Modelling the asymmetry of stock market volatility

Recent studies suggest that a negative shock to stock prices will generate more volatility than a positive shock of equal magnitude. This paper uses daily data from the Hong Kong Stock Exchange to illustrate the nature of stock market volatility. Regression-based tests for integration in variance are applied, providing contrasting results to the usual test based on the Wald statistic. A partially non-parametric model of the relationship between news and volatility is estimated and used in conjunction with tests for the sensitivity to both the size and sign of a shock as a metric to judge various candidate characterizations of the underlying data generating process.