ScholarWorks@경상국립대학교

초록

This study proposed a four-regime stochastic volatility model (FRSV) to better capture the time-varying properties of asymmetric volatilities and investigated the properties of asymmetric volatilities in Korean stock markets using KOSPI and KOSDAQ daily returns from January 2003 to March 2020. This process involved applying and estimating various asymmetric stochastic volatility models including FRSV model using the Bayesian Markov chain Monte Carlo method. I also compared models through a deviance information criterion (DIC) and value at risk (VaR) study. According to the results of the DIC and VaR estimates, FRSV model with the threshold and leverage effect, which is referred to as the negative correlation between returns and volatilities, revealed a good fit of the model for KOSPI returns. For KOSDAQ returns, FRSV model and the triple regime stochastic volatility model fit the data best. The estimation results show that the relationship between returns and volatilities depends on the magnitude of returns as well as the sign.

키워드