Inequality aversion in health: Eliciting aversion to univariate health inequalities across different health distributions.

Objectives: Health inequality aversion parameters can be used to inform health-related social welfare functions (HRSWFs) when trade-offs between maximising and equalising health arise in health policy. Empirical estimates of inequality aversion among the public can be valuable to ensure that approaches to the trade-off reflect public preferences. We aimed to elicit inequality aversion parameters in a large population sample across different health distributions to inform HRSWFs of the Atkinson and Kolm-Pollak type and compare responses from different distributions.

Methods: We recruited a representative population sample to complete two sets of benefit trade-off tasks with different degrees of baseline health inequality. This allows to test for intra-individual concurrence between elicited inequality aversion parameters across two different distributions. Elicited parameters from each task-set are reported and compared across the two task-sets. Our study used univariate health distributions.

Results: 946 responses were retained after application of exclusion criteria. In our sample from Norway, we find median Atkinson and Kolm-Pollak inequality aversion parameters of 3.52 and 0.05, respectively, when pooling all responses. However, we do not find a convincing concurrence between elicited inequality aversion parameters from the two task-sets.

Conclusions: Our results indicate that most respondents are willing to trade-off at least some total health gains to favour the worse-off. We do not find a general concurrence between individuals' inequality aversion parameters when presented with distributions with unequal baseline inequalities, which suggests that our sample's distributional preferences may not satisfy the assumption of constant relative or absolute inequality aversion.