Abstract
Recent work on inference in regressions with heteroscedastic disturbances leads to the possibility of more efficient estimators. The conditions needed to exploi...