Return Predictability in Federal Funds Futures: Is it There?

Abstract

This paper studies return predictability in federal funds futures. I show that over the period 1990 to 2018, predictor variables from the literature do not consistently outperform the expectations hypothesis when evaluated out-of-sample. Further, while forecasts from advanced forecasting methods such as Dynamic Model Averaging and Complete Subset Regressions are considerably more accurate than those from simple linear prediction models, they do not generate systematic economic value to investors. These results suggest that federal funds futures do not need adjustment for time-varying risk premia.