Introduction
Democratic policy-making is fundamentally about searching for a consensus about the basic rules that allow people to best pursue their ideas of the good life, both individually and collectively. People delegated to serve in state and public authorities are thereby selected and motivated to support this pursuit with their work, their selection and motivation being shaped by state institutions.Footnote 1
It is within the foregoing procedural perspective of government and state that this short article comments on the role of empirical subjective wellbeing research in public policy.Footnote 2 In essence, evidence from this new and intriguing development in the social sciences is here seen to serve as informational input into the political process at all levels and for all actors involved, citizens being central. This is indeed most valuable, as it provides a steadily increasing basis of insights that are promising.Footnote 3 However, this perspective also rejects the idea that, based on the promises of the approach, we should adopt a new policy perspective that is oriented toward a decision rule maximizing wellbeing in terms of some aggregate measure of subjective wellbeing. This social engineering perspective, implicit in much reasoning about wellbeing policy, proposing a technocratic decision rule at the core of political decision-making, is not only not necessary in order to make productive use of the new insights, but risks: (1) disregarding some motivation problems; and (2) neglecting some basic aspects of the role of institutions in subjective wellbeing, as well as procedural utility as a source of subjective wellbeing.
In the following, I first briefly highlight some advantages of the evidence from the research on subjective wellbeing and its value for public policy. It echoes the proposal on how to make use of proxy measures of individual welfare based on individual self-reports in public policy by Frijters et al. (this issue). As practical instruments to support decision-making, the life satisfaction approach and wellbeing cost–effectiveness analysis (WCEA) offer practical extensions of the toolbox. Second, I argue that public policy in general should, however, not be equated with the solving of technical valuation problems. Public policy is primarily about understanding and addressing societal problems in a discursive democratic process. In fact, there are various challenges for wellbeing evidence-based public policy, even when not part of a wellbeing maximization approach. I briefly mention some of the obstacles. Third, a series of motivational issues are raised, challenging the implicit view of a benevolent social engineer and an unbiased representation of people's wellbeing. Fourth, I take up procedural utility as a key source of individual wellbeing that is neglected when people merely report mental states. Overall, I argue that wellbeing research should be oriented toward gaining insights that improve the diagnosis of societal problems and help us to evaluate alternative institutional arrangements to address them, both as inputs into the democratic process.
The subjective wellbeing lens: valuable complementary perspective on individual welfare
Traditional public policy in many areas is challenged by evidence that links possible policy outcomes directly to individual welfare consequences in terms of reported subjective wellbeing. This is a most welcome development, as it invites reflection and reconsideration of beliefs about the costs and benefits of various policies. Complementary attempts to assess policy consequences in terms of some direct proxy measure for individual welfare naturally focus on ends rather than means to an end, such as individual or aggregate income. The orientation is toward the subjective wellbeing of current and future generations rather than toward their material wellbeing.Footnote 4
Proxy for individual welfare
Based on the accumulated evidence, it is fair to say that the evaluation metrics applied in empirical wellbeing research fit people's appraisals of their lives better than the alternative metrics constructed around income and are thus comparatively good proxies for individual welfare (see, e.g., Clark Reference Clark, Adler and Fleurbaey2016).Footnote 5 At the very least, we can learn from people's reported judgments in addition to what they choose. We get ex post evaluations of individuals’ experiences.
These qualities render measures of subjective wellbeing an attractive tool for evaluating policies and institutions. In particular, measures of subjective wellbeing allow us to move toward an analysis of the net effects of policy interventions (including spillovers between individuals) and the experiences of particular circumstances (see Odermatt & Stutzer, Reference Odermatt, Stutzer, Diener, Oishi and Tay2018, for many examples and references). This includes the study of covariates of subjective wellbeing, which helps us to understand the critical determinants of individual welfare, involving both tangible and intangible (psychological) aspects. And it facilitates identifying institutions that enable individuals to better meet their preferences. In sum, based on measures of subjective wellbeing, it is possible to uncover welfare consequences that are partly unobservable when traditional measures of economic and social progress such as national income are used.
