Univariate Time Series Modelling of Cedi-Dollar Nominal Exchange Rate
John Tumaku *
Department of Mathematics and Statistics, Ho Polytechnic, P.O.Box HP 217, Ho, Ghana
Solomon Yemidi
Department of Multidisciplinary Studies, Ho Polytechnic, P.O.Box HP 217, Ho, Ghana
François Mahama
Department of Mathematics and Statistics, Ho Polytechnic, P.O.Box HP 217, Ho, Ghana
*Author to whom correspondence should be addressed.
Abstract
This paper seeks to develop a statistical model capable of forecasting the Ghanaian Cedi US Dollar nominal exchange rate. The study uses end-of-month interbank exchange rate between January, 1994 and May, 2016. We employed the Autoregressive Integrated Moving Average (ARIMA) scheme in analysing the data. The study reveals that the exchange rate has a unit root implying that shocks to the exchange rate have a long term permanent effect. We have established that the end-of-month Cedi-Dollar rate follows ARIMA (0,1,2) process. The 12-month ahead forecast indicates that the value of the Cedi to the Dollar would decline in line with its past trend. The forecasts from the model compared favourably with actual realizations of the exchange rate. It is recommended that businesses whose cost is in Cedi but their revenue in Dollar should use the lower limit of the forecasts for the purposes of planning their business activities and vice versa.
Keywords: Cedi, Dollar, nominal exchange rate, ARIMA process, unit root