The Heterogenous Carbon Content of Food Consumption and Its Implication for Policy

Can countries reduce their carbon emissions by changing the composition of consumption instead of, or in addition to, changing the underlying production technology? If so, what type of policy would be most effective? I contribute to answering these questions by introducing a novel dataset linking barcode level food consumption choices to carbon emissions, describing new stylized facts and analyzing their policy implications. I first show that about of third of the heterogeneity of carbon intensity across products and households, defined as the amount of CO2 emitted per real euro spent, comes from differences within narrow product categories, rather than differences between product categories. I then derive four sets of results. First, a product-level optimal tax would reduce carbon emissions from food by about 10%, or 40% of the gap between current carbon emission and France’s 2030 objective. Second, well-targeted taxes are crucial for efficient policy: a product-level optimal tax achieves the same objective as a uniform tax at a 67% lower utility cost. Third, the impact of carbon taxes across households varies strongly with their consumption’s average carbon intensity, but not with their expenditure level. Last, I rank product categories according to the effectiveness of a carbon tax targeted at this category and show that setting a 5% tax on two detailed categories only would reduce carbon emissions from food by about 1%, at a utility cost of 1.5€ per household, or 0.1% of the average annual food expenditure in my data