Learning about individual weaknesses reducing individual wellbeing
This is particularly relevant for policy evaluations in areas that might involve suboptimal behavior and where, for example, consumption vices turn into diseases.Footnote 6 The wellbeing perspective is thus compatible with the notion in behavioral economics that people do not always behave in their own best interests (weakening the assumption of consumer sovereignty in rational choice analyses and the strong (or exclusive) reliance on observed behavior as a welfare criterion). It provides a tool for policy evaluation in areas where decision utility (as reflected in behavior) systematically diverges from experienced utility (contributing to the approaches in behavioral welfare economics extensively discussed in Bernheim & Taubinsky, Reference Bernheim, Taubinsky, Bernheim, DellaVigna and Laibson2018).Footnote 7 If some behavioral choices are suboptimal from an individual's perspective, research can then seek to identify the conditions under which such choices are less likely or at least less harmful to the individual. Research on subjective wellbeing can thus provide a valuable input for the discourse on people's limitations in pursuing their ideas of the good life.Footnote 8
Valuing non-tangible and public goods
The valuation of non-tangible and public goods such as environmental quality or the fear of certain risks happening in terms of subjective wellbeing has direct utility in some practical policy work in which, formally or informally, the costs and benefits of a given project are analyzed. Otherwise, the benefits from public goods are difficult to measure, as the marginal utility to consumers is not reflected in market exchanges. Given that an appropriate empirical design is found, the marginal utility of public goods (or the disutility of public bads) is captured by calculating the partial correlation between the amount of public goods (or public bads) and individuals’ reported subjective wellbeing, controlling for many other determinants. Given this information, two kinds of applied welfare analysis are possible. First, by comparing the marginal utilities of public goods to the marginal utility of income, the trade-off between income and public goods can be calculated. This is the idea behind the Life Satisfaction Approach (Welsch & Kühling, Reference Welsch and Kühling2009; Frey et al., Reference Frey, Luechinger and Stutzer2010) as an additional evaluation method complementing the stated preference and revealed preference approaches.Footnote 9 Second, rather than transforming the benefits back into monetary terms to include them in the same units as costs in a cost–benefit analysis, information on benefits can be kept in wellbeing terms and directly related to costs. This WCEA proposed in Dolan and Fujiwara (Reference Dolan, Fujiwara, Adler and Fleurbaey2016) and Frijters et al. (this issue) has the advantage that it is not affected by any biases in the calculation of the marginal utility of income (an issue that has turned out to be most challenging). However, with WCEA, only projects can be compared that express at least some of the benefits in terms of subjective wellbeing (applying the same metric; e.g., an 11-point life satisfaction scale). WCEA thus becomes more interesting and informative the more projects have been analyzed for their wellbeing benefits per euro or pound.
So should we then maximize individual welfare based on reported subjective wellbeing?
Given the promises of the new wellbeing measures, to what extent should we rethink public policy? For Frijters et al. (this issue), the advanced possibilities for valuing projects lend themselves to a new perspective in formulating public policy. The perspective reads as follows: “If we envisage a self-financing tax change, we simply evaluate how this alters the happiness of each member of the population and aggregate these changes.’ Or ‘if we are considering a new regulation, we simply add up its effects on happiness across all members of the population.’ In the following, I will argue against this position for a new public policy paradigm. In particular, public policy in general should not be equated with the solving of technical valuation problems. Public policy is primarily about understanding and addressing societal problems in a discursive democratic process, and contributions from wellbeing research should be oriented toward this process.
