A Comparison of Univariate and Multivariate Time Series Approaches to Modeling Currency Exchange Rate

Main Article Content

Ijomah Maxwell Azubuike
Opabisi Adeyinka Kosemoni

Abstract

This paper describes a study using Average Monthly Exchange Rates (AMER) of Naira (Nigerian currency) to six other currencies of the World to evaluate and compare the performance of univariate and multivariate based time series models. The data from 2002 -2014 was used for modeling and forecasting the actual values of the AMER for 2014 of the six currencies. The Mean Absolute Percentage Error (MAPE) forecast accuracy measure was also used in determining if Univariate Times Series Model or Multivariate Time Series Models is best for forecasting the future AMER value of a given currency. The result of data showed that the Univariate time series fits better for Dollar, Pounds Sterling, Yen, WAUA and CFA, while only Euro fits well for the Multivariate time series.

Keywords:
Autoregressive integrated moving average, vector autoregressive and mean absolute percentage error.

Article Details

How to Cite
Azubuike, I., & Kosemoni, O. (2017). A Comparison of Univariate and Multivariate Time Series Approaches to Modeling Currency Exchange Rate. Journal of Advances in Mathematics and Computer Science, 21(4), 1-17. https://doi.org/10.9734/BJMCS/2017/30733
Section
Original Research Article