On asymmetric volatility in equity markets

This dissertation presents empirical and theoretical analyses of the widely documented phenomenon in equity markets: asymmetric volatility, i.e., returns and conditional volatility are negatively correlated.;We first provide a unified empirical framework to simultaneously investigate asymmetric volatility at the firm and the market level and to examine two potential explanations of the asymmetry: leverage effects and time-varying risk premiums. The empirical application uses the market portfolio and portfolios with different leverage constructed from Nikkei 225 stocks. Although volatility asymmetry is present and significant at the market and the portfolio levels, its source differs across portfolios. We find that it is important to include leverage ratios in the volatility dynamics but their economic effects are mostly dwarfed by the volatility feedback mechanism. Volatility feedback is enhanced by a phenomenon that we term covariance asymmetry: conditional covariances with the market increase only significantly following negative market news. We do not find significant asymmetries in conditional betas.;Although there is much research on its modeling using Asymmetric ARCH models and stochastic volatility diffusion models, the determinants of asymmetric volatility are not clearly identified. In this dissertation we also develop a volatility feedback model where dividend growth and dividend volatility are the two state variables of the economy. We identify dividend volatility as the main determinant of asymmetric volatility. Our explanation is that the changing uncertainty regarding dividends drives return and return volatility in the opposite direction, generating asymmetric volatility. Since we specify a stochastic volatility dividend process our model extends significantly the Campbell and Hentschel (1992) volatility feedback framework. The model possesses several advantages since it distinguishes between "news about dividends" and "news about dividend volatility". The model is also able to provide interesting interpretation to some of the empirical characteristics they documented, including the clustering of volatility feedback during volatile periods of the market. The empirical exercises use both the returns data as well as conditional volatility proxied by short-dated, at-the-money options implied volatility. The model is estimated by simulated method of moments. We find that volatility feedback is significant both statistically and economically.