Technical and conceptual challenges for policy analyses based on wellbeing research
Independently of how one evaluates the effectiveness (as well as the desirability) of the movement calling for more evidence-based public policy, the production of high-quality evidence is demanding and costly. This might hold specifically for the causal effects of some circumstances or some policy (intervention) on people's subjective wellbeing. I briefly present four challenges. In my view, they are important to keep in mind when addressing technical valuation problems in the cost–benefit or cost–effectiveness analyses mentioned above (and I suggest that we are still at the beginning with these applications). More generally, the technical challenges also indicate ‘security gaps’ of a maximization approach regarding the risks of mistakes as well as of manipulation. A procedural perspective is likely less to be prone to these kinds of risks, as the corresponding evidence is less directly linked to policy decisions and has first to survive the political discussion process.
Causal effects
To get an idea of the never-observed counterfactual wellbeing of a person if she or he were not to be treated with some life circumstances or some policy is very difficult. Most of the evidence in empirical research on subjective wellbeing is correlational. Unobserved and thus often omitted third factors driving both life circumstances and wellbeing outcomes often offer plausible alternative explanations for observed statistical relationships. Moreover, more and more evidence emphasizes how the motivational force of being in a happy state of mind affects people's behavior (see Lyubomirsky et al., Reference Lyubomirsky, King and Diener2005, for a review). For both mechanisms that hamper the causal interpretations of statistical relationships, a case in point is the correlation between individuals’ incomes and their subjective wellbeing.
In response, the search is for evidence from research adopting some experimental or quasi-experimental design. For future policy-relevant research, this is probably one way to go, but so far, the evidence base that lives up to this standard is rather limited.
Effect heterogeneity, extrapolation and scaling
Attributing an average wellbeing premium or reduction to some change in individuals’ life circumstances is not the only difficulty, however. The concrete policy questions might also involve particular groups of people for whom the effect of some change in living conditions is expected to be smaller or larger than the average. For example, an income transfer to sick people might improve their lot relatively more if privately financed health care services are the critical inputs for living well. However, if the crucial condition is social contact, more or less financial means would be expected to affect wellbeing relatively less (see, e.g., Finkelstein et al., Reference Finkelstein, Luttmer and Notowidigdo2013, for a related analysis).
Some well-identified treatment effect of certain living conditions might thus be of little help if effect heterogeneity is likely. Experimental evidence often captures local estimates regarding some change in the affected outcome variable that is difficult to transfer to other contexts or to extrapolate. The wellbeing effects of education may serve to illustrate this. Based on advanced econometric designs, the effect of an extra year of compulsory education in the UK (following a reform in 1972) has been studied. While a first study reported a positive effect on life satisfaction (Oreopoulos, Reference Oreopoulos2007), a second study detected a negative one (Jung & Clark, Reference Jung and Clark2017), the latter estimation being included in the table of key findings in Frijters et al. (this issue). Independently of whether the reform was beneficial to those bound by the reform in their decision to leave school, it is unclear whether the findings offer an ‘externally’ valid estimate for policies that increase years of schooling. Imagine, for example, a policy proposal for a mandatory earlier (pre-)school enrollment. The new regime would replace private (pre-)school and household internal arrangements that previously existed, affecting wellbeing in a way that is difficult to compare with the (long-term) consequences for children and parents of attending school up to a higher age and entering the labor market later (as underlay the original policy analysis). Moreover, conditions exploited in instrumental variable designs might affect people's subjective wellbeing via various other mechanisms than the one econometrically modeled. Given how broadly measures of subjective wellbeing pick up consequences on people's lives, such effects seem particularly relevant in wellbeing research. Both arguments suggest that the evidence should be put more in the context of the underlying policy variation (or the instrumental variables), offering more or less meaningful lessons for specific policy proposals.
The effects of a successful intervention might not even be representative if the very same intervention is applied again or rolled out to more people. Neglecting any publication bias, the policy experiment that has proven most successful in a competition of interventions is in fact more probably a false positive (Al-Ubaydli et al., Reference Al-Ubaydli, List and Suskind2019). Moreover, experiments might have taken place where the expected positive effects were largest and the people engaging in them most motivated (see, e.g., Helliwell et al., Reference Helliwell, Sachs, Adler, Bin Bishr, de Neve, Durand, Diener, Helliwell, Layard and Seligman2019b, p. 20). These two aspects add to possible negative network or equilibrium effects (see below) and diseconomies of scale.
