Econometrica: May 1983, Volume 51, Issue 3

More Efficient Estimation in the Presence of Heteroscedasticity of Unknown Form<751:MEEITP>2.0.CO;2-W
p. 751-764

J. G. Cragg

Recent work on inference in regressions with heteroscedastic disturbances leads to the possibility of more efficient estimators. The conditions needed to exploit the possibility are discussed. A sampling experiment indicates that the gains can be achieved even in small samples.

Log In To View Full Content