The Importance of Regression Equations Specification in Measuring Uncertainty of Macroeconomic Variables

In this study measuring of uncertainty related to macroeconomic variables has been considered. Given that uncertainty is not directly observable and measurable, researchers suggest different proxies for its measuring. One of the popular approaches in the related literature is based on time series models. In this approach, the proper measurement of uncertainty requires correct specification of regression equations. Accordingly, we estimate the uncertainty estimates of the baseline model for several key series in Iran’s macro dataset and compare it to the corresponding estimates of the alternative models. Our results show that uncertainty estimates of macro variables are affected by the specification of the forecasting regression equations. The difference over time between the estimates for these variables is quite pronounced in some periods, suggesting that much of the variation in these series is predictable and should not be attributed to uncertainty. Finally, evaluation of the uncertainty forecasting accuracy of the models shows that the SV and asymmetric GARCH models have better performance for the in-sample and out-of-sample periods, respectively.