Recipient Organization
PENNSYLVANIA STATE UNIVERSITY
408 Old Main
UNIVERSITY PARK,PA 16802-1505
Performing Department
(N/A)
Non Technical Summary
In response to the obesity crisis, poor dietary quality, and substantial disparities, a combination of policy levers, including updated dietary guidelines and redesigned nutrition facts labels, together with industry-led product reformulations, may be improving the outlook for diet quality. In 2016, the back-of-package nutrition facts label was made changed so that existing information, such as total calories, was made more noticeable and new information was added.The literature has documented mixed results of how consumers will react to the new labeling regime, which might be explained by multiple mechanisms. A strong theme emerging from studies based on market data is that consumer preferences for nutrition are hard to change, and diet quality is not likely to respond to external influences. If rigid consumer preferences are the predominant mechanism at work, then nutrition awareness policies may be ineffective, at least in the short term. However, public health and nutrition policies may be more effective if poor diets and dietary disparities are determined, instead, by obstacles, barriers, or constraints that affect some consumers more than others. Furthermore, while these obstacles and constraints might be related to factors that are often observable, such as a household's income or food environment, they might also be related to factors more difficult to observe, such as a household's nutritional literacy, the ability to obtain, understand, and use nutrition information. Other difficult-to-observe factors include a household's level of food-preparation facilities, storage infrastructure, cooking skills, or time constraints. Finally, even households with a good level of nutrition literacy and cooking skills may still face obstacles involving poorly labeled food or missing nutritional information. In other words, some households that prefer a healthy diet, all else equal, may have insufficient time, facilities, or nutritional awareness to obtain it.This proposed research intends to investigate the impact of the updated nutrition facts label on consumer choices, product reformulations by food companies, and the overall diet quality of U.S. consumers. More importantly, it also seeks to untangle the interrelated issues of nutrition literacy, nutrition labeling, and preferences to healthy or unhealthy food choices; and mechanisms that underly the associations and linkages. To succeed in untangling and identify mechanisms, the proposed research uses two complementary methods: (i) experiments to isolate the causal role that individual factors, such as nutritional knowledge and nutritional literacy, play in food choices; and (ii) market-purchase data (so-called scanner data) applied to both reduced-form and structural causal models to investigate how nutrition information, as well as changes to the nutrition facts panel, affect food purchases.This proposed work stresses the link between nutrition information and public health outcomes, including obesity and diabetes, that stem from generally poor diets of Americans. While the Dietary Guidelines for Americans provide strong consensus on what constitutes a healthy diet and how diet quality should improve, there is ongoing debate regarding the most effective tools to achieve it. Diet quality depends on the nutrition profile of the products made available by food manufacturers as well as on what consumers choose to purchase and eat. An overarching goal of this proposal is to provide empirical evidence from multiple methods on the role and importance that nutrition information, nutrition labeling regimes, and consumer attitudes and preferences play in dietary choices. Our project's results would be particularly useful for a wide range of stakeholders, especially food and health policy makers, and advocates. For example, policy makers and nutrition advocates would find it useful to know if and how nutrition labeling regimes impact consumer behavior or if behavioral barriers intercede.
Animal Health Component
100%
Research Effort Categories
Basic
0%
Applied
100%
Developmental
0%
Goals / Objectives
The overarching goal of this proposal is to investigate the impact of the updated nutrition facts label on consumer choices, product reformulations by food companies, and the overall diet quality of U.S. consumers, a concept that - for the purposes of this proposal - we use interchangeably with food healthfulness. To accomplish this goal, we formulate four main objectives.Objective 1: Identify the mechanism through which nutrition labels may affect the healthfulness of households' food choices using economic behavior experiments.Objective 2: Develop both consumer demand and product supply models to investigate the equilibrium impact of nutrition labels on consumer choices, firms' pricing and product reformulation decisions, consumer welfare, nutritional intakes, and population health using reduced-form and structural analyses of store-level food purchase data.Objective 3: Evaluate the heterogenous impact of nutrition labels on households' overall diet quality across household types defined generally unobservable nutrition-related variables (e.g., nutrition label reading habits, nutrition literacy, and importance of nutrition to consumers), using a natural experiment and machine learning coupled with a combination of consumer-level food purchase data and economic behavior experiment data.Objective 4: Identify and evaluate the effectiveness of various nutrition labels related policies aimed at improving the healthfulness of food choices under alternative mechanisms.
