Modeling of Internally Generated Revenue Using Autoregressive and Moving Average of a Time Series Models: A Case Study of Akwa Ibom State (Published)
Modelling of Internally Generated Revenue using error variances for model comparison was the main focused of this research. This procedure varies from the familiar information criteria used to compare alternative models. The autocorrelation and Partial autocorrelation function of the stationary series give basis for the choice of Autoregressive Integrated Moving Average, ARIMA (1 1 1), ARIMA (1 1 2) and ARIMA (2 1 1) for the revenue series. From the estimates, Akaike Information and Schwartz’s Information Criteria (AIC and SIC) suggested ARIMA (2 1 1), while the error variance suggested ARIMA (1 1 2) respectively as the best model. The advantage in the use of error variance for model comparison is that the variance measures are positive. (not less than zero). The positive and negative signs in the AIC and SIC values are sometimes confusing, since absolute values are not considered in the BIC, SIC and AIC. Hence, this research relies on error variance for the model selection, which reputes ARIMA (1,1,2) to be the best model for the Akwa Ibom State Internally Generated Revenue Series.
Keywords: Autocorrelation Function, Partial Autocorrelation Function, moving average error variance.