ABSTRACT
This chapter discusses autocorrelated disturbances in dynamic models and time-varying variances of disturbances. The autocorrelation itself has nothing to do with the unbiasedness. The unbiasedness of the ordinary least squares estimator is subject only to the exogeneity of explanatory variables. In many applications, the autocorrelation disappears if lagged dependent and independent variables are included as regressors. Furthermore, dynamic models provide more useful information than static models because they can describe dynamic effects of explanatory variables on the dependent variable. Estimated regression functions can be used to predict future values of the dependent variable. The GARCH model assumes that the conditional variance depends on the magnitude but not the sign of it. This assumption somewhat appears inconsistent with empirical evidence on the behavior of stock market prices. Food safety issues have come to center stage as contamination incidents worldwide during the 1990s attracted press attention and US policy makers instituted the Hazard Analysis Critical Control Point regulatory process in 1996.
