SWJ) report the results of a study on import demand for malt using the linear approximation of almost ideal demand system (LAIDS). Several studies on consumer demand have showed that the LAIDS could produce extremely biased estimates of demand elasticities using consumers' consumption and price data (Pashardes 1993;Buse 1994;Hahn 1994;Moschini 1995;Chen 1998). Whenever possible, demand analysts are advised to estimate the nonlinear version of the almost ideal demand system (NAIDS) instead of its linear approximation.In this comment, we use the same data set from SWJ to assess whether a similar bias exist in the import demand estimates using LAIDS and, if it exists, to what extent? As estimates of price and expenditure elasticities are often used to provide guidance for marketing strategies and policy making, answers to the issue of bias is important. We reestimate SWJ import demand for malt using NAIDS and compare elasticity estimates from SWJ's LAIDS and NAIDS.
The NAIDS SpecificationAn n-goods NAIDS in share form is given as (Deaton and Muellbauer 1980):(1) w i = p i q i /M = the budget share of good i, p i = the price of good i, q i = the quantity consumed of good i, M = the total expenditures on n-goods, α i , γ ij , β i = parameters, ω i = the error term, P = the translog price index defined by:(2)To satisfy adding up, homogeneity and Slutsky symmetry of a standard consumer utility maximization problem, the following restrictions on the parameters are imposed: