A GENERALIZED VARIANCE BOUNDS TEST WITH AN APPLICATION TO THE HOLT ET-AL INVENTORY MODEL

This paper derives a variance bounds test for a broad class of linear rational expectations equilibrium models. According to this test, if o bserved data accord with the model, then a weighted sum of autocovaria nces of the covariance-stationary components of the endogenous state v ariables should be nonnegative. The new reinterprets West's (1986) var iance bounds test and extends its applicability by not excluding unobs ervable exogenous state variables, nonstationary exogenous and endogen ous state variables, and nonzero initial values for the endogenous sta te variables. The new test is computed together with West's test for t he Holt et al. inventory model using nondurable industry data. While W est's test almost always rejects the model, the latter almost always p asses the new test. This and other evidence provide strong support for the importance of supply shocks in explaining aggregate inventory beh avior via the Holt et al. model.