Recipient Organization
UNIV OF MASSACHUSETTS
(N/A)
AMHERST,MA 01003
Performing Department
Resource Economics
Non Technical Summary
The degradation of health caused by a higher body weight has been at the core of policy debates and proposals aimed at achieving a less costly health care system. We propose to fill a gap in the policy debate and in the academic literature by carefully analyzing how different economic environment factors as well as consumers' preferences influence consumers' diet over time and across regions of the United States. In a first step, we propose to assemble a unique and rich brand-level database of consumer grocery purchases (quantities and prices), nutrition information and advertising intensities across the country and over time (a decade). Once assembled, we propose to apply an innovative econometric method that allows the recovery of parameters that describe consumers' preferences for products' nutrients, prices and advertising intensities. These parameters will be recovered at different periods of time and in different parts of the country. The main contribution of our econometric method rests on its ability to isolate the role of consumer preferences from all other elements that might influence dietary choices (product offerings, nutrient composition, prices, and advertising).Estimates from our method will then allow us to ask the counterfactual question of whether consumption would differ significantly if consumers in one region (or one time period) were exposed to firms' strategies (prices, product offerings, nutritional content, and advertising) observed in a different part of the country (or a in different year). The policy relevance of these econometric exercises is that they can be used to produce quantifiable predictions on the effectiveness of different scenarios for reducing unhealthful eating. For instance, we can evaluate effects of limiting (or banning) certain unhealthy ingredients (or advertising), restricting (or taxing) certain products or, alternatively, focusing on changing consumers' preferences towards certain types of consumption. By being able to separate the role of consumer preferences from all other factors that influence diet choices, we will be able to shed light on the highly-debated issue of whether policy formulations should focus on the consumer or the firm side.
Animal Health Component
(N/A)
Research Effort Categories
Basic
100%
Applied
(N/A)
Developmental
(N/A)
Goals / Objectives
Our overarching goal is to inform public health policy by uncovering which factors most determine diet in the United States today. We propose to disentangle the influence of supply-side factors, namely the choices that firms in the food market make regarding product offerings, pricing, and advertising, from the influence of demand-side factors, namely, consumer choices about what to buy and eat. This analysis leads naturally to policy conclusions about the most effective methods for improving nutrition and decreasing obesity in the United States.The funds provided by this grant will jump-start our efforts to a) to assemble a comprehensive database of food product offerings and consumer food choices that is both easily updateable with future data and readily useable for other studies; b) to use novel econometric methods to isolate consumer preferences from supply-side forces influencing dietary choices, and c) to generate findings of immediate policy impact that also serve as academic contributions to the health economics and industrial organization literatures.
Project Methods
As detailed in the project narrative portion of our grant proposal, we plan to collect, compile, and construct a database from three sources of data: scanner, advertising and nutrition. The database will include brand-level information on five elements: product offerings, market shares, prices, advertising expenditures, and nutritional content. The first three elements are obtained from scanner databases, while the other two (advertising and nutrition) come from sources already detailed. Nutrition data are available at the national level, whereas advertising, product offerings, market share and price data are available at the metropolitan area level.To measure the overall healthiness of products, we utilize two established health indices: the Nutrition Profiling Index and the Nutrient-Rich Foods Index. Our study of the indices will be confined to descriptive results with the goal of gauging patterns over time and across regions. To perform this analysis we will average the nutritional quality of the food market for each market-year combination in various ways. First, we form the simple average across all products available in a market. This allows us to judge variation in supply-side forces, such as differing access to healthy cereals, across different markets. Second, we will calculate the same average across products weighted by market share. This captures, in a non-casual way, demand-side variation, such as whether consumers in one market purchased healthier cereals than those in other markets.Our structural work is aimed at addressing two questions. In the first we propose to determine what portion of the variation in diet quality over time can be explained by supply-side verses demand-side factors. Importantly, we are further able to distinguish between various supply-side forces such as prices versus advertising. In a second exercise, we ask the same questions regarding variation across different geographical regions within the U.S. (e.g., Los Angeles versus Philadelphia).We adopt the model proposed by DGN to estimate consumer preferences in the food market. The advantages of this model are that it: a) allows us to account for household purchases across a range of food categories, b) allows utility to be a function of product characteristics, and c) allows purchases to be non-discrete within a category and across categories. We propose to extend this approach to include advertising in the above specified demand model. Specifically, we propose to incorporate advertising expenditures in the demand estimation as an additional product attribute.Similar to DGN, we propose to carry out the estimation of the proposed linear demand equation for nine product categories: fruits, vegetables, grains, dairy, meat, oils, sweeteners, drinks, and prepared foods. In their analysis, DGN were restricted to considering only carbohydrates, protein and fat as product attributes; we incorporate in addition sodium and cholesterol and, as discussed, the product's advertising. We propose to estimate consumer demand separately for different time periods and for different geographical regions. In another exercise, we consider estimating two sets of demand parameters for two geographical regions. While interesting in their own right--informing, for example, whether consumer preferences for fat differ across regions--these estimations are primarily useful to our counterfactual analyses, used to test potential policies. In a nutshell, all counterfactual exercises boil down to predicting consumption levels given a change in one of the market forces studied.Our counterfactual analysis allows us to independently consider at least four forces potentially influencing diet quality: nutritional profile of food products offered, prices of those offerings, advertising, and consumer preference for a particular nutrient. Of these, only the last is demand-side, while the others are supply-side forces in the food market.