Template-Type: ReDIF-Paper 1.0
Series: Tinbergen Institute Discussion Papers
Creation-Date: 2026-02-26
Number: 26-007/III
Author-Name: Sicco Kooiker
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Janneke van Brummelen 
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Julia Schaumburg 
Author-Workplace-Name: Vrije Universiteit Amsterdam
Author-Name: Marcin Zamojski
Author-Workplace-Name: Vrije Universiteit Amsterdam
Title: Self-driving neural networks for term structure modeling
Abstract: We propose a factor model with time-varying loadings for term structure modeling and forecasting. While maintaining the interpretation of the factors as level, slope, and curvature through explicit identification restrictions, we allow the loadings to take flexible shapes by specifying them as neural networks that evolve over time using a “self-driving” updating scheme based on past forecast errors, with gradient scaling to improve robustness. Using an empirically calibrated simulation study and an application to U.S. Treasury yields across 24 maturities, we show that flexible and dynamic factor loadings improve forecasting performance relative to standard benchmarks, including Nelson-Siegel models and the random walk. The gains are strongest at medium maturities and shorter forecast horizons, highlighting the importance of capturing curvature dynamics. In-sample results further illustrate how time-varying loadings provide insight into changes in yield curve shape beyond traditional parametric specifications.
Classification-JEL: C38, C45, E43
Keywords: time-varying neural networks, observation-driven dynamics, yield curve
File-URL: https://papers.tinbergen.nl/26007.pdf
File-Format: application/pdf
File-Size: 22.729.135 bytes
Handle: RePEc:tin:wpaper:20260007