Label or taxes: why not both? Testing nutritional mixed policies in the lab

We run an incentivized framed-field laboratory experiment to evaluate the interaction of labelling (Nutri-Score) and pricing policies (fat taxes and thin subsidies) on the food shopping of a sample of French consumers. Taxes and subsidies, designed to fit Nutri-Score, are differentiated according to their magnitude (large or small), and their salience (explicit or implicit). We exploit a Difference-in-Difference design, whereby subjects shop for real from a catalog of 290 products twice, first without any labelling nor pricing policy, and then a second time with one of five different combinations of labelling and pricing policies. We focus on the impact of different policy mixes on the aggregate nutritional quality and the total expenditure of the shopping. Results show that: i) when implemented alone, taxes and subsidies are less effective than labelling, especially when implicit and when small in magnitude; ii) with high taxes and subsidies, the policy mix is complementary to labelling, but strongly sub-additive; iii) with low taxes and subsidies, the policy mix is counter-effective and leads to lower impacts; iv) a cost-benefit analyses indicates that any pricing policy results in negligible gains for the consumer and large costs for the state.