Skip to content
All library documents

Log Discount Factor Interpolation and Forward-Rate Robustness

Article Quant Q&A · Author: JoeBass

Summary

This document asks why yield curve construction may interpolate the logarithm of discount factors and what robustness means in that context. It relays a documentation claim that linear interpolation of log discount factors keeps instantaneous forward rates constant between curve grid points. The claim also says this approach can succeed in borderline cases where other interpolation methods fail during curve creation.

The text does not explain the mathematical reason behind the relationship or define the failure cases that make the method robust. It provides no worked example or comparison with alternative interpolation schemes. The core topic is the link between discount factor interpolation and the implied forward-rate behavior, with further study needed to understand the numerical stability claim and its limits.

Key ideas

  • Linear interpolation of log discount factors is described as producing constant instantaneous forward rates between grid points.
  • The cited documentation characterizes this interpolation approach as robust in some borderline curve construction cases.
  • The document asks why robustness arises but does not provide a derivation or examples.
  • Comparisons with other yield curve interpolation methods are not included.

Tags

Full text
# Why is the log of the discount curve used?


# Why is the log of the discount curve used?












I was reading here: SAP S/4HANA Software Documentation

> Calculation Methods for Interpolating the Yield Curve The option Linear Interpolation of Log( Discount Factor ) means the system keeps instantaneous or differential forward rates constant between the grid points. This method is very robust and can also work in borderline cases where the other calculation methods lead to an error during yield curve creation.

I vaguely recall seeing this in Bloomberg, also. What is the robustness that is mentioned above?

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.