APPLICATION OF LINEAR STOCHATIC MODELLING FOR NIGERIAN MONTHLY CRUDE OIL EXPORTS
Chima Godknows Igiri & Ette Harrison Etuk
Department of Computer Science, Rivers State University, Port Harcourt
Department of Mathematics, Rivers State University, Port Harcourt
Email: chimaigiri@yahoocom
ABSTRACT:
This research is concerned with autoregressive integrated moving average (ARIMA) modelling of monthly crude oil exports of Nigeria. The realization covered is from January 2006 to November 2016. The time plot reveals an initial generally negative trend up to mid-2009, and then a positive trend up to mid-2010 and then a negative trend up to 2016. An examination of the data reveals a measure of twelve monthly seasonality. The Augmented Dickey Fuller (ADF) Test adjudges the series as stationary but its autocorrelation structure shows the involvement of a secular trend in the series. This means that the series is still non-stationary. A non-seasonal differencing makes the series stationary. The correlation structure of these first differences suggest an ARIMA(1,1,1) fit. A twelve monthly differencing of the original series was done also. The correlogram of these seasonal differences shows evidence of secular trend in them. A non-seasonal differencing rids the seasonal differences of the trend. The autocorrelation structure of the resultant series gives evidence of twelve-monthly seasonality and suggests a SARIMA(1,1,0)x(0,1,1)12 fit. However it is observed that on the basis of AIC, the ARIMA(1,1,1) fit is the more adequate model. Hence, forecasting or simulation of the export series may be done on its basis.
Keywords: Nigerian Crude Oil Exports, ARIMA Modelling, SARIMA modelling
