The fundamental idea behind this study was to forecast a number of variables all at once, taking advantage of relationships between them. The goal of this study is to compare the forecasting ability of several VAR and ARIMA models using Bangladeshi economic indicators. Secondary sources in Bangladesh included the Bangladesh Bank Bulletin, Bangladesh Economic Review, Monthly Economic Trends of Bangladesh Bank, and Bangladesh Statistical Yearbook. The accuracy of the stationary VAR and ARIMA models for forecasting these financial variables was verified by comparing ME, RMSE, MAE, MPE, MAPE, and MASE of the variables. For strongly correlated variables such as GDP vs. GNP, Export vs. Import, and so on, the VAR model performed better than ARIMA models in this study. For fairly low correlated variables, however, ARIMA and VAR models performed almost identically. That is, because the variables were comparably weakly correlated, the multivariate time series model could not provide a better forecast than the univariate time series model. Finally, the researchers stated that before predicting, the authority should verify correlations among the variables, and that for highly correlated variables, the VAR model should be used for forecasting, while for uncorrelated variables, any model can be employed.
Please click here : https://journalajpas.com/index.php/AJPAS/article/view/30243
No comments:
Post a Comment