The Efficacy of SARIMA Models for Forecasting Inflation Rates in Developing Countries : The Case for Turkey

This paper analyzes the efficacy of SARIMA models in view of forecasting the inflation rates in the Turkish economy. We perform rigorous tests on the stationarity and show that seasonality in the Turkish inflation rate is both deterministic and stochastic in nature, with the latter form dominating the inflation process. Further, we provide the first study that tests for fractional integration in a Turkish inflation series from 2003 to 2009. The proposed SARIMA model is derived by a systematic modeling strategy with the step- wise selection procedure of the novel Hyndman-Khandakar (HK) algorithm. Our results suggest a single best SARIMA model that provides a parsimonious and accurate representation of the Turkish inflation process from 2003 to 2009