Food carbon taxation and distributional equity: Comparing LCA and IO approaches

Food systems account for a large share of global greenhouse gas emissions but remain largely outside formal carbon pricing frameworks. This paper develops a household-level microsimulation model to assess the environmental, distributional, and nutritional impacts of a uniform food carbon tax in Ireland. Weekly food expenditures from the 2015 Irish Household Budget Survey (6,864 households) are converted into quantities using CPI-adjusted supermarket prices and linked to emissions intensities from both Processed Life Cycle Assessment (LCA) and Environmentally Extended Input–Output (EE-IO) models. The analysis compares the two emissions-accounting approaches and tests whether aggregation bias alters inequality estimates. Results show that food carbon taxation is regressive: under a €56/tCO 2 tax, the poorest decile faces median burdens of about 3.8% of income with LCA and 2.2% with EE-IO, compared with 0.5–1% in the richest decile. Nutritional vulnerability is also concentrated among low-income households, whose protein- and energy-linked food expenditures reach 2.0–2.3 times the population mean as a share of income. Despite notable item-level differences between LCA and EE-IO, both methods yield statistically similar inequality outcomes, and aggregation bias is limited. These findings emphasise that distributional effects are driven more by household income patterns than emissions-accounting choices. • Food carbon taxes are regressive. Median burdens in the poorest decile reach 3.8% of income (LCA) and 2.2% (EE-IO), compared with 0.5–1.0% in the richest. Kakwani values (–0.16 to –0.19) confirm regressivity. • Nutritional vulnerability is income-linked. Protein, energy, and fat linked expenditures account for 2.0–2.3× the population mean (as a share of income) in the poorest decile but only 0.5–0.6× in the richest. • Emissions-accounting choices change levels but not inequality. Post-tax Gini values shift only slightly (0.3102 to 0.311–0.314), and aggregation bias is minimal. • Untargeted recycling reinforces regressivity (Kakwani = –0.2479), while targeting the transfer to poorer households brings the distributional impact close to neutral (Kakwani = 0.0005).