Forecasting Exchange Rate Across Countries with Gold Price as Exogenous Variable Using Transfer Function and VARI-X Model

Abstract

Investors and collectors hold gold as protection for their savings and wealth at
large. Gold does not pay interest like treasure bonds or savings accounts, but current gold
prices often reflect increases and decreases of an asset. This research aims to provide a
model for the relationship between the exchange rate, which is vital in exporting gold, and
gold prices across countries. The Australia, Brazil, and South Africa exchange rates are
used as a case study against the gold price. The ARIMA model is used for forecasting gold
price as an input for the Transfer Function and VARIX models. The Transfer Function
model only considers the relationship between gold prices as input with the exchange rate
in each country, whereas the VARIX model also considers the interrelationship between
exchange rates in these countries. Daily data is used for the period 1st June 2010 to the
28th February 2018. The RMSE and MAPE are used as criteria for selecting the best
model. The results show that VARIX is the best model for forecasting the Australian
exchange rate, while the Transfer function is the best model for forecasting South African
and Brazilian exchange rates.