Local Influence Analysis Of GARCH Model

In financial economics, we often relate to volatility, it is widely used in risk management, the derivatives pricing, asset pricing, portfolio and other. For volatility forecasting and research to measure this precisely today’s financial market has become a major task. However, we often encounter problems in the study when heteroskedasticity financial time series data, which often exhibit fat tail, volatility clustering characteristics. Therefore Heteroscedasticity emerged. GARCH model which is a typical model of Heteroscedasticity financial time series. GARCH model than the traditional model can more effectively explain the clustering and yield volatility thick tail phenomenon. Thus the GARCH model becoming an important tool in researching and forecasting volatility.In this article we will use Bayes approach to statistical inference, the main contents include:(1) Integral method using Byaes through the calculation of the posterior distribution of the unknown parameters to estimate the parameters of GARCH model.(2) Use Bayes method to diagnosis GARCH models. First, we did Bayes Data Delete impact analysis for GARCH model, and then we did Bayes local influence diagnostic for Bayes local influence joint small perturbation data distribution and the prior distribution of GARCH model.(3) Volatility clustering, fat tail, multi-peak, point mutations and abnormal findings by simulating GARCH model to better portray the financial time series data.