| Summary: | The survey of price trends duringthe period from 2005 to 2010 was aimed atestimating the best quality of ARIMA models usedto forecast consumer price index and/or inflation inBiH in 2011. A very process to create and selectARIMA models was based on the recommendedBox-Jenkins procedure involving data collection forthe selected time series, identification of itsstationary level, creation of model, diagnosticchecking, model stationary checking (simulation ofreal series) and finally inflation forecasting basedon the selected models. This is an iterative process,i.e. it is repeated over and over again to obtain aset of desired and efficient models used for inflationforecasting. Out of 64 tested models following theAkaike Info Criterion (AIC), Schwarz Criterion(SIC), Root Mean Squared Error (RMSE), MeanAbsolute Error (MAE) and Theil InequalityCoefficient (TIC) criteria, three ARIMA modelsidentified to fit best the forecasting procedure wereselected. Their efficiency and predictive power willbe tested during the course of 2011.
|
|---|