Yield curve forecasting using the Nelson-Siegel model. A comparison of ARIMA, VAR and Random Forest approaches – evidence from the USA
Yield curve forecasting using the Nelson-Siegel model. A comparison of ARIMA, VAR and Random Forest approaches – evidence from the USA

Summary/Abstract: The aim of this article is to compare different approaches to forecasting the US yield curve factors derived using the Nelson-Siegel (NS) model. Using daily US swap yield data from 1990 to 2026, we assess the Autoregressive Integrated Moving Average (ARIMA), Vector Autoregression (VAR) and Random Forest (RF) models in a 1-day-ahead and 1- to 20-day-ahead forecasting competition. The principal finding of this study is that ARIMA significantly outperforms the RF in forecasting the NS model factors, as does VAR, although only in terms of the Level and Curvature factors. The results of this study thus suggest that the use of machine learning methods, in the case of the US yield curve, is not always superior.

• Details
• Contents