Global measures
While global measures (e.g., of satisfaction with life) have the great advantage that they leave the aggregation and weighting of wellbeing judgments across different domains of life to the individuals surveyed, they come with the inevitable disadvantage that they are less sensitive to changes in some individual circumstances. Accordingly, effects on subjective wellbeing must either be very large to be detected with some statistical precision in an empirical evaluation or else very large samples are necessary. Specific evaluations then become rather costly. Thus, a trade-off emerges between adopting a comprehensive but less sensitive measure on the one hand and more specific indicators that help to assess the consequences for wellbeing on the other.
Equilibrium effects
If research provides information about outcomes linked to individual wellbeing, it is tempting to suggest policy interventions that strengthen positive outcomes and diminish negative ones by trading them off in order to obtain an optimal policy mix that maximizes some empirical indicator of subjective wellbeing. Unemployment is a case in point. One might propose that, in order to improve the lot of the unemployed, the social stigma related to unemployment should be attenuated, and, in order to reduce the economic insecurity of the employed during a crisis, employment protection should be extended (see, e.g., Luechinger et al., Reference Luechinger, Meier and Stutzer2010, for evidence on the wellbeing costs of general unemployment). However, such conclusions might be premature. Regarding the supporting of the unemployed, the personal psychological costs of unemployment due to a strong social work norm (see, e.g., Clark, Reference Clark2003; Stutzer & Lalive, Reference Stutzer and Frey2004) might be functional or even necessary in order to maintain a generous and financially sustainable social benefit system. These costs help to limit moral hazard; namely, that citizens do not search less intensively for re-employment when receiving higher unemployment benefits. With regard to the aggregate level, increased job protection might benefit those who already have a job (see, e.g., Clark & Postel-Vinay, Reference Clark and Postel-Vinay2009). However, it is also likely to make employers more reluctant to hire new workers, which is detrimental for the job prospects of those who are unemployed. Thus, stronger labor protection rights might lead to longer individual unemployment spells and to higher general unemployment.
These classic arguments highlight the importance of equilibrium effects and potential (unintended) side effects when thinking about policy interventions. Moreover, they highlight the limitations of an approach that focuses on an optimal outcome mix.
Incentive issues in wellbeing policy
Motivation and manipulation
Engagement for the explicit goal of an increase in people's wellbeing might provide a direction many people in the public sector can associate themselves with, contributing to positive public sector motivation. This very goal might indeed have been the reason that they chose to work in the public sector in the first place. Next to some intrinsic motivation, people involved in public policy face various external incentives and pursue their ideas of the common good as well as their own private good. These fundamental aspects are neglected when a wellbeing maximization approach is proposed that implicitly rests on the view of a benevolent and unconstrained government. The critique of traditional welfare economics (see, e.g., Mueller, Reference Mueller2003) thus also applies and holds lessons for productively dealing with wellbeing policy (Frey & Stutzer, Reference Frey, Stutzer, Congleton, Grofman and Voigt2019).
In general, some mechanical implementation of wellbeing measures in the policy process is likely to induce strategic interactions between all of the actors involved, including citizens. Once a specific aggregate happiness indicator has become established as being politically relevant, the government, public bureaucracy and various interest groups have an incentive to manipulate it. This has proved to be true for various policy-relevant indicators before, such as gross national product, measures of poverty and crime or the rates of unemployment and inflation. For example, when the unemployment rate had become a politically crucial indicator, governments started to influence it in order to paint a better picture of the state of the labor market than was actually the case (see, e.g., Hughes, Reference Hughes1976; Gregg, Reference Gregg1994; Levitas, Reference Levitas, Levitas and Guy1996, for the experience in the UK).
