A Stochastic Volatility and Leverage: Application to a Panel of S&P Stocks

Abstract

We estimate stochastic volatility leverage models for a panel of stock returns for 24 S&P 500 firms from six industries. News are measured as differences between daily return and a monthly moving average of past returns. We estimate the models by maximum likelihood using an Efficient Importance Sampling method which produces numerically highly accurate estimates of the likelihood and related test-statistics. We find significant leverage effects for all 24 stocks. These effects are fairly consistent within each industry but there are significant differences across two groups of industries. Our models produce significant improvement in volatility predictability.