| Summary: | 碩士 === 銘傳大學 === 財務金融學系碩士班 === 97 === Financial information accumulates globally around the clock. However, the daytime trading period of a stock market is typically half the length of the overnight non-trading period. Inevitably, not all price sensitive financial information becomes available during trading hours. Recent evidence indicates that there is considerable public information accumulating overnight. In addition, the information generated from countries located in different time zones would affect the next opening price, and result in significantly difference between overnight return and trading period return. We introduce a stochastic volatility model, which conditions on lagged overnight information, distinguishes between the non-trading periods of weeknights, weekends, holidays and long weekends, and allows for an asymmetric leverage effect on the impact of overnight news. This paper designed six different stochastic volatility models. Furthermore, we implement MCMC method and Gibbs sampler to estimation the parameter of the SV model. The empirical result indicates that there is substantial predictive ability in financial information accumulated during non-trading hours for a set of Taiwan Stock Exchange Index, Nikkei 225 Index and SH & SZ 300 Index, and also shows that it is necessary to divide the non-trading period into four areas, including weeknights, weekends, holidays and long weekends. Finally, the research found that asymmetric SV model has better interpretation for sample information, and then used this general asymmetric SV model to investigate whether the asymmetric leverage effect exist in stock index.
|
|---|
