Introduction and structure of the response
We are very happy to see the engagement of so many eminent scholars in our field with our modest proposal for how to proceed with wellbeing policy formation in the UK. We are also heartened by the apparent endorsement of so many of them of the basic goal and essential characteristics of our proposal. Indeed, we see the contributions of John Helliwell, Arie Kapteyn, Mariano Rojas and Alois Stutzer as extensions and further background information to our proposal.
Our response is thematic. We first respond to the themes that add further perspective, depth and applicability before going on to the more critical themes that challenge our view.
Be even more ambitious
John Helliwell's main messages are: that the goal is life satisfaction; that the inequality of wellbeing matters more than the inequality of income; and that social connections are key determinants of wellbeing.
Arie Kapteyn tells us that wellbeing should change political dynamics between countries because of international status races.
Mariano Rojas argues that we should understand development in terms of wellbeing and not gross domestic product (GDP; or some other materialistic measure).
Alois Stutzer argues that wellbeing measures and their determinants should inform the choice between different democratic institutional frameworks.
Simply put: we agree. These are also like-to-haves that flow naturally from how we think. However, we deliberately kept our proposal tightly anchored to the institutions in the UK and the practical technocratic steps for how to proceed there.
Be strategic
Arie Kapteyn urges us to stay away from the status-race insights that come from the Easterlin Paradox discussions because he thinks they are politically a hard sell. He urges us to focus more initially on the less controversial parts of wellbeing knowledge.
We agree in principle that one needs to be pragmatic and evolutionary, but the Easterlin insights are not merely a political liability. They are a reality, unique insights arising from wellbeing analyses and also what combines the wellbeing agenda with sustainable development, inclusive growth and other more holistic goals for society that get us away from the orientation towards material welfare. So it is a politically double-edged sword to leave out the insights on status races. In fact, rather than being a political hard sell, insights on status races arising from wellbeing analyses can be considered a selling point of wellbeing data.
In addition, given that our role is that of advising social scientists, we cannot hide a key result of the wellbeing literature for fear that it is politically unpalatable. That would be more damaging to the credibility of wellbeing proposals than making specific suggestions for how status considerations can be brought into policy thinking.
We can add that in a 400-page handbook, Frijters and Krekel (Reference Frijters and Krekelforthcoming) go into greater depth on this issue, where they essentially take status preference formation as an engineering problem. A quote from Adam Smith exemplifies that line of thought: ‘the great secret of education is to direct vanity to proper objects.’
Institutional design
Nick Chater, Alois Stutzer and Mariano Rojas argue that we should inform the public and think of better institutions that engage them, rather than specific policy analysis.
We do not really agree with them. Academics do not of course make policy, but they should do their best to say what policy choices would be rational. It would be the ‘trahison des clercs’ (treason of the clerks) for academics to say ‘I strongly recommend to you a method of government but I refuse to advise that government on how to act.’ Instead, we should develop tools and evidence that can be used at any level of government or by any public-spirited non-governmental organization. We envisage that our kind of analyses would be publicly available and subject to all kinds of discussion with citizens and civil society. This is not a purely top-down approach.
The UK is of course more centralized in its general approach to policy-making than many other countries, and so the natural institutional approach there is to wonder how wellbeing considerations can be accommodated within a top-down approach. It is in this regard useful to remind ourselves that the UK already has a top-down approach to central resource allocation, guided by a ‘Green Book’ that HM Treasury publishes, which informs government departments, agencies and other UK institutions on how they can argue for central funds.
The 2018 Green Book already has an overall stated goal, which is: ‘Economic appraisal is based on the principles of welfare economics – that is, how the government can improve social welfare or wellbeing, referred to in the Green Book as social value.’ Hence, our proposal respects the reality of top-down resource allocation towards a specific goal and suggests an approach to make life satisfaction the goal rather than the economic surplus measures that are implicitly the goal in most cost–benefit analyses done in the UK at the moment.
