Forecasting Exchange Rate: A Univariate Out-of-Sample Approach (Box-Jenkins Methodology)

In this paper, the author has tried to build a univariate model to forecast the exchange rate of the Indian rupee in terms of different currencies like SDR, USD, GBP, Euro and JPY. This paper uses the Box-Jenkins Methodology of building the ARIMA model. Sample data for the paper was taken from March 1992 to June 2004, out of which data till December 2002 were used to build the model, while the remaining data was used to conduct out-of-sample forecasting and check the forecasting ability of the model. The data was collected from the Indiastat database. The paper shows that the ARIMA models provide a better forecasting of exchange rates than the simple autoregressive models or moving average models. The author was able to build a model for all the currencies except USD, which shows the relative efficiency of the USD currency market.