The Empirical Research on Volatility Measurement Model Based Multiplicative Error Model

Abstract

Volatility is a very important factor of measuring financial risk. This paper introduces the volatility measurement method of high frequency financial time series involving the nonnegative-Multiplicative Error Model. This paper takes the high frequency data of HS300 index of Chinese stock market as the research object, building the TARCH model according to leverage, and uses the "realized volatility" to build ARFIMA model, multiplicative error model respectively, then carries on the comparative analysis on accuracy after using the three models to predict with the mean square error method. The analysis results show that the multiplicative error model gives the best prediction effects, and ARFIMA model is the second.