Modelling exchange rate movements in South Africa: an ARDL application

Abstract

This study examined the monetary, unbiased forward rate hypothesis (UFRH), and random walk models of exchange rate in South Africa using the auto regressive distributed lag (ARDL) approach. The estimated results show that there is a long-run relationship between variables in each model, suggesting the examined models are able to predict the exchange rate. In the long run, this paper found a sticky-price monetary model with correct signs but restrictions on money supply are violated. However, both the UFRH restrictions and random walk model were upheld. The models were then used to predict forecasts of the out-of-sample period as compared to the random walk. We found that UFRH outperformed other models. In this context, forward rate is predominantly an unbiased estimator of the future spot exchange rate, suggesting that the South African exchange rate market is efficient.