Identifying Linkages Between the Money and Currency Markets Using Artificial Neural Networks.

This paper designs an artificial neural network (ANN) model to identify the connection strengths that link the money market to the foreign exchange market. By extending the one-way arbitrage model to include currency futures rates, a dual interest parity relationship is derived which serves

as the basis for selecting the input parameters for the network architecture. Although the network's input parameters rest on fundamental relationships, the neural network model itself is free from any functional specifications. The ANN model is used to forecast the 90-day future spot rate for three currencies: the British pound, the German mark, and the Swiss franc. To evaluate the network's forecasting performance, an accuracy test and a correctness test were conducted. The results for both tests showed the network's predictions outperformed those made by both the forward and futures rates for the time horizon specified by the model. (JEL F30)