Font Size: a A A

Time-varying Volatility Persistence And Its Intrinsic Mechanism

Posted on:2020-11-19Degree:DoctorType:Dissertation
Country:ChinaCandidate:S N ChenFull Text:PDF
GTID:1489306518457254Subject:Financial engineering
Abstract/Summary:PDF Full Text Request
Volatility persistence refers to autocorrelation of volatility.GARCH and HAR models describe this phenomenon with linear autocorrelation model.The latest research finds that it is more accurate for volatility forecasting after calibrating the autocorrelation coefficient of HAR model with time-varying error terms.But there are few economic explanations for the time-varying error term.The most classical interpretation of the volatility persistence is that sequential arrival of exogenous information arrivals related to fundamentals cause volatility persistence.However,this interpretation does not explain the intrinsic mechanism of time-varying volatility persistence.Based on theoretical study of volatility and state-dependent investor attention,this paper conjectures that the characteristics of time-varying volatility persistence are related to investor attention.When the exogenous information shock arrives,market participants trade based on their own information,which induces price formation.Price as the endogenous information can attract investors' attention.Since endogenous information does not affect the fundamental value,demands for exogenous information increase.In that exogenous information and its accompanying uncertainty are incorporated into the market through trading,thus leading to time-varying volatility persistence.We investigate the above price feedback mechanism induced by investors' attention towards endogenous information as follows.First,the paper verifies that market endogenous information variables have significant impact on volatility persistence.In this paper,the linear combination of return and the realized volatility is used to construct the proxy variable of the market endogenous information.Calibrating HAR model with market endogenous information,we find that performance of the model has improved.Secondly,according to the calibrated model,the paper further explores the influence of endogenous information proxy on the volatility persistence.On the one hand,this paper examines that positive and negative endogenous information have asymmetric impact on time-varying volatility persistence.On the other hand,this paper examines the threshold effect of volatility persistence.Only when the endogenous information is greater than thresholds,will it affect the volatility persistence,and those thresholds change dynamically with the previous realized volatility.This conclusion is consistent with the attention allocation theory,that is,investors learn uncertainty from historical information,and only allocate limited attention to important information shocks,thus causing market reaction.Based on the above characteristics,it is speculated that the change of endogenous information impact is related to investor attention.Finally,this paper provides direct evidence for the relationship between investor attention and the above-mentioned time-varying characteristics and asymmetry of volatility persistence.On the one hand,this paper uses the amount of Guba postings and readings as the proxy variables of attention.It finds that the market endogenous information variables have a significant positive impact on investor attention,and attention has significantly positive impact on volatility persistence.On the other hand,using the positive and negative postings to construct asymmetric attention proxy,we find that the asymmetric attention has a significantly positive impact on the asymmetric volatility of the next trading day.In summary,the main contribution of this paper is to provide a new explanation for the intrinsic mechanism of time-varying volatility persistence from the perspective of micro-investor behaviour,and fill the gap in this research area.
Keywords/Search Tags:Time-varying volatility persistence, investor attention, asymmetric attention, Investor behaviour
PDF Full Text Request
Related items