A principal components analysis of the UK term structure and the influence of fiscal policy
This study examines the use of principal component techniques in analysing term structures of interest rates. It employs original methods of estimating B-Spline models with endogenous knot positions and applies the method of Dierckx (1981) to generate new data sets for the study. The variability of the knots suggests that natural market boundaries do not exist in the UK gilts market. Few, if any, of the previous studies of term structures using principal components have subjected the components to statistical testing. This is remedied in this thesis. The results suggest that only two components, a level and a slope component, are required to describe most of the variability in the term structures irrespective of the data used, but these components are not stable over time. The thesis extends the method to include partial common principal components, and using this method demonstrated the difference in the major components of selected data sets. The thesis found that changes in the principal component scores could not be accounted for by regularly published economic news, including news about the PSBR. A macromodel was estimated. This showed that the term structures in the sample were altered by changes in government spending but the movement in interest rates would depend upon how this was funded and what maturity of interest rates was studied. The model also showed that significant changes would take a long time to manifest themselves and that there was evidence that some forms of funding had unstable effects. These results provide an explanation of why news effects are difficult to discern and why there is no consensus on whether or not fiscal variables affect the term structure.