Numerous experiments find substantial aversion to health inequalities. However, aversion to health inequality may not be independent of the source of that inequality. Some inequalities may be due to factors entirely beyond an individual's control, while others are not. Some inequalities may be considered fair or deserved, while others may not. The priority given to the worse-off may, therefore, depend not only on their health but also on other equity-relevant characteristics. Few experiments are designed to make this distinction. Ours is the first paper to separately identify health inequality aversion parameters and equity weights across multiple non-health domains - sex, income and smoking status. To do this, we develop a novel experiment design where participants allocate scare resources that differentially impact the health hypothetical individuals in society. In separate treatments, each with 10 rounds, recipients are 1) anonymous, and 2) identified by a) biological sex, b) income, and c) smoking status. Choices made within and across the treatments makes it possible to disentangle aversion to health inequality from weights according to non-health characteristics. We gather rich experimental data (42,900 obvs.) in an interactive online experiment with a representative sample of the UK adult population (n=582) recruited through Prolific. We specify social welfare functions and estimate the inequality aversion and equity weight parameters at the participant-level using a random behavioural model. Results show that the majority of participants exhibit a significant degree of health inequality aversion. We find that, on average, participants give a high weight to non-smokers over smokers, a moderate weight to the poor over the rich, and a small weight to females over males. The estimated health inequality aversion parameters combined with domain-specific equity weights have clear consequences for the allocation of resources across groups defined by the non-health characteristics and the prioritisation of those groups in response to baseline health inequalities between them. We illustrate these policy consequences using baseline data on Quality Adjusted Life-Expectancy. This highlights problematic recommendations of existing Distributional Cost-Effectiveness Analysis methods. We uncover reasons for heterogeneity across domains and between participants using detailed questionnaire data. This reveals that beliefs surrounding responsibility, luck and fairness drive heterogeneity in equity weights across domains and between participants.

Keywords: Inequality Aversion, Health, Experiment, Responsibility, Fairness, Social Welfare