This also answers some of the criticisms of Anna Alexandrova, Martine Durand, Ramandeep Singh and others who suggest that we should wait around for a more perfect measure before we become serious about wellbeing. The reality is that measures that are far more imperfect than life satisfaction (such as additional material consumption) are at this very moment driving resource allocations in central government in the UK and elsewhere. The choice is hence not between life satisfaction and some future, more perfect measure, but between what is used now and life satisfaction.
On the argument that one should sometimes provide information rather than choose anything for the population (Alois Stutzer, Martine Durand, Ramandeep Singh and Anna Alexandrova), all we can say is that we agree, but do not see this as a critique. Academics advise, governments decide. And in any case, governments themselves also disseminate a lot of information in order to help citizens choose. The UK, for instance, has weather bureaux and laws on business ventures that give information and guide interactions. Governments also bring out information on wellbeing for the citizenry, such as in the UK via the Office of National Statistics, which regularly produces detailed maps of wellbeing for the whole country.
Alois Stutzer argues that public policy in general should not be equated with the solving of technical valuation problems, but is primarily about understanding and addressing societal problems in a discursive democratic process. As we have said, this a strange and false dichotomy. It is not either/or. There are different layers of public policy, with politicians accountable to the electorate deriving broad policy guidelines in a discursive process and non-elected bureaucrats accountable to politicians implementing these policy guidelines. The latter have a rather technocratic aspect, and this technocratic aspect needs an explicit goal and methods to get there. Solving technical valuation problems is, therefore, a reality in government practice, and a wellbeing orientation should be reflected in that practice. If not, then wellbeing to most intents and purposes remains window dressing.
Beware of imperfection
‘Never let the perfect be the enemy of the good.’
Dan Benjamin, Kristen Cooper, Ori Heffetz and Miles Kimball say that life satisfaction is imperfect and not yet ready for primetime. Their alternative is the suggestion in their 2014 article to have a weighted average of many life goals, where the weights are determined by an aggregation of how individuals trade them off in hypotheticals posed in surveys (Benjamin et al., Reference Benjamin, Heffetz, Kimball and Szembrot2014).
On the first point, we have to say that life satisfaction is more ready for primetime than GDP or the EQ-5D was when they went primetime in the 1950s and the 1990s, respectively. We have millions of observations and thousands of studies to draw upon, compared to maybe a few dozen when GDP went ‘live’ despite all its problems.
Because we offer an institutional trajectory wherein better measures can over time replace the incumbent best measure for wellbeing, we do not really see the comments of these authors as decisive.
The alternative they offer does not seem practical for the same reason that indices combining many questions are not practical. Asking respondents lots of hypothetical trade-offs is known to be problematic, needs a lot of time in surveys and would require policy-makers to have causal estimates of every policy instrument on all of these trade-off dimensions. By the time some estimates come online, several more dimensions, questions and hypothetical trade-offs would need to be added to the measurement instrument, as we have seen with the indices of sustainable development that now have something like 169 variables in them. It is in addition not clear that these stated preferences can take into account spill-overs between individuals (where something that I want makes someone else worse off), whereas these automatically appear in measures of subjective wellbeing.
Martine Durand, Anna Alexandrova and Ramandeep Singh argue that life satisfaction is imperfect and what we know about it is imperfect. How about context and changes over time in what matters to life satisfaction, they ask?
Everything government does is imperfect, including all of the data it uses and all of the procedures it implements. So whilst we in principle agree that more localized and contextual information could improve things, we should also note that perfection is an illusion.
Using the best information one has at present is entirely normal for bureaucracies, even if that means presuming British citizens react similarly to certain stimuli as do the Swedish. This is the default already when it comes to medicines and basic human proclivities, unless there is clear evidence to the contrary. Indeed, we assume a certain degree of similarity in how different individuals respond to the same situation, which is why the laws we apply in Cornwall are the same as those in Devon, and the education curriculum in Lancaster is the same as that in Kent. Of course, there are some differences in how people respond, and institutions do incorporate some of that knowledge. But the way our institutions learn is not by knowing everything beforehand, but by initially applying a reasoned general rule of thumb and adjusting along the way as things unfold.
