Modeling Inflation Volatility Using Arimax-Garch

Abstract

Forecasting inflation is necessary as a basis for making decisions and high quality good planning in economic development in Indonesia particularly for the government and businessmen. The forecasting generally uses time series data. However, there is a time series data which is difficult to obtain stationary, i.e., the variance on financial time series data such as the stock price index, interest rates, inflation, exchange rates, and etc. It is mainly caused by the inconsistency of variance (heteroscedasticity). This study developed Autoregressive Integrated Moving Average (ARIMA) model using exogenous factors, namely the price of oil and outlier detection to forecast inflation. Another modeling which is expected to solve the problem of heteroscedasticity is a Generalized Autoregressive Conditional Heteroscedasticity (GARCH) model. In this study, the asymmetric GARCH of Glosten Jagannathan Runkle-GARCH (GJR-GARCH) was carried out. This model could accommodate the volatility in the form of negative shocks that can leverage the effect. The data used in this study was the Inflation rate of Indonesia and world oil prices in January 1991 to December 2014 respectively. The results showed that ARIMAX-GJR GARCH is the best model to forecast national inflation volatility.