GDP-Linked Bonds as a New Asset Class

Abstract

We show that GDP-linked bonds can provide diversification benefits to investors. We use a stochastic spanning methodology which makes no assumptions on the distri- butional characteristics of the returns of these novel instruments and test both floaters and linkers. None of these types of GDP-linked bonds are spanned by a broad bench- mark set of stocks, bonds, and cash assets, thus providing a new asset class. Spanning is ruled out for a wide range of bond design parameters. Out-of-sample testing doc- uments economically and statistically significant diversification benefits for investors, with increase in Sharpe ratios 0.10-0.43 for floaters and 0.05-0.17 for linkers over the benchmark portfolio. The benefits depend on the risk premium for linkers, whereas floaters are less sensitive to the premium, but benefits persist for a wide range of premia estimated in existing literature and are robust to a randomized test. We also document the finance and macro factors that drive GDP-linked bonds performance, using generalised method of moments regressions.

Keywords: stochastic spanning, contingent debt, risk premium, diversification