Volatility Forecast with Long Memory: Evidence from Jordan Stock Market

Various volatility estimators and models have been proposed in the literature to measure volatility of asset returns. The particular emphasis of this paper is on assessing empirical performance of various long memory models (ARFIMA, FIGARCH models, and MF multi-fractal model which has recently been introduced as a new model)in comparison to short memory such as GARCH model, using time-series data from 1987-2004 of 90 stocks traded on the Amman Stock Exchange (ASE). Since long memory models should have a particular advantage over long forecasting horizons, we consider predictions of volatility models by one-day, five-day, ten-day, one-moth, two-moth, and three-month ahead. Two different measures are used to evaluate the forecast accuracy, RMSE and RMAE. Our results indicate that conditional volatility (ARFIMA ,FIGARCH and MF models) dominate over GARCH model. However, while FIGARCH and ARFIMA also have a number of cases with dramatic failure of their forecast, the MF model does not suffer from this shortcoming and its performance practically always improves upon the naïve forecast provided by historical volatility.