Dynamism, variations in time, is the inherent characteristic of most economical phenomena. In econometrics, ignoring this dynamism may cause many problems due to over simplification of the problems. One such case occurs in application of static regression models to problems with dynamic nature. In so doing, one ignores the fact that the parameters of the model change in time, which may lead to misleading results.
In this paper, we intend to exhibit the importance of this negligence in the context of a practical problem, namely modelling the weekly average rate of conversion of us dollar into Iranian currency, Rial.
Thus, two approaches of static and dynamic modelling are compared with respect to their efficiency in tracking the path of the variations of the US dollar rate according to various criteria such as MAD and MSE of predictions. Using a dynamic time series model along with required interventions at outlier points
superiority of the Bayesian dynamic model is shown.
JEL Classification: C22, F31.
', './data/jte/coversheet/cover_en.jpg'))