Forecasting the Equity Risk Premium: The Ups and the Downs

Abstract

Recent academic advances in predicting the market equity risk premium (ERP) employ constraints motivated by economic theory in order to generate non-negative ERP forecasts. We show that the improved accuracy of these constrained models is primarily due to the fact that the equity markets have exhibited positive average excess returns across the business cycle. However, we find that in recessionary periods, in periods of negative ERP realisations and in periods of increased volatility, they generate larger forecasts errors than a standard unconstrained linear model. This finding constitutes a significant challenge for the practical benefit of constrained forecasting models when used by a mean-variance investor who allocates dynamically between the equity market and a risk-free asset; we show that these models generate significant economic benefits over a long sample period, from 1927 to 2013, but generate substantial losses during recessionary and volatile periods, exactly when it matters the most. Our results have important implications for the design of new ERP forecasting models.