View article

Specification error resulting from the choice of an incorrect functional form can lead to biased estimates. A new, more flexible and more general demand system is derived in this paper, and applied to evaluate the effects of functional form choices on the estimated effects of beef and pork advertising on the demand for meat in the United States. The new demand system, the Nested PIGLOG (NEP) model, combines three types of generalizations used previously in the literature:(1) the inclusion of pre-committed quantities kMM to the Linear Expenditure System (LES) generalization of the Cobb-Douglas model),(2) nesting the Almost Ideal (AI) and Translog (TL) models together (Lewbel 1989), and (3) using the Fourier flexible functional form to augment other functional forms (Chalfant 1987). In particular, the new demand system augments the expenditure function of the Generalized Almost Ideal Translog (GAITL) model (Bollino and Violi 1990) with the expenditure function for the Fourier flexible functional form (Gallant 1981, 1982). M

The NEP model nests all of the previously existing demand systems that are consistent with PIGLOG preferences, and five new models. In this paper, we estimate the NEP demand system including advertising variables, using US meat consumption data, and test the appropriateness of the sets of parametric restrictions in the NEP model that define its nested special cases. We also test the statistical (and economic) significance of advertising effects, given each of the functional form restrictions.