Forecasting exchange rates in the frequency domain

Abstract

Since a long a time that the exchange predictability is a hot topic for finance practitioners and researchers. In this dissertation we study the exchange rate predictability using a method that has never been applied in the literature: the frequency domain. The uncovered interest rate parity, studied by Fisher (1896) was the model selected for this investigation. Instead of the original time series applied in Rossi (2013), we applied the Faria and Verona (2017) methodology in the Rossi (2013) framework. The frequency-decomposed predictor method tested in the interest rate differential model, does not improve the exchange rate predictability across the sample and time horizon selected. This conclusion come from a horse race analysis of different exchange rates, different filters and different frequencies.

Date of Award	17 Dec 2018
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Gonçalo Faria (Supervisor)