I: Against ‘a-contextual’ happiness economics

The push to position wellbeing at the crux of policy-making rests on having a high degree of confidence in the core characteristic of Frijters et al.'s chosen measure of life satisfaction, viz. its quantifiability. The reduction of wellbeing to a single number is what allows for the extensive econometric analysis of large-scale panel datasets that the authors reference to arrive at their estimates of key effects. The alternative – maintaining a multitude of indicators that reflect the complexity of wellbeing (its emotional, cognitive and social aspects) – would surely make this task more difficult, as it is hard to know the effect of a policy on wellbeing if wellbeing is not a single quantity. We understand the predicament and grant that trade-offs are necessary between having a manageable number and acknowledging that no such number properly represents the object in question. Still, we note that these authors project more certainty than the literature warrants. They claim that life satisfaction is singled out by the Stiglitz, Sen and Fitoussi report as ‘the most useful for those looking for a summary measure,’ when in fact the report is vocal that ‘because no single measure can summarise something as complex as the wellbeing of the members of our society, our system of measurement must encompass a range of different measures’ (Stiglitz et al., Reference Stiglitz, Sen and Fitoussi2009, p. 12). The multiplicity and diversity of wellbeing measures are by now well-known, and these authors paint a consensus where there isn't one.Footnote ¹

But for the present purpose, the more troubling oversimplification is the following: once we have estimates of the effects of various policies and circumstances on life satisfaction, for these authors these effects are generalizable in their applicability. Their methodology is built around the goal of identifying the effect of a variable on wellbeing as a uniform contribution of this variable in all circumstances and in all places (like the contribution of gravity to the motion of all bodies). In Frijters et al.'s table 1, the size of this effect is indicated in the column ‘Effect on 0–10 life satisfaction’. Sometimes the numbers listed are accompanied by a country; for example, –0.40 is the specific Australian effect of violent crime. However, in their section entitled ‘Cost–effectiveness’, the authors propose the ‘critical ratio’, which in their vision is clearly based on some universal net effect. This assumption provides little space for recognizing the importance of context, be it historical or cultural, such as the way that, say, poverty's effect on wellbeing may be mediated by place and time instead of being comparable and transferable. This latter point is familiar from the critiques of certain strands of evidence-based policy (Deaton & Cartwright, Reference Deaton and Cartwright2018; Cartwright, Reference Cartwright2019), but it is worth developing it in relation to this brand of happiness economics specifically.

Take again the aforementioned table 1, which summarizes the key findings from the literature on life satisfaction across a range of domains, such as work, education and environment. The evidence from the UK shows that an extra year of compulsory education is associated with a 0.03-point decrease in life satisfaction (measured on a scale of 0–10). Do such statistical relationships provide any meaningful insights for happiness-minded policy-makers or practitioners? Perhaps the authors are proposing to defund education, but given the absence of evidence on what factors, if any, might mediate and/or moderate the effects of education (and other such variables) on life satisfaction, the policy-maker is none the wiser.

Ideally, we should be trying to, in the words of Stiglitz et al. (Reference Stiglitz, Sen and Fitoussi2009, p. 12), ‘document the diversity of peoples’ experiences,’ rather than merely estimating average levels of wellbeing as Frijters et al. do. This, granted, is a lot to ask, and perhaps the authors are driven by pragmatic considerations of minimizing complexity, but the ambition to recognize context motivates existing efforts to take into account distributional concerns and inequalities in wellbeing. These assessments are predominantly sector-specific; they do not presume that there exists an average effect of a policy on wellbeing, but instead they estimate the impact of a policy on different population strands and on different aspects of their wellbeing (e.g., air quality, accessibility, affordability, etc.). This is the spirit of the UK Government's Department of Transport's (2011) guidelines for considering distributional impacts in the case of transport appraisal, and the Department for Environment, Food & Rural Affairs’ (2008) segmentation model for evaluating pro-environmental behaviours (which, recognizing the importance of context and differences, divides the public into seven subgroups). The fact that Frijters et al. call for a ‘common language on and understanding of wellbeing across government’ underscores their dismissal of these efforts and shows just what a radical change they are envisaging.