Limited knowledge has not stopped institutional change in the past, and the current situation is full of missed chances. Wellbeing policies thus bring more attention to things such as mental health, relationships and the limits of what material welfare brings, which one can see as the missed chances of the current situation due to the imperfections of the previous measures. Wellbeing policies will help overcome those missed chances for improvement and no doubt create new changes due to its imperfection. This will lead to change over time, just as policies in health, defence, education and even democratic institutions have changed over time when mistakes became known. To insist that we cannot push towards a better future because we are not sure about every pebble on the road ahead effectively favours the status quo and implies a distrust in our ability to adjust.
Regarding the specificity of context, the changes over time in what matters to life satisfaction and the critique that one focuses only on average effects, we agree in principle that this can be problematic and only be a second best. Ultimately, however, this calls for more wellbeing research rather than being a fundamental critique of the method.
Martine Durand suggests that we should have a dashboard as in New Zealand, saying that no OECD country yet uses life satisfaction in the way envisaged. We agree that our proposal is novel, though really it represents only a small step within the UK institutional context. Dashboards and indices are part of Rojas’ stage 2 of the wellbeing movement (the pre-policy phase). One problem is that if one bases trade-offs on a dashboard, one must choose weights. Another is that, for policy formation, one must in principle have causal estimates for all the elements in the dashboard. Perhaps most importantly, dashboards can be easily gamed by anyone in the policy arena.
The New Zealand initiative is a good example of the problems you get with dashboards. As we understand their institutions from how one of their architects, Tony Burton, presented it to us in London in 2019, the New Zealand Treasury effectively allows different government departments to pick their own goals from a large dashboard, not asking them to work out how their actions affect other goals and not openly adopting any weighting on the different goals. This allows departments to keep doing what they were already doing, picking whatever labels suit them best, thus effectively protecting them from having to engage with wellbeing knowledge.
That approach is surely better than nothing, but it is not far away from window dressing. If one truly cares about wellbeing, one goes with the best measure available, partially because that forces institutions to truly learn the general lessons on wellbeing.
Anna Alexandrova, Nick Chater, Martine Durand, Alois Stutzer, W. Kip Viscusi and others argue that there is (possibly) more to life than life satisfaction. There probably is, but: (1) what we suggest is a trajectory of improvement, not an overnight revolution that wipes away all policy development as we know it; and (2) there is more to life than what governments take account of right now, so the question is life satisfaction versus the status quo.
Nick Chater and Martine Durand argue that actual policy formation and human motivations are extremely messy and cannot be captured or replaced by processes oriented around a single measure. This is true, but we think that these authors underestimate the degree to which central bureaucracies and large human groups gravitate towards overall goals to guide trade-offs, meaning that the question becomes what measure is best. Part of the relentless logic of a hierarchical bureaucracy is the need for accountability, and this leads to the emergence of a focal point, as it has done for decades in the UK. Wellbeing understood as the aggregation of what the citizenry think about their own lives is a focal point that, in the UK at the moment, would displace a more materialistic focal point.
The deeper point is that one is trying to make ideals become reality rather than accepting things as they are. Our proposal is meant to help quite longstanding ideals shared by many humans over many cultures and time periods become more of a reality. The alternative that Nick Chater and Martine Durand implicitly offer is to presume that muddling on is somehow optimal already, and that our proposal can only upset the fine balance of current policy formation.
