The Potential for High-Value Agricultural Products Under the North American Free Trade Agreement: The Case of Beef in Mexico and Canada: Comment

In his recent article, Onianwa used the absolute price version of the Rotterdam demand model to estimate Mexican and Canadian import demands for U.S. beef products. The specification of the Rotterdam model allows straightforward testing of the basic theoretical properties of demand. Based on theory, a system of demand equations should obey adding up, homogeneity, symmetry, and negativity (Theil 1971, 1975; Deaton and Muellbauer). Adding up is guaranteed or automatically satisfied in the Rotterdam model and other similar demand models like the almost ideal demand system (AIDS). Hence, adding up cannot be tested. Negativity can be checked by calculating the eigenvalues of the Rotterdam Slutsky matrix (all eigenvalues should be nonpositive). The remaining two properties, homogeneity and symmetry, can be straightforwardly tested by using the likelihood ratio test, as in the present paper, or the Wald or Lagrange multiplier tests; homogeneity can also be tested separately for each demand equation using the F-test.