Essays on bond return predictability and liquidity risk

If there is valuable information for predicting bond prices over time, how can we use this information to improve investor's risk-return trade-off and term structure modelling? This thesis aims at answering this question. The first chapter discusses the predictive role of alternative measures of the liquidity premium of TIPS relative to Treasury bonds for government excess bond returns. Results show that the liquidity premium predicts positive (negative) TIPS (nominal Treasury) excess returns. I also find that the out-of sample forecasting power of liquidity for nominal Treasury excess returns appears to have been addressed by the events during the recent financial crisis. By contrast, I have evidence of out-of- sample forecasting ability during both normal and bad times for TIPS’ excess returns. In the second chapter, I explore whether or not the TIPS liquidity premium can be considered as an unspanned factor that forecast bond returns, but that is not necessarily spanned by the U.S. yield curve. I consider a joint Gaussian affine term structure model for zero-coupon U.S. Treasury and TIPS bonds, with an unspanned factor: liquidity risk. In the model, the liquidity factor is restricted to affect only the cross-section of yields but it is allowed to determine the bond risk premia. I present empirical evidence suggesting that the liquidity factor does not affect the dynamic of bonds under the pricing measure, but does affect them under the historical measure. Consequently, the information contained in the yield curve appears to be insufficient to completely characterize the variation in the price of curvature risk. In the third chapter, I estimate the non-parametric optimal bond portfolio choice of a representative agent that acts optimally with respect to his/her expected utility one period forward, provided that he/she observes the ex-ante liquidity signal. Considering alternative measures of liquidity, I find that the liquidity differential between nominal and TIPS bonds appears to be a significant determinant of the portfolio allocation to U.S. government bonds. In fact, conditional allocations in risky assets decrease as market liquidity conditions worsen, and the effect of market liquidity decreases with the investment horizon. I also find that the bond return predictability translates into improved in-sample and out-of-sample asset allocation and performance.