification: C8, E32, E43, G12, N22

Suggested Citation

Hallerbach, Winfried George and Houweling, Patrick, Ibbotson’s Default Premium: Risky Data (December 9, 2011). The Journal of Investing, Summer, Vol. 22 No. 2, pp. 95-105, Available at SSRN: https://ssrn.com/abstract=1898178 or http://dx.doi.org/10.2139/ssrn.1898178

Winfried George Hallerbach (Contact Author)

Fintelligence CCT ( email )

Salernes, Var 83690
France

EDHEC Business School - Department of Economics & Finance ( email )

France

HOME PAGE: http://https://www.edhec.edu/

Patrick Houweling

Robeco Asset Management ( email )

Rotterdam, 3011 AG
Netherlands

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
2,685
PlumX Metrics