Can we Exploit Predictability in Bond Markets?

56 Pages Posted: 21 Mar 2011 Last revised: 27 Feb 2013

Francisco Barillas

Emory University - Goizueta Business School

Date Written: March 15, 2011

Abstract

This paper investigates the optimal bond portfolio choice of an investor in a model that captures both the failure of the expectation hypothesis and recent findings that variables not in the term structure of interest rates drive expected bond returns. I estimate a daily multifactor affine term structure model with a large set of unrevised macroeconomic data in which one of the state variables is unspanned by the contemporaneous yield curve. By characterizing the optimal bond portfolio choice under different information sets I quantify the importance of using information not reflected in the yield curve. I find that investors who only condition on bond prices would pay a significant amount of wealth to observe the (unobservable) risk-premium in real time. This implies that bond-risk premia cannot be filtered precisely solely from fixed-income securities. Instead, I find that investors who condition on macroeconomic variables perform much better and that payrolls, industrial production and gross domestic product are the most relevant to portfolio decision. These results show that expanding the information set to include macroeconomic information that is not already incorporated in prices can substantially increase the ex-ante utility (and ex-post performance) of bond portfolio investors. I document the importance of hedging demands at intermediate levels of risk-aversion and discuss some shortcomings of the affine framework when applied to bond portfolio decisions.

JEL Classification: G11, E44, G12

Suggested Citation

Barillas, Francisco, Can we Exploit Predictability in Bond Markets? (March 15, 2011). Available at SSRN: https://ssrn.com/abstract=1787567 or http://dx.doi.org/10.2139/ssrn.1787567

Francisco Barillas (Contact Author)

Emory University - Goizueta Business School ( email )

1300 Clifton Road
Atlanta, GA 30322-2722
United States

Paper statistics

Downloads
169
Rank
146,464
Abstract Views
840