The lack of sensitivity to the role of context is further evident in the discussion of factors that explain the variation of national life satisfaction across countries. Frijters et al. argue that the ‘idea for policy is that variation between countries gives us hints as to what can be achieved by moving from the institutions, culture and systems operating in countries with low wellbeing to those with high wellbeing.’ In the long run, there is a call to ‘adopt the lessons of other countries with high wellbeing,’ for instance by ‘celebrating our lives and communities in the manner of the Costa Ricans’ or ‘embracing the positive psychology lessons of the primary school system in South Australia.’ Through this lens, institutions and culture are seen as entirely fungible – two tools among a set of innumerable others available in the policy-maker's toolbox for squeezing out the maximum ‘extra happiness per pound of expenditure’. The implicit assumption is that the key relationships are broadly similar across countries and regions.

To be fair to the authors, this assumption is shared elsewhere in the literature, such as by Blanchflower (Reference Blanchflower2008), who claims that the ‘structure of a happiness equation has the same general form in each industrialized country.’ This, however, stands in contrast to other developments in the literature on subjective wellbeing, which attend far more carefully to variations and cultural effects (e.g., Diener et al., Reference Diener, Oishi and Tay2018). Research by Lucas Stanca (Reference Stanca2010, p. 132) on the cross-country distribution of the relationship between economic conditions and wellbeing argues that ‘geography, culture and institutions must be explicitly taken into account.’ Stanca's (Reference Stanca2010) two-step methodology first estimates the effect of economic conditions on wellbeing at the individual level for each sample country and then investigates the relationship between country-specific sensitivities of individual wellbeing and macroeconomic and social conditions and geographical location. For example, Stanca (Reference Stanca2010) found evidence of spatial dependence in the cross-country distribution of the micro-level effects of unemployment on wellbeing, even after controlling for macroeconomic and sociodemographic conditions. The fact that the geographical pattern of the sensitivity of wellbeing to unemployment cannot be accounted for by socioeconomic conditions in neighbouring countries points to the role that local culture and institutions play in mediating the effects of unemployment on wellbeing.

There are hopeful signs in the literature – scholars are adopting methods and approaches specifically attuned to picking up differences. But Frijters et al. and related publications from this group such as Clark et al.'s (Reference Clark, Flèche, Layard, Powdthavee and Ward2019) Origins of Happiness are not well equipped in this regard. Even as their data reveal a substantial variation in the ways that local norms and institutions mediate the impacts of unemployment and poverty, their eyes are set on a universal range of factors and on estimating their context-free average coefficient. There are rich traditions of research using qualitative and mixed methods and of different kinds of professional knowledge about wellbeing – traditions capable of filling these evidential gaps and that practitioners of wellbeing policies value (Bache, Reference Bache2019) – but this does not cross the authors’ horizons. Notably, today's economics profession is beginning to acknowledge this problem and call into question the tendency to ‘dismiss disciplines that rely more on qualitative analysis and verbal theorising’ (Naidu et al., Reference Naidu, Rodrik and Zucman2019, p. 4), urging a corresponding reform of economic education (Bowles & Carlin, Reference Bowles and Carlinin press). Happiness research needs such methodological diversity more than ever.Footnote ²

II: Against imperial happiness economics

As we argue in the previous section, Frijters et al. see wellbeing as a tractable quantity that exhibits stable relationships with various factors in the environment. Panel datasets and randomized controlled studies enable the detection and estimation of these relationships. This is the model of knowledge production that these authors favour. The next step – a model of policy-making – makes total sense in light of their previous steps: if we assemble a central library of these stable relationships, we can direct policy far away by consulting this library. An imaginary message from the librarian to the client could go this way: ‘Is wellbeing suffering in your community? Let me check what our library offers … You could start cognitive behavioural therapy with such and such coefficient of effectiveness or you could invest in preschool programmes with a different coefficient. Do what you have the budget for.’

What could be wrong with this model? In our view, the problem is the power asymmetry between the librarian and the client. The librarian is the one with authority and the keys to knowledge, not the client. The best way to appreciate this is to dwell on the remarkable analogy that Frijters et al. themselves supply: the imperial botany at the nineteenth-century Kew Gardens. Usually, commentators wishing to ring the alarm bells have to go digging for the troubling precedent themselves, but here the authors surprise us with their openness. Just as Kew Gardens systematized knowledge about plants and supplied samples and best practices across the Empire, they argue, so should the What Works Centre do for wellbeing once it fully adopts Frijters et al.'s model.

Not so fast. Historians of imperial botany have long recognized that Kew's monopoly on knowledge in the name of communicability, order and cumulativity had a very dark side, namely erasing local and indigenous knowledge of species and local agricultural practices, all the while strengthening imperial control (Drayton, Reference Drayton2000; Bonneuil, Reference Bonneuil, Bourguet, Licoppe and Sibum2002). In much the same way, Frijters et al. hand the reins of the wellbeing agenda to the expert technocrat. The irony is particularly stark given that the authors express the ideal that ‘no single individual or group owns this agenda,’ and they even call their model ‘decentralized.’ It is neither: only the expert-librarian gets to define what wellbeing is and gets to decide what works.

