The paper proposed an improved multivariate time series model for tax revenue using vector autoregression (VAR) approach. The advantage of this study is to verify the error variance decomposition on both the original and transformed data. To check the efficiency of this approach the four stages of VAR were observed with R package to examine the mean absolute error MAE, mean absolute percentage error MAPE, mean error ME, and root mean square error RMSE. The result shows that, the error variance decomposition in transformed data is less than the error variance decomposition in Original data, which means there‟s significant difference in the error variance decomposition The models suggested by information criterion procedure are different because VAR (1) model is selected for transformed data for all the criterions while by information criterion procedures for the original data are VAR (5) for AIC, BIC, FPE, and VAR (2) for SIC model are the most suitable model for the data sets based on the model adequacy checking and accuracy testing. The result of forecast error reveals that the outsample data yield fewer error values than the in-sample data for both the original and transformed data. The study recommended that data should always be treated before using it for prediction.
Volume 12 | 07-Special Issue
Pages: 294-303
DOI: 10.5373/JARDCS/V12SP7/20202110