A Variance Bounds Test of the Linear Quadratic Inventory Model

This paper develops and applies a novel test of the Holt et al. linear quadratic inventory model. It is shown that a central property of the model is that a certain weighted sum of variances and covariances of production, sales, and inventories must be nonnegative. The weights are the basic structural parametersof the model. The model may be tested by seeing whether this sum is in fact nonnegative. When the test is applied to some nondurables data aggregated to the two-digit SIC code level, it almost always rejects the model, even though the model does well by traditional criteria.