GLP-1 Medication Adoption and Household Demand for Ultra-Processed Foods

This paper investigates how the introduction of Glucagon-Like Peptide-1 receptor agonist (GLP-1) medications affects household food-at-home purchasing behavior. Using household scanner data, we link detailed food purchases with household prescription purchases to identify the timing of GLP-1 adoption on food purchasing behaviors. To measure shifts in food processing levels, we classify more than five million products into four NOVA categories (i.e., minimally processed, culinary ingredients, processed and ultra-processed) using a hybrid system that combines best-practice hand classification with a high-accuracy machine-learning algorithm based on product descriptions and ingredient lists. We estimate causal effects of GLP-1 on food purchasing behaviors using the staggered-adoption difference-in-differences approach, comparing GLP-1 adopters with households initiating non-GLP-1 diabetes medications. We find that the introduction of GLP-1 leads to a statistically significant reduction of approximately $56 per month in total food spending and induces a systematic reallocation of the household food budget away from ultra-processed foods and toward minimally processed foods. Although dollar-value changes in ultra-processed purchases exhibit heterogeneity, the decline in the budget share of ultra-prcocessed foods and the increase in minimally processed foods are robust across specifications. These findings suggest that GLP-1 medications generate meaningful improvements in dietary composition, with implications for public health policy and for food manufacturers likely to be affected by shifts in consumer demand as GLP-1 adoption increases.