Forecasting the Term Structure of Interest Rates Near the Zero Bound - A New Era?

Abstract

We investigate the forecasting performance of popular dynamic factor models of the yield curve after the global financial crisis (GFC). This time period is characterized by an unprecedented low and non-volatile interest rate environment in most major economies. We focus on the dynamic Nelson-Siegel model and regressions on principal components and use a dataset of monthly US treasury bond yields to show that subsequent to the GFC both models are significantly outperformed by the random walk no-change forecast. Especially for short and medium term yields the random walk is up to ten times more accurate. Interestingly, these results are not picked up by traditional global forecast evaluation metrics. We show that combining forecasts mitigates the model uncertainty and improves the disappointing forecasting accuracy especially after the GFC.