If the authors share our concerns, they could retort that a single analogy with unsavoury baggage does not by itself incriminate their whole proposal. But our point is that their model is a model that, at least on the face of it, produces and enforces a universalist, ‘a-contextual’ and technocratic vision of both science and governance. There is no space in this model for participatory knowledge, for grassroots citizen science, for genuine pluralist debate and for autonomous deliberative decisions according to a community's values. In the literature on public policy, all of these are well-recognized antidotes to the rule of experts and its attendant dangers, be they paternalism or ignorance of local specificities.Footnote ³ It is hard to exaggerate the importance of avoiding these, given that the costs of alienating the public from the process of governance are so painfully clear in today's politics. In Frijters et al., there are undoubtedly humanist intentions, but along with those we detect a distinctive ideology of ‘improvement’ on the terms of the powerful, and their rhetoric helps to cement it. Thus, the imperial tinge of their model will not go away with abandoning the Kew Gardens analogy. Nor will it go away if they remove another colonial phrase they use – ‘the Anglo-Saxon tradition’ (even the French don't get any credit?). It will not do to clean up the text. Their version of happiness economics needs decolonizing.

III: Taking critics more seriously

As expressed, we harbour a deep uneasiness with the implicit imperial sensibilities of Frijters et al.'s proposals. Some of our concerns are epistemic (wellbeing science would be better off with richer measures and more diverse methods), while others are ethical (the model of policy-making they propose does not recognize the need for legitimacy, deliberation and respect). This is entirely unsurprising given that their vision is grounded in classical utilitarian ideas. Our final objection is to their attempts to make it look uncontroversial, continuous with other traditions and widely accepted by all reasonable people. The authors note ‘ideological resistance to wellbeing: a suspicion that the pursuit of happiness is futile and childish’ (the implication being presumably that the authors’ approach is evidence-based rather than ideological), and this is as far as they go in considering ethical objections to their proposals. According to them, their critics have nothing more to offer except for these implausible worries. This is not so, and we wish to set the record straight.

Take their treatment of cost–effectiveness analysis, where the only consideration is the extra happiness a policy delivers relative to the extra expenditure involved, measured in ‘net terms,’ and compared to a ‘critical ratio.’ This involves simply adding up the effects, not accounting for the fact that interventions may have differential impacts on individuals grouped according to their characteristics (e.g. socioeconomic, geographical),Footnote ⁴ nor the fact that these individuals may be entitled to fair treatment or to access to certain crucial resources, or that they may have some other rights that any decent society depends on. Utilitarianism on its own famously has no resources for dealing with values such as justice, rights and respect for people's own priorities (which can include goods other than happiness), and a major task for political theory and accounts of public policy is to figure out ways in which values should be balanced.Footnote ⁵

These are old and well-known arguments, and they are no more ‘ideological’, ‘futile’ or ‘childish’ than what Frijters and coauthors themselves have to say. Is it ideological to object to making welfare payments conditional on submitting to therapy? Is it ideological to worry that positive psychology's focus on individual treatment of misery obscures the social and structural determinants of this misery? Or that promoting happiness therapies and resilience training amounts essentially to victim-blaming, whereby those of us who cannot deal with the world as it is are expected to retrain ourselves in order to keep up? These are all published concerns against Richard Layard's – one of the coauthors – approach to mental health policy.Footnote ⁶ For Frijters et al. to claim that objections to their vision are exhausted by scepticism about the very idea of happiness – notably its unattainability and childishness – is a grossly unfair misrepresentation. Unfortunately, this is not the first time such has happened. At the December 2016 launch event for The Origins of Happiness, Layard referred to his detractors as ‘puritans,’ another trope that falsely implies the absence of any criticisms worth attending to.

This is not to say that the critics are right, but only that they raise substantive concerns and that the authors have not engaged with them.

Final thought

The three problems we discussed here point to a better version of happiness economics – one that recognizes context, is properly inclusive and pluralistic and attends to the moral and political complexities of policy-making with a richer set of concepts beyond bare classical utilitarianism. Some aspects of the vision presented in ‘A happy choice: wellbeing as the goal of government’ are certainly compatible with these ideals and, if the authors find them attractive, we would welcome their thoughts on how they can be accommodated.