Many policymakers aim to reduce health disparities across groups. In fact, the distributional cost effectiveness analysis (DCEA) approach aims to incorporate treatment value based on the extent to which they reduce inequality. However, what types of inequality are we trying to reduce: ex-post or ex-ante? Ex post inequality means reducing inequality of outcomes at a given point in time. Ex-ante inequality focuses on equal access to care at a given time–even if outcomes do vary.
A key question is how can one estimate inequality aversion from an ex-post or ex-ante perspective?
A paper by Hansen, Kjær and Gyrd-Hansen (2025) summarize current approaches for esimation.
[an] ex post perspective…involves asking respondents to act as social decision makers and to distribute resources across current patient groups to which they do not belong…This approach aligns with the extra welfarist tradition of social welfare representing the sum of individuals’ utility. The “extra” refers to an expansion of the evaluation space, allowing for the reweighting of utilities to reflect the public’s inequality aversion…
Other researchers have sought to disclose individual distributional preferences using an ex ante perspective, whilst seeking to ensure that individuals are detached from their own position. In alignment with Rawls “veil of ignorance” approach…one strategy is to ask respondents to make choices on behalf of their imaginary unborn grandchild. An alternative approach is to follow Harsanyi’s strategy, which involves presenting individuals with equal probabilities of ending up in every possible future position. This approach has been applied in the literature to study distributional preferences for income…Harsanyi argues that this strategy will lead to fair social distributions that are aligned with the welfarist approach of expected utility as individuals will seek to minimize the risk of ending up in a particular bad situation.
The authors claim that only ex ante considerations are important for option value, which the authors define as “the value that the citizen places on securing access to a public (health care) service even if there is little likelihood of the citizen actually using it.”
Which approach generates more inequality aversion? Then Hansen paper aims to answer this question using the following approach:
We compare the preferences that are generated when applying the ex post perspective and the ex ante perspective using Harsanyi’s average expected utility approach [
Harsanyi 1955, 1975]. The approaches are different in that the ex post perspective focuses on eliciting weights to be attached to certain outcomes, whilst the ex ante approach seeks to evaluate individuals’ expected value of future uncertain outcomes including the option value. We hypothesize that because the ex ante perspective elicits distributional preferences inclusive of option value, this approach may elicit stronger preferences for equal distributions if private risk aversion is a stronger driver of distributional preferences than social inequality aversion.
Here is what their approach found:
Overall, we find that distributional preferences differ across the two perspectives, with equality objectives receiving more sup-port when the ex ante perspective is applied. Whilst the ex post distributional preferences mainly reflect preferences for access to treatment and health gains to others when need is certain, the ex ante distributional preference includes the private option value, that is, the knowledge of knowing that access to a specific treatment is ensured should one need it. Our results suggest that the insurance perspective, as presented in this study with equal probabilities of ending up in future states, drives a stronger preference for equality than the ex post perspective. This is most likely due to the high option value of knowing health care is available irrespective if a need arises.
These results seem very abstract but the paper itself provides key examples of ethical tradeoffs that people face. Should you give more access to treatments for people worse off or people for whom the treatment would work better? Should you fund access to therapies for more severe diseases or should the quantity of health gain be the only thing that matters? You can read the full paper here.
