In this article, tests of the implications of consumption-based asset-
pricing models are developed. Four of these tests are based on varianc
e bounds for intertemporal marginal rates of substitution introduced b
y Hansen and Jagannathan. The tests provide one means of quantifying t
he effects of sampling error when the bounds are used as a diagnostic
device. The tests are used to construct confidence regions for the par
ameters of an asset-pricing model using U.S. data. Monte Carlo simulat
ion is used to determine the small-sample properties of the tests.