Robust Investing in Corporate Bonds with Factor Investment Styles

By Martin Nybye, Mikael Venø and Michael Holte, Corporate Bond Portfolio Team, Jyske Capital

This paper discusses factor investment styles for corporate bonds and the benefits of combining the factors in a non-technical way. We will show that a factor approach can create robust returns in a corporate bonds universe – even in different credit cycles without losing the upside. We will show evidence for excess performance of the single factor investment style, but also evidence for robust excess performance if the factors are combined1. The model was Implemented at Jyske Capital in May 2012. In-production results from the model will be used in the discussion.

Background

Factor investment styles pioneered by Fama and French have become widespread tool for investing in equities and the academic research on the topic is extensive. Opposite to equity factor models the litterateur on factor models for corporate bonds is quite scares.

One of the reasons for the scarcity of litterateur is that corporate bonds are more complex instruments than equities. There is usually only one stock price and its maturity is infinite. However, a company can issue bonds with different maturities, call features, different covenants and different collateral structures. All factors that makes it more difficult to model the price of a bond. Even getting the correct price data of a corporate bond is not straight forward compared to the stock market due to the smaller liquidity in the corporate bond market.

However, the complexity of corporate bonds and the lack of attention from research side makes factor style investing in corporate bonds potentially more attractive then equity factor style investing. The risk that factor styles get arbitraged away is significantly smaller for corporate bonds.

The paper is organized the following way. Section 2 discusses the model and the three single factor investment styles value, momentum and quality. The factor investment styles are combined in the end of the section. Section 3 validates the model based upon in-production results. Section 4 concludes.