<![CDATA[Inflation is an important indicator that can provide information on the development of prices of goods and services consumed by public. Forecasting inflation is important in order to assist the government in taking monetary policy to maintain economic stability in the future. In general, forecasting inflation can be done with time series approach, causal approach, and a combination of time series and causal approaches. Models with a combined approach that is widely used for forecasting inflation is ARIMAX model that includes Transfer Function and Intervention Model or also known as dynamic regression models. In addition, Generalized Autoregresive Conditional Heteroscedasticity (GARCH) for variance has also been applied to models in forecasting inflation. This study explains the procedure of building the ARIMAX models and GARCH detection using a case study of inflation data in denpasar city. Predictor variables consist of metric data variable (ie number of foreign tourists) and non-metric data variables (the increase of fuel oil (BBM) ), basic electricity tariff (TDL) and Bali bombings). The best model for in-sample data is intervention model with the smallest value of AIC and SBC, whereas the best model for data out-sample is transfer function model with the smallest RMSE value. GARCH detection results with Langrange Multiplier test shows no evidence of heteroscedasticity in ARIMAX model.]]>
Copyrights © 2014
