Multivariate Time Series Modelling of Nigerian Gross Domestic Product (GDP) and Some Macroeconomic Variables

Abstract

This paper focused on modelling Nigeria’s Gross Domestic Product and some macroeconomic variables, which include, Agriculture, Crude Oil/Mineral Gas and Telecommunication using different classes of multivariate time series models. Multi-Dependent Linear Regression Model (MLRM), Vector Autoregressive Model (VARM) and Multivariate Autoregressive Distributed Lag Models (MARDLM) have been fitted to the multivariate time series. The basic statistics of the estimates and errors reveal the competitiveness of VARM and MARDLM. This was also evidently using the model selection criteria. But the mean square error of forecast places VARM on a higher comparative advantage than MARDLM. The results of the Granger causality tests showed that Crude Oil/Mineral Gas granger causes Gross Domestic Product and also granger causes Agriculture, but not vice versa in each case. This paper establishes the fact that Crude Oil/Mineral Gas is a good predictor of Gross Domestic Product and Agriculture as a major contributor to the nation’s economic development. The need to consistently juxtapose causal relationships between major economic sectors and Gross Domestic Product is vehemently advocated for proper evaluation of sectorial contributions and formulation of economic driven policy in the country.