| There are many policy questions concerning how to encourage consumers to buy healthier food. Among these concerns are the effectiveness of subsidizing low-calorie, high-nutrition foods such as fruits and vegetables to improve public health. Many food policies and food assistance programs, such as the Supplemental Nutrition Assistance Program (SNAP), are targeted at low-income households where per capita fruit and vegetable consumption is the lowest. For policies and programs like these, quantitative information on demand for fruits and vegetables at the household level and for different segments of the population is required to inform public policy. In addition to providing information on demographic and other socio-economic variables, this dissertation focuses on quantifying consumer responses to prices and incomes in a complete demand system framework.;Chapter One estimates demand for three fruit and vegetable products (fresh fruits, fresh vegetables, and processed fruits and vegetables) by two income groups (high and low) as a weakly separable group using data from the Consumer Expenditure Survey (CEX) from 2002 to 2006. A censored demand system, using a correlated random effect specification, is applied and the model is estimated by quasi-Maximum Likelihood Estimation. Results show that the demand for fresh fruits, fresh vegetables, and processed fruits and vegetables (conditional on expenditure for all fruits and vegetables) for low-income households are more price elastic than those for high-income households. All three fruit and vegetable products are found to be gross complements and net substitutes. Season, region, and a number of demographic variables such as household heads' race and gender and household size and composition affect consumption of fruits and vegetables for both categories of households. In contrast, urban status, household heads' educational level, and age have an impact on high-income households only.;Chapter Two uses a different approach to reexamine conditional demand for fruits and vegetables using the 1996--2010 CEX data. Because the panel data used in the first chapter only have two time periods, this study overcomes the limitation by constructing a "pseudo-panel" from the mean of the grouped observations. The groups, which are selected according to the time and income variables, can be tracked over time, although individuals within the group cannot. This study extends the theoretical framework of single-equation pseudo demand to a demand system where cross-equation restrictions can be imposed. A common problem in the empirical work is that time-invariant group effects are neglected during the estimation, which leads to an inconsistent Ordinary Least Squares (OLS) estimator and an invalid hypothesis testing of the fixed effects estimator. However, this study accounts for the time-invariant group effects and proposes an efficient Generalized Method of Moments (GMM) estimator. Results are consistent with Chapter One except that processed fruits and vegetables and fresh vegetables are found to be net substitutes, but the corresponding estimated cross-price elasticities are insignificant.;Chapter Three investigates how a price subsidy affects demand for three fruit and vegetable products for two income groups of households. This study combines the results from Chapter Two and develops a two-stage budgeting approach to estimate a complete demand for fruits and vegetables using 1986--2010 quarterly CEX data. Results show that low-income households have larger total expenditure elasticities but smaller unconditional price elasticities than high-income households. Fruits and vegetables and all other goods are found to be net substitutes. Assuming that supplies for fruits and vegetables are perfectly elastic, a 10% price subsidy increases consumption of processed fruits and vegetables, fresh vegetables, and fresh fruits by 3.27% (10.68%), 3.29% (10.73%), and 3.50% (11.42%) for low-income (high-income) households and only causes a small change in consumption of all other goods. |
