Forecasting Bond Risk Premia with Unspanned Macroeconomic Information

Job Market Paper

Rui Liu
University of Houston
September 1, 2016

Abstract

This paper provides out-of-sample evidence for the existence of unspanned macro risks. I propose a novel framework to forecast bond risk premia using a macro-.nance term structure model with unspanned output and in.ation risks. I account for model uncertainty by combining forecasts with and without unspanned macro information optimally from the forecaster.s objective, and I take advantage of the no-arbitrage condition by imposing risk premium restrictions for the purpose of forecasting. Incorporating macro information generates signi.cant gains in forecasting bond risk premia relative to yield curve information at long forecast horizons, especially when allowing for time-varying combination weight. These gains in predictive accuracy signi.cantly improve the utility of a bond investor.