Taylor Rule Based Exchange Rate Models with Wealth Effects

Student thesis: Doctoral ThesisPhD

Abstract

This thesis focuses on the relationship between the exchange rate and its determinants using an endogenous monetary policy rule as represented by the Taylor rule. Compared to the recent literature on out-of-sample exchange rate predictability, I extend the model of Molodtsova and Papell (2009) by including two variables representing wealth effects, as has been suggested in the standard Taylor rule models. Using quarterly data from 1975-2008, I first investigate the econometric properties of the Taylor rule applied to U.K., Australian and Swedish data against the US dollar. Various unit root tests indicate that variables commonly used in such models are likely to be integrated of order one. However, by accounting for structural breaks, I can conclude that all variables are stationary. Parameter estimates suggest wealth effects are strongly related to the nominal exchange rates in these countries, in contrast to the standard monetary variables. Furthermore, I evaluate short-horizon exchange rate predictability with the Taylor rule fundamentals model for the U.S. dollar against the Australian dollar, Swedish Krona and British Pound. Following the recent literature, a robust set of out-of-sample statistics, including the Clark and West statistic, Diebold-Mariano statistics and Theil’s U ratio are used to evaluate the forecast performance. Current results from the Theil’s U ratio and CW statistics shows the Taylor rule incorporating the wealth effect improves the short run exchange rate forecast performance. Finally, we model the exchange rate from 1975 to 2008 as a Smooth Transition Regression (STR) based model in which a series of economically meaningful transition variables drive the movement across exchange rate regimes. The overall findings show strong evidence supporting the nonlinear relationship between the exchange rate and economic variables. Moreover, the STR Taylor rule models of the exchange rate substantially outperform both the random walk model and the linear Taylor rule model in forecasting the exchange rate.
Date of Award24 Jun 2015
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorBruce Morley (Supervisor), Javier Ordonez (Supervisor) & Michalis Stamatogiannis (Supervisor)

Cite this

'