The Almost Ideal Demand System (AIDS), become the most popular model in applied demand analysis. The popularity of the AIDS model can be explained by two main reasons. First, it satisfies almost all desirable properties. That is, it gives an arbitrarily approximate to any demand system; it satisfies the axioms of choices exactly; it aggregates perfectly over consumers without invoking parallel linear Engel curves; its functional form is consistent with known household-budget data; and it easily allows testing of the standard restrictions of classical demand theory. Second, the AIDS is apparently easy to estimate because it is, in general, approximated to a linear system (LAIDS). The reason for this linearization, which typically uses the Stone price index, is to avoid difficulties involved in the estimation process of the AIDS model, which is a nonlinear system.;The true link between the translog price index and the Stone price index was established. This allows us to analytically evaluate the conditions under which it is adequate to approximate the AIDS model with the LAIDS model. We found that conditions that allow the linearization of the AIDS model impose serious restrictions on the form of the original model specification. Thus, it is not appropriate to use the Stone price index to approximate the AIDS model with the LAIDS model.;For an alternative to the commonly used estimation method for the AIDS, this thesis had proposed a Bayesian approach to directly evaluate the AIDS model without compromising model specification. Our results suggest that it is more advantageous to use the Bayesian method than the classical iterative methods. Using meat consumption data, two cases of AIDS model specification were studied: first the AIDS model was evaluated when preferences are assumed constant over time, and second, we analyze the AIDS model when possibility for structural changes in preferences is assumed. Our results suggest that it may be inappropriate to use aggregated meat data for demand analysis.
展开▼
