《《4 An Almost Ideal Demand System1980 70.3.312-326》.pdf

An Almost Ideal Demand System By ANGUS DEATON AND JOHN MUELLBAUER* Ever since Richard Stone (1954) first In Section II, the model is estimated on estimated a system of demand equations postwar British data and we use our results derived explicitly from consumer theory, to test the homogeneity and symmetry re- there has been a continuing search for strictions. Our results are consistent with alternative specifications and functional earlier findings in that both sets of restric- forms. Many models have been proposed, tions are decisively rejected. We also find but perhaps the most important in current that imposition of homogeneity generates use, apart from the original linear expendi- positive serial correlation in the errors of ture system, are the Rotterdam model (see those equations which reject the restrictions Henri Theil, 1965, 1976; Anton Barten) and most strongly; this suggests that the now the translog model (see Laurits Christensen, standard rejection of homogeneity in de- Dale Jorgenson, and Lawrence Lau; Jorgen- mand analysis may be due to insufficient son and Lau). Both of these models have attention to the dynamic aspects of con- been extensively estimated and have, in sumer behavior. Finally, in Section III, we addition, been used to test the homogeneity offer a summary and conclusions. We be- and symmetry restrictions of demand the- lieve that the results of this paper suggest ory. In this paper, we propose and estimate that the AIDS is to be recommended as a a new model which is