ARCH models often lie at the boundary of the parameter space under con
ditional homoskedasticity, which invalidates the usual chi(2) distribu
tion of LR and Wald tests. Although LM tests are not affected, the one
-sided nature of the alternative hypothesis should result in more powe
rful tests. We propose a simple one-sided version of the LM test, whic
h is closely related to the Kuhn-Tucker multiplier test. We also prese
nt critical values for LR, Wald and one-sided LM tests. The results of
a Monte Carlo comparison suggest that one-sided tests are indeed more
powerful than their two-sided counterparts. (C) 1998 Elsevier Science
S.A. All rights reserved.