Determinants of ultra-processed food purchases: Evidence of between- and within-household variation in France

Ultra-processed food consumption is associated with obesity and chronic diseases, but its determinants are not well documented. This understanding is essential to inform policies aimed at improving public health. To this end, we construct a unique longitudinal dataset of French households' food purchases from 2004 to 2015, which we merge with nutritional and food processing data. Using panel data econometric analysis, we identify the determinants of ultra-processed food purchases by exploiting variation between and within households and capturing the persistence of purchasing behavior over time. Moreover, we combine this analysis with an instrumental variable approach to obtain price elasticities. Demographics explain differences between households. Higher ultra-processed food purchases correlates with obesity, younger cohorts, poverty, rural areas, and Paris and northern region of France. Within a household, the purchase of ultra-processed food is influenced by income constraints. Time constraints, such as those imposed by having young children or being a one-adult household, also play a role, highlighting the link between ultra-processed food consumption and convenience, as well as limited time for food preparation. Nevertheless, a tax policy could effectively reduce ultra-processed food purchases. A 20% price increase would reduce ultra-processed food purchases by 19.5% within a year, with even larger declines in the long run, around 32%, due to persistent purchasing patterns. We show how different groups react differently to taxes, which has important consequences for designing effective policies aimed decreasing ultra-processed food consumption.