Return Predictability in the Treasury Market: Real Rates, Inflation, and Liquidity

Abstract

Estimating the liquidity differential between inflation-indexed and nominal bond yields, we separately test for time-varying real rate risk premia, inflation risk premia, and liquidity premia in U.S. and U.K. bond markets. We find strong, model independent evidence that real rate risk premia and inflation risk premia contribute to nominal bond excess return predictability to quantitatively similar degrees. The estimated liquidity premium between U.S. inflation-indexed and nominal yields is systematic, ranges from 30 bps in 2005 to over 150 bps during 2008-2009, and contributes to return predictability in inflation-indexed bonds. We find no evidence that bond supply shocks generate return predictability. There is wide consensus among financial economists that returns on nominal U.S. Treasury bonds in excess of Treasury bills are predictable at different investment horizons. Predictor variables include forward rates (Fama and Bliss, 1987), the slope of the yield curve (Campbell and Shiller, 1991), and a linear combination of forward rates (Cochrane and Piazzesi, 2005). There is however an ongoing discussion about what drives this predictability. We contribute to this discussion by conducting a joint empirical analysis of the sources of excess bond return predictability in nominal and inflation-indexed bonds in the U.S. and the U.K. Specifically, we provide evidence that both time-varying inflation risk premia and time-varying real interest rate risk premia are quantitatively important in explaining time variation in the expected excess return on nominal bonds. The question whether expected excess returns of inflation-indexed bonds are time-varying is important on its own. It remains relatively unexplored, partly due to the short history of U.S. inflation-indexed bonds (Campbell, Shiller, and Viceira, 2009). We contribute to the emerging research on inflation-indexed bonds by examining and distinguishing among three potential sources of excess return predictability in inflation-indexed bonds: time-varying real interest rate risk, timevarying liquidity risk, and market segmentation. Our work differs from recent work on inflationindexed bond return predictability by Campbell, Sunderam, and Viceira (2013) and Pflueger and Viceira (2011) in that it finds and corrects for an economically significant liquidity risk premium, which otherwise would likely distort estimates of real rate and inflation risk premia. Novel and unique to our work is the finding that a liquidity component in breakeven inflation, or the spread between nominal and inflation-indexed bond yields of similar maturity, predicts the return differential between nominal and inflation-indexed bonds due to liquidity. The estimated U.S. return differential due to liquidity exhibits a significantly positive stock market CAPM beta, suggesting that U.S. TIPS investors bear systematic liquidity risk and should be compensated in terms of a positive return premium. While the return differential due to liquidity is not directly