ANOTHER HETEROSKEDASTICITY-CONSISTENT AND AUTOCORRELATION-CONSISTENT COVARIANCE-MATRIX ESTIMATOR

A root T-consistent estimator of a heteroskedasticity and autocorrelat ion consistent covariance matrix estimator is proposed and evaluated. The relevant applications are ones in which the regression disturbance follows a moving average process of known order. In a system of I equ ations, this 'MA-l' estimator entails estimation of the moving average coefficients of an I-dimensional vector. Simulations indicate that th e MA-I estimator's finite sample performance is better than that of th e estimators of Andrews and Monahan (1992) and Newey and West (1994) w hen cross-products of instruments and disturbances are sharply negativ ely autocorrelated, comparable or slightly worse otherwise.