Impact of Covariance Matrix Estimator Choice on Yield Curve Decomposition With PCA

8 Pages Posted: 26 Dec 2023

Date Written: November 26, 2023

Abstract

This investigation evaluates the impact of different covariance matrix estimation methods when used as an input for principle component analysis in the context of yield curve decomposition, focusing on the economic interpretations of Level, Slope, and Curvature components. It was found that the constant correlation shrinkage estimator consistently outperforms others, including the sample covariance matrix. The investigation also reveals that short time periods yield less reliable results, with the issue attributed to the length of time sampled rather than sample sizes.

Keywords: PCA, principle component analysis, yield curve, covariance, constant correlation, yield curve decomposition

Suggested Citation

Holtes, Grant, Impact of Covariance Matrix Estimator Choice on Yield Curve Decomposition With PCA (November 26, 2023). Available at SSRN: https://ssrn.com/abstract=4644700 or http://dx.doi.org/10.2139/ssrn.4644700

Do you have negative results from your research you’d like to share?

Paper statistics

Downloads
42
Abstract Views
179
PlumX Metrics