City Research Online

Abstract

Factor analysis has contributed imperatively towards solving the dimensionality problem and identifying the underlying factor structure governing term structure of interest rates. The estimated latent factors are known as level, slope, and curvature. These factors can be estimated using Principal Component Analysis (PCA) or Nelson-Siegel (1987) framework as reparameterized by Diebold and Li (2006). The two statistical methods have been shown to produce the same three factors.

The thesis contributes towards testing of level, slope, and curvature factors extracted using the statistical models. We investigate the issues of stability in the eigenspace variables governing level, slope, and curvature. We develop a stability testing procedure to examine the presence of significant structural changes in the latent factors estimated using PCA. Bootstrapped critival values have been employed in order to draw inferences. Monte Carlo evidence suggests good finite sample size and power properties of the tests. Empirical test results on zero coupon bond yield curves show significant structural changes in factors.

Further, we propose some extensions to estimating level, slope, and curvature factors for term structures where the interest rate maturities are coupled together into dependence clusters. In this, we extend the Nelson-Siegel (1987) framework to the case of modelling yield curves with correlation clusters. We identify the short maturity and long maturity clusters governing the term structure and propose a block dynamic representation to model the factors. We find that the proposed model generated superior forecasts than the benchmark model proposed by Diebold and Li (2006).

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HG Finance
Departments:	Bayes Business School > Finance Doctoral Theses Bayes Business School > Bayes Business School Doctoral Theses

Export

Downloads