A model of exchange rate innovation that accounts for short and long term determinants

This paper develops a model of exchange rate determination within an error correction framework. Using daily data, the intention is to identify both short and long-term determinants that can be used to forecast the AUD/USD exchange rate. Specifically, the overnight interest rate differential, Australia's foreign trade-weighted exposure to commodity prices and exchange rate volatility all provide explanatory power for the AUD/USD exchange rate over the post float period 1984-2004. When forecast out of sample, the error correction model is found to perform better then a naïve random walk model based on three different metrics.