A0421
Title: Real time monitoring of a change in the persistence of stochastic volatility
Authors: Emily Whitehouse - University of Sheffield (United Kingdom) [presenting]
Abstract: Stochastic volatility models are commonly used to describe the time-varying nature of volatility in asset returns. Much financial and econometric literature has assumed stochastic volatility to be highly persistent. Still, recent research suggests that structural breaks are common in both the level and persistence of stochastic volatility and that a failure to account for these structural breaks can cause over-estimation of the true persistence in many series. We propose a real-time monitoring test procedure for structural breaks in the persistence of stochastic volatility. We exploit the autocorrelation structure of the log-squared price return series to propose several simple test statistics based on the autocorrelations of this series. A two-tailed version of a recently proposed real-time monitoring algorithm is considered to allow the detection of both increases and decreases in persistence. Monte Carlo simulations show that our test procedure has promising levels of power to detect structural breaks of this nature.
