CMStatistics 2018: Start Registration
View Submission - CFE
A1540
Title: Forecasting and trading monetary policy effects on the riskless yield curve with regime switching Nelson-Siegel models Authors:  Massimo Guidolin - Bocconi University (Italy) [presenting]
Manuela Pedio - Bocconi University (Italy)
Abstract: The aim is to investigate whether accounting for regime shifts enhances the forecasting power of a Dynamic Nelson-Siegel (DNS) model in which exogenous factors that interact with the dynamics of regime shifts are used to capture different aspects of the monetary policy conducted by the Federal Reserve, with special emphasis to the emergency measures adopted in response to the 2008-2009 financial crisis. We also implement a set of systematic trading strategies relying on the forecasts of our models to evaluate the economic values of both a baseline DNS with and without regimes and of an extended econometric framework in which DNS, regimes, and factors capturing the monetary policy stance are interacted. The empirical results suggest that the Markov switching structure generally improves the out-of-sample forecasts of the single-state model, particularly during the crisis. Moreover, we find that the models augmented with a variety of macroeconomic variables significantly improve the predictability of different features of the yield curve. This is especially true during the crisis, coherently with the stated purposes of unconventional monetary policies. Finally, we obtain evidence that switching DNS models augmented to include monetary policy variables is economically beneficial in risk-adjusted terms.