The Derivation and Application of a Theoretically and Economically Consistent Version of the Nelson and Siegel Class of Yield Curve Models

A popular class of yield curve models is based on the Nelson and Siegel (1987) (hereafter NS) approach of fitting yield curve data with simple functions of maturity. However, NS models are not theoretically consistent and they also lack an economic foundation, which limits their wider application in finance and economics. This thesis derives an intertemporally-consistent and arbitrage-free version of the NS model, and provides an explicit macroeconomic foundation for that augmented NS (ANS) model. To illustrate the general applicability of the ANS model, it is then applied to four distinct topics spanning finance and economics, each of which are active areas of research in their own right: i.e (1) forecasting the yield curve; (2) investigating relationships between the yield curve and the macroeconomy; (3) fixed interest portfolio management; and (4) investigating the uncovered interest parity hypothesis (UIPH). In each application, the ANS model allows the formal derivation of a parsimonious theoretical framework that captures the essence of the topic under investigation and is readily applicable in practice. Respectively: (1) the intertemporal consistency embedded in the ANS model results in a vector-autoregressive equation that projects the future yield curve from the current yield curve, and forecasts from that model outperform the random-walk benchmark; (2) the economic foundation for the ANS model leads to a single-equation relationship between the current shape of the yield curve and the magnitude and timing of future output growth, and empirical estimations confirm that the theoretical relationship holds in practice; (3) the ANS model provides a theoretically-consistent framework for quantifying risk and returns in fixed interest portfolios, and portfolios optimised ex-ante using that framework outperform a passive benchmark; and (4) the ANS model allows interest rates to be decomposed into a component related to economic fundamentals in the underlying economy, and a component related to cyclical influences. Empirical tests based on the fundamental interest rate components do not reject the UIPH, while the UIPH is rejected based on the cyclical interest rate components. This provides empirical support for suggestions in the theoretical literature that interest rate and exchange rate dynamics associated with cyclical interlinkages between the economy and financial markets under rational expectations may contribute materially to the UIPH puzzle