How Persistent is Stock Return Volatility? An Answer with Markov Regime Switching Stochastic Volatility Models
提出带马尔可夫区制转换的随机波动模型，发现股票收益波动率的持续性和平滑性远低于传统GARCH或随机波动模型的估计，且低波动时期更易出现持续短区制，高波动时期持续性更弱，表明波动率估计和预测比通常认为的更难。
Abstract: We propose generalised stochastic volatility models with Markov regime changing state equations (SVMRS) to investigate the important properties of volatility in stock returns, specifically high persistence and smoothness. The model suggests that volatility is far less persistent and smooth than the conventional GARCH or stochastic volatility. Persistent short regimes are more likely to occur when volatility is low, while far less persistence is likely to be observed in high volatility regimes. Comparison with different classes of volatility supports the SVMRS as an appropriate proxy volatility measure. Our results indicate that volatility could be far more difficult to estimate and forecast than is generally believed.