| This dissertation concerns several issues, all related in that they deal with estimation of demand functions. In the first essay, an approach to account for income distribution effects on market demand was developed. The proposed procedure contains two steps: estimation of the underlying income distribution and estimation of market demand. Given the rejection of Lewbel's concept of mean scaling for distribution of income in the United States, correcting for income aggregation bias is essential in demand analysis. The proposed approach addresses income distribution effects in a relatively simple way. In an empirical application to meat commodities, beef, pork, and poultry, the own-price elasticities were over-estimated and the income elasticities were under-estimated when income distribution effects were not properly taken into account.;In the second essay, a procedure that corrects for selectivity bias was proposed to estimate demand functions using cross-sectional data under the assumption that prices vary across households. This procedure, which extends the work by Cox and Wohlgenant, includes a two-equation system of expenditure and unit value functions. Using the 1987-88 Nationwide Food Consumption Survey, the model was applied to household expenditures for beef steaks and roasts. Both the price elasticity and the income elasticity were under-estimated if selectivity bias was not taken into account.;The remaining three essays relate to count data models. Poisson and negative binomial (NB) models have been extensively used in this context. However, these models impose arbitrary restrictions on the relation between the conditional mean and variance of the dependent variable, limiting their generality. In this study, tests for selection among the Poisson and NB models were first proposed by formally demonstrating that the log-likelihood function (LLF) of a general NB model parametrically nests the LLF of the Poisson, NBI and NBII as special cases. Subsequently, this general NB model procedure was applied to food away from home visit frequency analysis using National Panel Diary data. Finally, a count data model with endogeneity and missing regressor problems was derived for analyzing price effects on food away from home visits. |
