Simulating Yield Curves with Dynamic Nelson-Siegel Factors

Summary

The document asks how to generate evolving yield curves for evaluating bond hedges. It explains that the Nelson-Siegel curve is a static representation on its own, then points to the Diebold-Li approach: interpret its factors as level, slope, and curvature, and model those factors over time with autoregressive processes. The simulated factor paths can then be translated into yield curves at different maturities for a hedging analysis.

The answer cautions that simple short-rate models may not capture enough of the curve’s shape for bond strategy evaluation. It offers a multifactor Heath-Jarrow-Morton model as a more involved alternative for simulating the term structure. The response is a brief direction to relevant modeling approaches rather than a worked calibration or simulation procedure; it gives no empirical comparison, parameter choices, or hedge performance evidence. Model selection therefore depends on the desired curve dynamics and the complexity an analyst can support.

Key ideas

• Nelson-Siegel provides a static description of the yield curve unless its factors are given time dynamics.
• The Diebold-Li method treats the factors as level, slope, and curvature and evolves them with autoregressive models.
• Simulated factor values can be mapped into maturity-specific yields for bond hedge evaluation.
• Simple short-rate models may not represent the full yield curve shape needed for strategy analysis.
• A multifactor Heath-Jarrow-Morton model is a more elaborate alternative for term-structure simulation.

Tags

# How can I simulate the Yield Curve?


# How can I simulate the Yield Curve?












I would like to simulate the Yield Curve and Yield Curve changes and then use this data to evaluate bond hedging strategies. I certainly need a model like Nelson-Siegel, but how can I simulate changes in a model and how can I get the data from the model, so I can use it for my hedging analysis. Thanks for any help / advice!

## Answer by David Duarte (score 1, accepted)

https://quant.stackexchange.com/a/55176

Nelson-Siegel is a static analysis, but you can use Diebold Li's approach. Check out Forecasting the term structure of government bond yields (2005), Francis X. Diebold, Canlin Li. Basically, they use the fact that the factors can be interpreted as level, slope and curvature, and then use autoregressive models to make the curve evolve in time.

Apart from that, because you want to evaluate bond strategies, simple short rate models will probably not give you enough information on the curve shape, but if you feel brave enough, you can try to use a multi factor HJM model to simulate a term structure throught time

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.