Asymmetric lead-lag relation, nonsynchronous trading, time-varying risk premium, and cointegration

This thesis consists of two essays.;In the first essay, using a set of bivariate GJR-GARCH forecasting models, we provide empirical evidence that in addition to the lead-lag relation documented by Lo and MacKinlay (1990b) and the asymmetric directional phenomenon documented by McQueen, Pinegar and Thorley (1996), a non-synchronous trading variable is significantly positively correlated with the returns from a portfolio of small firms, and it is negatively correlated with the returns from a portfolio of large firms. Since portfolio returns are not perfectly measured when trading is non-synchronous, a partial explanation of the predictability of small firm returns from lagged large firm returns is mismeasurement of return.;The second essay of this thesis employs Johansen's (1988) vector cointegration procedure and the Gonzalo and Granger (1995) decomposition to investigate the long-run behavior of values of five size based portfolios in the NYSE and the AMEX. We present evidence that there is one common cointegration relation among the price-weighted values of the five size based and portfolios. Estimation and tests of the cointegration suggest that the three largest firm portfolios (portfolio 3, 4, and 5) play the most important role in the stationary cointegration relations. Although all portfolios, except portfolio 4, are active in determining the permanent components of the Gonzalo-Granger decomposition, small firm portfolios generally have larger and much more volatile transitory components.