TY - JOUR T1 - Factor Investing in Corporate Bond Markets: <em>Enhancing Efficacy Through Diversification and Purification!</em> JF - The Journal of Fixed Income DO - 10.3905/jfi.2019.1.074 SP - jfi.2019.1.074 AU - Thomas Heckel AU - Zine Amghar AU - Isaac Haik AU - Olivier Laplénie AU - Raul Leote de Carvalho Y1 - 2019/10/03 UR - https://pm-research.com/content/early/2019/10/03/jfi.2019.1.074.abstract N2 - We show that factors from value, quality, low risk, and momentum styles play an important role in explaining the cross-section of corporate bond expected returns for the US and Euro Investment Grade and US BB-B Nonfinancial High Yield universes. We demonstrate the importance of purifying factor data by neutralizing a number of risk biases that are present in the factors: controlling for sectors, option-adjusted spread (OAS), duration, and size biases significantly increase the predictive power of style factors. We propose a new simple approach for efficiently neutralizing the biases from multiple risk variables and demonstrate its superiority relative to stratified sampling and optimization as alternative control methods. We also measure the added value from diversifying the number of factors in each style. Finally, we show that the results are robust in relation to transaction costs and can be used to design strategies that aim at outperforming traditional benchmark indexes.TOPICS: Analysis of individual factors/risk premia, factor-based models, style investingKey Findings• Factors from value, quality, low risk and momentum styles play an important role in explaining the cross-section of corporate bond expected returns for the US and Euro Investment Grade and US BB-B non-Financial High Yield universes.• The forecasting efficacy of style factors increases significantly if biases such as sectors, OAS, duration and size in the factor data are neutralized. Diversifying the number of factors in each style also significantly improves the forecasting efficacy.• We propose a new simple approach for increasing the forecasting efficacy of style factors by efficiently neutralizing the biases from multiple risk variables. We demonstrate the superiority of this approach over stratified sampling and optimization. ER -