Specific issues
Alois Stutzer worries about the manipulation of surveys if resources depend on it. We think this is actually less likely than it may seem, because most policies would be based on expected and experimentally proven effects, not on the results of instantaneous measurement. In addition, as we flagged, we are thinking of a trajectory of measurement changes such that one relies more on measures that are harder to manipulate where that is a strategic worry. And precisely in order to prevent the doctoring of lots of results, we advocated an open Intergovernmental Panel for Climate Change-type process for deciding on and advertising key current presumed causal effects so that they might be challenged by better estimates.
Arie Kapteyn tells us our figures 2 and 3 should not be read causally and may not be the most representative of the results out there. However, our goal in the research behind these figures was to escape from the multiplicity of non-comparable studies by adopting one outcome (life satisfaction) and simultaneously estimating the predictive power of all of the main explanatory factors. We explicitly recognize that in this endeavour it is not possible to adequately and simultaneously instrument each of these factors. Moreover, when it comes to mental health, we would emphasize that it is not the same as low life satisfaction, which can come from many possible factors – one of which is the vulnerability measured by having ever been diagnosed as mentally ill.
W. Kip Viscusi argues for the superiority of economists’ calculations of the value of a statistical life year (VSL). The VSL comes out of standard economic thinking and so the critique of Kapteyn applies. The VSL presumes things that are no longer tenable as general assumptions, such as the idea that we cannot compare utilities but we can compare money, that there are no major consumption externalities, that what people want is more or less fixed and that individuals are rational about their choices. Clearly, the past two decades of behavioural economics have proven otherwise. Moreover, it is not obvious why the valuation of a truncated remaining lifetime should, as Viscusi claims, follow the same time path with age as the wellbeing experienced on a year-by-year basis.
Viscusi does, however, raise important issues on the relationship between traditional cost–benefit analysis and cost–effectiveness analysis using wellbeing as the measure of benefit. These issues are discussed in a handbook by Frijters and Krekel (Reference Frijters and Krekelforthcoming) and in an article by Frijters and Layard (Reference Frijters and Layard2018).
Traditional cost–benefit analysis measures benefits by willingness to pay in money terms. For some policy areas, this kind of evidence is very useful – areas such as transport or industrial policy. But this depends on people revealing their valuations through their choices. For most of government expenditure, there is no such evidence – such as in health, social care, child protection, law and order and redistribution. In such cases, discounted measures of wellbeing provide the only useful measure of benefit.
But how can we get coherent policy if benefits are measured in different ways? The answer is that they are not – because the money measures of benefit should always be weighted by the marginal utility of income by the person concerned, which turns it into a measure of utility or wellbeing.
Martine Durand presses the point that some data show low correlations between life satisfaction and positive or negative affect. Here, we refer to and concur with the comments of John Helliwell, who on that same basis argued that governments should pick something like life satisfaction as the goal, and not positive or negative affect. The low correlation found there then simply underscores that we do not currently have any clear policy use for measures of positive or negative affect.
In conclusion
We thank all of the reviewers for the time they have spent on our proposal and are heartened to see so many of them agreeing with the basic goals, adding perspective and further applications.
In terms of the main criticism that one would like more certainty about many aspects of wellbeing knowledge before becoming serious about policies and policy institutions, we fundamentally disagree with the implicit presumption that the status quo is a safe haven. We see many problems in this world and in our countries that are urgent, including the unsustainability of simply producing more and more material stuff at the expense of the environment. The misery that is there at the bottom of our societies also should not be ignored whilst waiting for even better measures to tell us just how miserable the bottom is. Current measures in UK cost–benefit usage insufficiently incorporate wellbeing knowledge, meaning that the status quo is not a safe haven. In short, let not the perfect be the enemy of the good.
We think that the 250-year wait since enlightened opinion adopted the idea that happiness should be the goal of government is more than long enough. The sensible way forward is to go with the immense amount of knowledge we have at the moment whilst, of course, continuing to work on improvements in all aspects of our wellbeing knowledge.
We urge our colleagues and friends working on wellbeing to pursue their ambitions and their hopes. Happier populations are a worthwhile goal to work towards.
