摘要：In forecasting financial time series such as inflation, there is a reason to believe that variance of error term is volatile. Reseachers are likely to find periode of high volatility with large errors, then it is followed by periode of low volatilty with smaller errors. The variability could occur because the financial market is very sensitive to changes in government monetary and fiscal policies, even non economic factor such as political upheavals, rumors etc. The variance of errors is not constant but varies from one period to another period. It contains some kind of outocorrelation in the variance of errors. This model is so-called autoregressive conditional heterscedasticity (ARCH).The goal of this study is to apply ARCH model in estimating financial time series in Indonesia with montly data of inflation during 1994.1-2002.4 period and to compare it with OLS model. The inflation data exhibit volatility, suggesting that variance of inflation varies over time. By using the ARCH model, the results prove that ARCH-M model with maximum likelihood estimation gives better results than the OLS one.
