The effect of food price changes on consumer purchases: a randomised experiment.

Background: Most evidence on health-related food taxes and subsidies relies on observational data and effects on single nutrients or foods instead of total diet. The aim of this study was to measure the effect of randomly assigned food price variations on consumer purchasing, where sets of prices emulated commonly discussed food tax and subsidy policies, including a subsidy on fruit and vegetables, a sweetened beverage tax, and taxes on foods according to sugar, sodium, and saturated fat content.

Methods: In this study, adult participants (≥18 years) in New Zealand completed up to five weekly shops in a virtual supermarket. Each shopping occasion was randomly allocated to control (no change in prices) or one or more pricing options simulating the following: a fruit and vegetable subsidy (20%), a sweetened beverage tax (20% or 40%), a saturated fat tax (NZ$2 per 100 g or $4 per 100 g saturated fat), a salt tax ($0·02 per 100 mg or $0·04 per 100 mg sodium), or sugar tax ($0·40 per 100 g or $0·80 per 100 g sugar). The primary outcome was the healthiness of the total shopping basket for each weekly shop (% of total unit food items defined as healthy). Low and high price change options were combined in analyses (eg, results for a saturated fat tax are an average of $2 per 100 g or $4 per 100 g).

Findings: Between Feb 1, and Dec 1, 2016, we randomly assigned 1132 shoppers, of whom 1038 (91·7%) completed at least one shop and 743 (71·6%) completed all five shops. Overall, data from 4258 shops were included in the analysis, including 645 control shops, 2545 shops where one policy was activated, and 1068 shops with two (or more) policies activated. In the control condition, 67·90% (SD 13·01) of food purchases were classified as healthy. Three of the five policies increased this proportion by a small, but significant amount (saturated fat tax mean absolute difference 1·77%, 95% CI 1·03 to 2·52, p<0·0001; sugar tax 1·09%, 0·26 to 1·91, p=0·0099; and salt tax 1·31%, 0·50 to 2·13, p=0·0016). The sweetened beverage tax and fruit and vegetable subsidy resulted in non-significant increases of 0·18% (95% CI -0·49 to 0·85, p=0·60) and 0·41% (-0·26 to 1·07, p=0·23), respectively. Both the saturated fat tax and salt tax resulted in the following important substitution effects: an increase in fruit and vegetable purchases as a percentage by weight of all food purchases (saturated fat tax 4·0%, 0·9 to 7·1; salt tax 4·3%, 0·9 to 7·7); but also an increase in sugar as a percentage of total energy (saturated fat tax 5·0%, 2·1 to 7·9; salt tax 3·2%, 0·0 to 6·5). Interaction terms for combined policies were mostly non-significant, consistent with additive effects of policy combinations.

Interpretation: Price changes representing saturated fat, sugar, and salt taxes increased total healthy food purchases. As we observed important substitution effects, a combination of different tax and subsidy policies might be the most effective way to improve diets and decrease diet-related chronic diseases.

Funding: Health Research Council of New Zealand.