Quantified Magnitude of Fat Tax / Food Subsidy Incidence

What the evidence says

A simulated New Zealand tax of NZ$2 per 100g of saturated fat raises retail prices in proportion to a food's fat content: butter by 71%, full-fat milk by 19%, sausages by 16%. A parallel simulated sugar tax of NZ$0.4 per 100g raises cordials/fruit drinks by 10.8%, tomato sauce by 12.9%, and sugar itself by 82.9%. These are modeled pass-through effects from a single simulation study, not measured retail outcomes.

Where an actual tax has been implemented and measured, the Mexican sugar-sweetened beverage (SSB) tax decreased purchases of taxed beverages by an average of 6% (−12 mL/capita/day) across 2014 relative to a pre-tax counterfactual, with the decline growing to 12% by December 2014. The reduction was not uniform by income: low-socioeconomic-status households cut purchases by an average of 9.1% (−19 mL/capita/day) across the year, reaching 17.4% by December, versus roughly 5.5%–5.6% for middle- and high-SES households. Purchases of untaxed beverages rose 4% (36 mL/capita/day), driven mainly by bottled plain water, with the largest substitution among middle-SES households (+5.9%, 53 mL/capita/day) and the smallest among high-SES households (+1.5%, 15 mL/capita/day).

Simulation-based dietary modeling of a saturated fat tax finds decreases in daily sodium intake (−59 mg/day) and total energy intake (−136 kJ/day). A fruit and vegetable (F&V) subsidy modeled the same way increases F&V intake (+42.0 g/day) but also increases sodium intake (+48 mg/day) and energy intake (+236 kJ/day) — the authors describe the sodium/energy increase as undesirable. When modeled as a standalone intervention rather than part of a tax package, the F&V subsidy does not produce a clear net health benefit: mean −13,000 DALYs, with a 95% uncertainty interval (−44,000 to 18,000) that spans zero.

A separate methodological finding concerns how food-expenditure response is modeled. Applying item-specific price elasticities conventionally to a saturated fat tax simulation produces a 1.92% decrease in total food expenditure, which the source authors identify as running opposite to economic theory; the same conventional approach applied to a sugar tax and an F&V subsidy produces "revealed" total food expenditure elasticities of −0.56 and −0.22 — also implausibly negative. Rescaling the model using an estimated total food expenditure elasticity (TFEe) reverses the direction for the saturated fat tax, to a 2.93% increase in food expenditure. This adjustment is not cosmetic: it reduces the modeled health gain from the saturated fat tax by 46%, from an unadjusted estimate down to 1,805,000 HALYs (95% UI 1,337,000 to 2,340,000) over the lifetime of the 2011 New Zealand population. The TFEe parameter itself is reasonably well bounded across studies, with New Zealand estimates of 0.68–0.83 and international high-income-country estimates of 0.46–0.90 (excluding a UK outlier of 0.04).

No claim in the current sources measures real-world health outcomes — obesity, disease incidence, or mortality — from an implemented saturated fat or sugar tax. The Mexican SSB tax evidence covers purchasing behavior only, through December 2014; the health-gain figures (HALYs, DALYs) for saturated fat taxes and F&V subsidies come entirely from simulation models, not observed health outcomes.

Where the evidence splits

Within the Blakely simulation, conventional price-elasticity application and TFEe-adjusted price-elasticity application disagree on the sign of the saturated fat tax's effect on total food expenditure: −1.92% under the conventional method versus +2.93% under the TFEe-adjusted method. This is not an unexplained cross-study conflict — the source states the difference directly: the conventional method produces a result the authors judge economically implausible (opposite to theoretical expectation), and the TFEe adjustment is applied specifically to correct that direction. The same conventional-method problem shows up for the sugar tax and F&V subsidy, which under conventional PE application "reveal" implausibly negative total food expenditure elasticities of −0.56 and −0.22 respectively.

The F&V subsidy's own reported effects run in opposite directions on different outcomes: it increases fruit and vegetable intake (+42.0 g/day) while also increasing sodium (+48 mg/day) and energy intake (+236 kJ/day). The source does not treat this as a data conflict to be resolved — it explicitly frames the sodium/energy increase as an "undesirable" side effect coexisting with the intake increase, attributing it to cross-price elasticity effects on other foods.

Thresholds and dose-response

The Mexican SSB tax's effect on purchases grew over the course of a single year rather than holding constant: purchases of taxed beverages fell by an average of 6% across 2014 but reached a 12% decline by December 2014, and among low-SES households the decline reached 17.4% by December versus a 9.1% average across the year. No source states a price level, tax rate, or expenditure cutoff at which an effect appears or reverses — the pattern established is a within-year growth in effect size over time, not a level-based threshold.

Open empirical questions

• Whether real-world retail price pass-through of a saturated fat or sugar tax matches the modeled pass-through rates (71%/19%/16% for saturated fat; 10.8%/12.9%/82.9% for sugar) — no measured pass-through data appears here.
• Whether the measured decrease in Mexican SSB purchases (6% average, growing to 12% by December 2014) translates into measured changes in body weight, obesity prevalence, or other clinical outcomes — the sources measure purchases only, through December 2014.
• Whether the F&V subsidy's modeled increase in sodium and energy intake, from cross-price substitution toward other foods, holds up in experimental settings beyond the two cafeteria studies the source notes as having evaluated competing-food substitution.
• What the true total food expenditure elasticity (TFEe) value is precisely, given the sensitivity range reported (0.42–0.96) and how that precision would change the size of the modeled saturated fat tax health gain (currently 1,805,000 HALYs, 95% UI 1,337,000–2,340,000).