How do consumers respond to nutritional Front-of-Pack labels? Evidence from a natural experiment

The recent increase in diet-related health problems has led to much discussion about whether and how we can improve the quality of dietary intakes in the population. Policy makers have implemented various measures aiming to improve individuals’ diets. This paper evaluates one recent UK policy that aimed to improve information provision. We examine the introduction of Front-of-Pack (FOP) nutrition labels on the quantity and quality of individuals’ diet using the timing of introduction of FOP by British retailers as a natural experiment.

We use scanner panel data from the Kantar World Panel on about 1m transactions at British retailers by over 30,000 households between 2005 and 2010. The UK Food Standards Agency (FSA) recommended retailers to adopt nutrition labeling on five food categories. Using a difference-in-difference approach, we compare the consumer response for these five categories in supermarkets where nutrition labelling was adopted to that in supermarkets that did not introduce labelling, exploiting the differential timing of labelling in supermarkets as a natural experiment. 

Theoretical work by Stigler and Becker, Grossman and others suggest that informational interventions change individuals’ knowledge about product characteristics, but their effect on consumption might depend on individuals’ valuation of health and socio-economic characteristics. 

The empirical literature on the effect of labeling on individuals’ behavior is rather mixed. One difficulty is finding sources of exogenous variation in the provision of information. A strand of this literature has used experiments in supermarkets and laboratory settings. A second strand has investigated quasi-experiments primarily focusing on the introduction of the Nutrition Labeling and Education Act in the United States. Neither strand has found conclusive evidence on the effectiveness of labeling and on the format of labels that are most effective.

We contribute to both strands of the empirical literature. Compared to the experimental literature, we consider a very large number of actual purchases and exploit a natural experiment. Our paper is the first to investigate the effectiveness of such labels in an observational dataset in the UK. We consider the effect on quantity of purchases as well as quality of diet measured by the Nutrient Profile Score defined by the FSA. We also consider a number of socio-economic characteristics and consumers’ stated preferences towards healthy food.

Very preliminary results indicate that labeling induces consumers to buy fewer products and substitute to healthier food. This effect is stronger for consumers who indicated a preference towards healthier products prior to the introduction of the labeling. Future analyses includes the use of online vs in-store shopping, and a comparison of branded (unlabeled) and private-brand products in a triple difference approach; these results will be available shortly.