Skip to content
All library documents

Diebold–Li Estimation as Fixed-Shape Nelson–Siegel Regression

Article Quant Q&A · Author: HeinKK
Read the original

Summary

The note clarifies that Diebold–Li is a way to estimate a Nelson–Siegel yield curve, rather than a separate curve model with its own distinct estimation procedure. The key modeling choice is to fix the Nelson–Siegel shape parameter; with that parameter held constant, the remaining coefficients can be estimated from a cross-section of yields using linear regression.

It points to software that supplies the Nelson–Siegel factor loadings used in this setup, while distinguishing that support from a dedicated Diebold–Li parameter-estimation function. This is a concise conceptual answer to a software question, not a full implementation guide: it gives no code, calibration details, data requirements, or comparison of alternative packages. It also does not discuss how the shape parameter should be chosen or how estimation performance should be assessed.

Key ideas

  • Diebold–Li estimation uses the Nelson–Siegel yield curve specification.
  • The method fixes the Nelson–Siegel shape parameter.
  • The remaining curve coefficients can be estimated by linear regression on cross-sectional yields.
  • Software that constructs Nelson–Siegel factor loadings can support this estimation approach.

Tags

Fixed incomeStatistics
Full text
# Is there a packages to estimate Diebold-Li Parameters in R?


# Is there a packages to estimate Diebold-Li Parameters in R?












"The models implemented are: Nelson-Siegel, Diebold-Li and Svensson" is written in the description of the packages "YieldCurve". However, I can't find a specific estimation function for the diebold-li parameters. How can I fix this problem with this packages or do you know other packages for the estimation?

## Answer by Enrico Schumann (score 2, accepted)

https://quant.stackexchange.com/a/55648

Diebold/Li is not a separate yield-curve model: for a given cross-section of yields, they fix the shape parameter in Nelson-Siegel and then estimate the remaining parameters via linear regression. See Calibrating the Nelson-Siegel-Svensson Model. The function `NSf` in package in package NMOF, which I maintain, sets up the factor loadings for Nelson-Siegel, as used by Diebold/Li.

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.