Gold returns dynamics and forecast

Abstract

This study uses frequency-domain approaches to forecast gold returns, considering the high-frequency, business cycle-frequency, and low-frequency of each variable. The low-frequency component of the Equity Risk Premium, extracted using wavelet filtering, emerges as a strong predictor of gold returns, outperforming the other variables, and remaining stable over a 24-year out-of-sample period. The predictive power of different variables for gold returns during expansive and recessive business cycles is also explored. The low-frequency component of the Equity Risk Premium emerges as the most significant predictor during economic expansions. In contrast, the high-frequency component of the Default Yield Spread exhibits greater predictive power during recessions. These findings demonstrate how different frequencies of macroeconomic predictors can individually capture important information regarding the future dynamics of gold returns.

Date of Award	18 Dec 2023
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Gonçalo Faria (Supervisor)