Title: Conditional heteroskedasticity of return range processes
Authors: Yan Sun - Utah State University (United States) [presenting]
Guanghua Lian - University of South Australia (Austria)
Zudi Lu - University of Southampton (United Kingdom)
Jennifer Loveland - Utah State University (United States)
Isaac Blackhurst - Utah State University (United States)
Abstract: It is well known that price range contains important information about the asset volatility. To utilize this information towards volatility modeling, we propose to view the [low, high] daily price range as an interval-valued time series and develop an extended GARCH model, called Int-GARCH, that allows for interval-valued input, for the corresponding return range process. The aim of Int-GARCH model is to enrich the classical GARCH model that is solely based on closing price, with price fluctuation information in the surrounding time interval. Theoretical properties of the Int-GARCH model are developed under the framework of random sets, and a metric-based least squares method is presented for estimating model parameters. From the empirical analysis with stocks and indices data, the Int-GARCH model consistently outperforms GARCH for both in-sample estimation and out-of-sample prediction.