Semiparametric GARCH Model Compared With GARCH, Nonparametric GARCH Model

In recent years, China's securities are in an opportunity and unexpected risks era, investment and financing environments are becoming more and more complex and changeable。In this case, for all financial investment institutions concerned, The key to survival is whether they can control and manage their stock market investment risk or not. The fluctuating of the prices has been the main feature of securities market since it produced. For securities market each stakeholder, how to accurately describe the securities market price and identify market future yield are concerned about. Therefore, The study of volatility are particularly important.A lot of empirical researches have shown that the financial data exist some phenomenon, such like pointed peak thick tail, a small but lasting memories, volatility agglomeration etc. So it is a bit inappropriate if we continue to use the traditional time series of the regression model to fit the volatility of the financial data. GARCH model can well fit fluctuation agglomeration and pointed peak thick tail which exist in the data. So since it was put forward, GARCH models have been influenced by the scholars of favor and have become a powerful tool of measuring the volatility of financial market data. Currently the method used to estimate the GARCH model is mainly three kinds:parameter GARCH model, nonparametric GARCH model and semiparametric GARCH model. Due to simple in form and easily to estimates, parameter GARCH model is most widely used. But it may exist the risk that the model established by mistake. In view of this situation, scholars researched and proposed the nonparametric GARCH model. The nonparametric GARCH model is able to overcome the problem of the parametric GARCH model, but there are more difficult to explain model result and "dimension disaster" problem. Therefore, subsequently, we introduced the semiparametric GARCH model, it combined by parametric GARCH model and nonparametric GARCH model, so it can not only overcome parameter model's problem that established by mistake, but also can set the risk of solving the nonparametric model's problem that inadequately explained and dimension disasters problems, therefore semiparametric GARCH model are great practical value for scholars and investorsThe innovation of thesis mainly embodies in the research method and empirical analysis. It state the models of GARCH, semiparametric and nonparametric GARCH, and at the same time, it select the 180 index and shenzhen component index as the research object. carring out quantitative modeling for volatility of its index returns, and the three models made comparison analysis, the results of the study indicate:comparing with parametric GARCH model, the fitting of volatility are more suitable for market and closer to a true value using nonparametric GARCH model and semiparametric GARCH model; And semiparametric GARCH model can not only overcome parameters of nonparametric GARCH existing "dimension disaster", but also to explan the results of the estimating.