Modeling Stock Volatility with Trading Information
This paper studies volatility in individual stocks of the Toronto Stock Exchange (TSE), using a recently developed nonlinear approach, a stochastic threshold model. Trading information is embedded into the determination process for volatility in the stochastic threshold model with a generalized conditional heteroskeasticitic variance (STGARCH). We use the number of price changes (quote changes) to approximate the trading information. This trading variable has significantly positive impact on stock volatility following a declining market and ambiguous impact on the stock volatility following a rising market; there is higher probability to fall into a highly volatile state after a declining market than after a rising market. The GARCH-type persistence in volatility is reduced significantly in our nonlinear model for individual stocks with high persistence. The STGARCH model also gives satisfactory fitting in terms of alternative model selection criteria.