Project Methods
For Objective 1, "Identify the mechanism through which nutrition labels may affect the healthfulness of households' food choices using economic behavior experiments", we plan to casually identify the impact of nutritional information on nutrition literacy, and its impact on healthy eating across consumers with different socio-economic statuses (SES). We will collect information on participants' nutrition literacy and food choices that reflect eight different conditions using a four-by-two experimental design: First we vary the nutrition information participants receive (i.e., no information, specific information on added sugars, specific information on serving sizes, or general nutritional information). Second, we vary the labeling system participants see (i.e., an old nutrition facts label or a new, updated nutrition facts panel). Participants will be presented ten food items and asked to rank them according to their nutritional scores, and separately, rank them according to their preferences. The order of these two rankings will be randomized. Both rankings will be incentivized: We will reward participants for correctly ranking the food items according to their actual nutritional score. For the food-preference ranking, we will deliver a subset of the chosen foods to participants by randomly selecting the number of items in their preference ranking to deliver. After the experiment, participants will be surveyed to collect household traits and SES characteristics, nutrition label reading habits, shopping habits, and general nutritional literacy. We will then prepare a manuscript summarizing the findings of the economic experiment and the survey. The experiment will allow us to identify the role of the nutrition facts label and nutritional information treatments in correctly identifying the healthfulness of different food items; identify how the nutrition facts label and/or nutritional information treatments affects participants' food choices; and identify heterogeneous effects for different SES households regarding nutrition facts label and nutritional information treatments on nutrition literacy and food choices.For objective 2," Develop both consumer demand and product supply models to investigate the equilibrium impact of nutrition labels on consumer choices, firms' pricing and product reformulation decisions, consumer welfare, nutritional intakes, and population health using reduced-form and structural analyses of store-level food purchase data."This objective relies on detailed IRI retail-level food-purchase data acquired in collaboration with USDA's Economic Research Service. Quantitative analyses of the data begin with reduced-form models that analyze the impact of added-sugar and serving-size changes to the nutrition facts panel on product sales. Next, structural demand models estimate demand as a function of product attributes and will be used to investigate specific cases chosen to isolate the role of serving size changes and added sugars. A structural product-supply model will also be estimated to complete the equilibrium, thus allowing for policy-based counterfactuals and simulations. All estimated models will be investigated to see if labeling regimes affect heterogeneous households differently and if behavioral impacts vary across product types.We will then prepare two manuscripts summarizing the findings of the scanner scanner data analysis.In objective 3, "Evaluate the heterogenous impact of nutrition labels on households' overall diet quality across household types defined by generally unobservable nutrition-related variables using a natural experiment and machine learning coupled with a combination of consumer-level food purchase data and economic behavior experiment data", we plan to further examine the heterogeneous effects of the nutrition facts label change across households with different types defined by generally unobservable nutrition-related variables (e.g., nutrition label reading habits, nutrition literacy, and importance of nutrition to consumers). Different from the previous two stages, which focus on a single or several food categories, we plan to estimate how overall diet quality is affected by nutrition labels through the natural experiment of nutrition fact label change. Understanding the potential heterogenous effects will further shed light on which mechanism is dominating consumer's food choices, consumer preferences or external obstacles, and identify the corresponding cost-effective public health and nutrition policies. We will match data collected in the experiment on individuals' nutrition label reading practices and other nutrition-related characteristics with detailed consumer-level food purchases found in the IRI Consumer Panel. The study team will use machine learning methods to identify in the IRI data those households that are more likely to read nutrition fact labels, have higher levels of nutritional literacy, and have food preferences linked to nutrition, all variables that would normally be unobservable in the IRI data. After matching the data, the study team will estimate a reduced-form model to investigate the impact of updated nutrition facts labels on consumers' dietary choices among consumers with different nutrition-related unobservables.In objective 4, we plan to Identify and evaluate the effectiveness of various nutrition labels related policies aimed at improving the healthfulness of food choices under alternative mechanisms.The analysis of each of the previous objectives will shed light on which mechanism is dominating consumer's food choices, consumer preferences or external obstacles. We will further perform policy analysis or policy simulations to compare and identify the cost-effective public health and nutrition policies to promote a healthier diet. The varied background of our co-PIs and communications with economists in ERS will ensure that we develop targeted and accessible outputs for all audiences.