ABSTRACT
Universal volatility phenomenon exists in economic and financial time series,
while fluctuation is a core problem in financial market research and stochastic volatility
model is an important method for financial market research. In recent years, there are
great progresses in stochastic volatility model. For instance, a number of expansion
model were raised by researchers as well as many parameter estimation methods were
proposed as its parameter estimation specificity. This thesis were main to simulate and
emulate the different kinds of SV expansion models on the basis of Gibbs sampling
Monte Carlo(MC) methods and contrast these models with each other.
(1) The thesis summarized the various types of stochastic volatility models’
parameter estimation methods, with an emphasis on the MCMC method.
(2) This thesis gave a comparative analysis of Shanghai Composite index
simulation result of five types SV model by using WinBUGS software, and evaluated
the merits and demerits of these fitting model. According to the contrast, for simulating
Shanghai Composite index simulation, SV-MT model is the most representative model
in describing Shanghai stock market volatility level, while Leverage SV model is the
best in simulating the data of Shanghai stock market, which mean that Shanghai stock
market exists low leverage effect.
(3) The definition、relationship and difference of unusual fluctuation points and
variable structure points were introduced. And we used related methods to diagnose the
unusual points in volatility process with the sample of Shanghai stock market index and
analysis the cause of the appearance of these unusual points. Accordingly, we came to
the conclusion that Shanghai stock market is a typical policy-oriented market. And then,
we built the models aim at the above results by using the first-order expansion of SV
model and variable intercept SV model.Universal fluctuation phenomenon exists in
economic and financial time series, while fluctuation is a core problem in financial
market research and stochastic fluctuation model is an important method for financial
market research.
Key Words: stochastic volatility, Monte Carlo, leverage effect, unusual fluctuation
points, variable structure points, policy-oriented market