Modeling interest rate volatility: A Realized GARCH approach

We propose using a Realized GARCH (RGARCH) model to estimate the daily volatility of the short-term interest rate in the euro-yen market. The model better fits the data and provides more accurate volatility forecasts by extracting additional information from realized measures. In addition, we propose using the ARMA-Realized GARCH (ARMA-RGARCH) model to capture the volatility clustering and the mean reversion effects of interest rate behavior. We find the ARMA-RGARCH model fits the data better than the simple RGARCH model does, but it does not provide superior volatility forecasts.