Should subjective wellbeing, as measured by life evaluation, be the sole criterion for policies? This article answers this question negatively based on three arguments. First, it is important to distinguish between people's life evaluations, their emotional experiences and their sense of purpose; each has different drivers and consequences, implying that no single measure can adequately subsume the others. Second, while subjective wellbeing provides information missed by more conventional measures, the reverse is also true. This implies that information on the intrinsic importance of other key wellbeing dimensions cannot be derived from just looking at their instrumental value in raising subjective wellbeing. Third, the ‘utilitarian calculus’ implicit in subjective wellbeing regressions shines little light on normative decisions such as the attention we should focus on the worst off or on future generations. In contrast to the ‘automatic pilot’ approach to policies advocated by Frijters et al., this article favours an approach based on dashboards of indicators used to inform all of the stages of the policy cycles, as recently implemented by several OECD countries.

Happiness economics as advocated by Frijters et al. makes three assumptions: that policy should be based on facts about the net effect of a factor on happiness; that wellbeing policy should be technocratic and centralized; and that the only credible objections come from critics who do not value happiness. We argue that all three should be rejected and that the science and policy of wellbeing should instead be pluralistic, context-sensitive and participatory.

The value of a statistical life (VSL) monetizes the expected loss in wellbeing associated with the risk of death. The utility loss resulting from a fatality is central to the empirical framework for estimating the VSL. The VSL trajectory over the life cycle exhibits an inverted-U shape, following a trajectory similar to that of lifetime patterns of consumption. The U-shaped pattern displayed by happiness measures over the life cycle is the opposite of the inverted-U shape pattern displayed by the VSL. It is consequently inappropriate to use happiness measures as ordinal ranking substitutes for the VSL for the purposes of estimating the benefits of mortality risk reduction. Compared to ordinal wellbeing scales, the VSL also offers a variety of additional capabilities by providing a cardinal index of the unit benefits for changes in mortality risks.

Frijters et al. make a powerful and lucid case for a top-down approach to government, in which the maximization of wellbeing should be the ultimate goal. I argue, by contrast, for a bottom-up approach: that the variety of goals, plans, norms and rules that govern our lives should be the starting point for political discussion. From this standpoint, the goal of individual and collective decision-making of all kinds is the reconciliation of conflicting objectives and priorities on a piecemeal basis. The distinction between top-down and bottom-up approaches in political decision-making parallels debates between ‘foundationalists’ and ‘coherentists’ in epistemology.