The Two Perils of Symmetry-Constrained Estimation of Demand Systems

Abstract

When the error covariance matrix Σ of a demand system is unknown and is estimated, (1) symmetry-constrained estimates based on such a Σ estimate may have large mean squared errors, and (2) the conventionally computed standard errors may seriously underestimate the variability of these estimates. Several procedures are compared; the conventional one (which uses a Σ estimate consisting of mean squares and products of LS residuals) performs worst.

This article first appeared in Economics Letters, 13 (1983), 105–111. Reprinted with the permission of Elsevier Science Publishers V.B. (North-Holland). Research supported in part by the McKethan-Matherly Eminent Scholar Chair, University of Florida.