TY - JOUR T1 - Forecasting Swap Spreads: <em>A Bayesian Approach</em> JF - The Journal of Fixed Income SP - 40 LP - 53 DO - 10.3905/jfi.2016.26.2.040 VL - 26 IS - 2 AU - Daniel Klein AU - Elena Nikitina AU - Jean-Christophe Curtillet Y1 - 2016/09/30 UR - https://pm-research.com/content/26/2/40.abstract N2 - In this article, we analyze the determinants of swap spreads applying Bayesian model search (BMS). BMS is a flexible methodology that explicitly accounts for model risk and has become increasingly popular in the financial forecasting literature. We apply BMS to explore the data-rich environment of the candidate explanatory variables to predict the dynamics of the Euro area swap spreads using monthly data from 2000 to 2015 within a linear regression framework. Our empirical results suggest that the BMS approach shows consistently superior forecast performance compared to a model selection approach based on the Schwarz information criterion and a naïve benchmark model.TOPICS: Derivatives applications, big data/machine learning, developed ER -