Forecasting Inflation based on Stochastic Differential Equations and Alternative Models (A Comparative Study)

In this paper، it is tried to propose a robust model for predicting inflation in Iran among alternative models. For doing this، monthly data from April 1990 to the end of September 2009 is used. Firstly، it is tried to determine whether the CPI data is chaotic or stochastic. It is shown that it is chaotic rather than stochastic. Therefore، it is predictable. Then، a stochastic differential equation model is estimated (specifically a geometric Brownian motion) for CPI in Iran. In order to compare the prediction power of the model other alternative models of prediction like ARMA، non-linear GARCH، EGARCH، TGARCH are also used to extrapolate inflation during a six month prediction period. Based on RMSE، MAE، U-Tail، it is revealed that stochastic differential equation model is much more robust than the alternative models mentioned above.