This paper develops a parametric family of models of generalized autor
egressive heteroskedasticity (GARCH). The family nests the most popula
r symmetric and asymmetric GARCH models, thereby highlighting the rela
tion between the models and their treatment of asymmetry. Furthermore,
the structure permits nested tests of different types of asymmetry an
d functional forms. Daily U.S. stock return data reject all standard G
ARCH models in favor of a model in which, roughly speaking, the condit
ional standard deviation depends on the shifted absolute value of the
shocks raised to the power three halves and past standard deviations.