GLP-1 Drugs and Food Demand
Paper Session
Monday, Jan. 5, 2026 8:00 AM - 10:00 AM (EST)
- Chair: Jill J. McCluskey, Washington State University
Weight-loss Drugs Shedding Away Demand for Ultra-processed Foods
Abstract
U.S. Obesity rates continue to rise even though average caloric intake has plateaued in recent years, prompting significant interest in identifying both underlying causes and effective interventions. One line of inquiry centers on the increased consumption of ultra-processed foods, which have been associated with poorer diet quality and higher all-cause mortality. Concurrently, glucagon-like peptide-1 (GLP-1) medications, originally developed to treat type-2 diabetes, have gained popularity for weight management due to their appetite-suppressing effects. These medications may not only reduce overall caloric intake but also shift users’ food preferences. Inthis paper, we explore how initiating GLP-1 therapy affects household demand for foods grouped by their level of processing. We employ a rich dataset that merges household scanner data from 2020 to 2023 with pharmacy prescription data to identify when individuals begin taking GLP-1 medications. This linkage allows us to measure how households alter their purchasing patterns once they start using the medication. To classify purchased items, we rely on the NOVA Food Classification system, which consists of four categories: (1) unprocessed or minimally processed foods, (2) processed culinary ingredients, (3) processed foods, and (4) ultra-processed foods. We manually labeled 17,000 products and then trained a natural language model to classify the remainder of purchased products. We use the difference-in-differences approach proposed by Callaway and Sant’Anna (2020) to identify the causal effects by exploiting variation in the timing of GLP-1 medication initiation. Our results reveal a significant reduction in monthly household food expenditures of approximately $53 after the start of GLP-1 therapy. While spending declines in all four NOVA categories, we detect notable shifts in budget shares across these groups. Specifically, the share devoted to unprocessed or minimally processed foods increases by 3%, whereas the share allocated to ultra-processed foods decreases by 2%. The shares of processed culinary ingredients and processed
GLP-1 Use and Protein Demand
Abstract
Consumers are increasingly using glucagon-like peptide-1 (GLP-1) receptor agonists to lose weight, and food manufacturers are changing the contents and marketing of their products in response. However, we know little about how the demand for food is changing as a result of GLP-1 use. Noting the prevalence of protein consumption in GLP-1-related discussions, this research focuses on how aggregate food spending and protein demand changes in response to GLP-1 use. Prior research suggests that these impacts may be confounded with other consumer characteristics and, as such, we estimate GLP-1 effects on protein demand by extending causal inference methods to structural food demand modeling. We obtain GLP-1 use information and choice data from a publicly-available survey. We use matching procedures to balance measured covariates between GLP-1 users and non-users. We then estimate the effects of GLP-1 use in an interval-censored model of food expenditures, a multinomial logit model of protein choice, and a protein-focused Almost Ideal Demand System. We find that GLP-1 use increases household food-away-from-home expenditures by $11.26 to $13.17 weekly but does not have statisticallysignificant effects on food-at-home expenditures. Willingness-to-pay for protein products increases when using GLP-1 medications, with the magnitude of the effects varying across product and outlet. Further, we find smaller own-price elasticities for some products (i.e., own- price elasticities become up to 0.40 closer to zero). Our results indicate shift effects of GLP-1 use on protein demand and that causal inference methods can improve heterogeneity assessments in structural demand modeling.
Discussant(s)
Brian E. Roe
,
Ohio State University
JEL Classifications
- D1 - Household Behavior and Family Economics
- I1 - Health