Misrepresentation and reactance
Similarly, when individuals become aware that the wellbeing level they report influences the behavior of political actors, they have an incentive to misrepresent it. Imagine, for example, a fiscal equalization scheme that takes into consideration people's subjective wellbeing across regions. Misrepresenting one's state of wellbeing in self-reports has little cost. Moreover, I am not aware of an incentive-compatible mechanism for the measurement of subjective wellbeing. People's motivation to truthfully report their perceived wellbeing is thus a crucial precondition for any wellbeing evidence-oriented politics. How can this motivation be maintained? How can non-deliberate or even strategic answers be prevented? Answers to these questions should be part of any broader wellbeing perspective on public policy.
The idea that people react if they feel threatened in their freedom is well established in psychology as reactance (Brehm, Reference Brehm1966). A counter-response might thus occur if people feel that they are part of an engineering problem that includes the reporting of their subjective wellbeing but little other involvement.Footnote 10
Avoidance and outcome orientation
Reactions to a wellbeing policy that is perceived as technocratic might be strategic or apathetic and neglectful. Under both conditions – for different reasons, however – a relationship between some outcome measure and people's subjective wellbeing may no longer be observed once policy measures try to exploit it. This might hold in particular for policy measures that involve any kind of transfer, be it in terms of money, services or personal attention. The problem that people avoid an intervention (or game the system) is well known in economics and has been dubbed the ‘Lucas critique’ in macroeconomics.
A subtler problem arises when the outcome measure in terms of subjective wellbeing becomes the primary orientation and replaces the necessary constant debate regarding the importance of various societal problems and the corresponding readjustment and reorientation of public policy toward more intermediate outcomes, as well as the further development of basic rules to provide a more productive constitutional framework. This essential part of the democratic process is seemingly superfluous when the policy problem can be reduced to the ranking of projects in terms of their contribution to some wellbeing indicator. It takes away an important and productive task in which traditionally parties and political entrepreneurs engage themselves alongside citizens and the public administration.
Procedural utility versus individuals as ‘metric stations’
Ideas about citizens’ involvement in government decision-making differ depending on the policy perspective. Conceptually, citizens play no active role when a wellbeing policy is oriented toward the maximization of reported wellbeing. Citizens are in fact reduced to ‘metric stations’. In other words (as cynics might express it), rather than having a say on issues in politics, they become targets of mental state management. This disregards the fact that citizens have preferences for processes over and above outcomes. They gain procedural utility from living and acting under institutionalized processes, which offer them possibilities for self-determination in the economy and polity, contributing to a positive sense of self. Specifically, processes contribute directly to people's wellbeing if they address innate needs for autonomy, relatedness and competence (for an introductory review, see Frey et al., Reference Frey, Benz and Stutzer2004). Lane (Reference Lane1988) emphasizes the procedural goods of democracy when people feel respected and treated with dignity and perceive some personal control, understanding and public resonance.Footnote 11 The ‘gilets jaunes’ movement in France might be a behavioral reflection of the consequences when people feel the absence of procedural qualities.
The political agency of citizens and an engineering perspective of wellbeing policy are thus difficult to reconcile. This holds for the reporting of subjective wellbeing, as well as the reporting of weights that should then be applied in the maximization of some social welfare function (for the aspect of asking people about the weight given to various groups in the aggregate wellbeing function, see, e.g., O'Donnell & Oswald, Reference O'Donnell and Oswald2015; Frijters et al., this issue).
Concluding remarks outside of the maximization paradigm
The policy paradigm matters for the choice of research questions and thus for the kind of knowledge wellbeing research aims to provide, as well as for the people seen as addressees. Based on the various considerations about a wellbeing-oriented public policy, I would like to draw six conclusions.
First, measures of subjective wellbeing offer a most interesting complementary indicator for evaluating public policies. While they can serve as close proxies for individual welfare, they should still not be used as a primary policy goal. Instead, they serve best in the diagnosis of societal problems and alternative institutional approaches to addressing them.
Second, keeping survey questions about subjective wellbeing as one of several inputs informing public policy rather than establishing them as key performance indicators that are directly linked to policy decisions will hopefully allow for the maintenance of the motivation of citizens to truthfully report their wellbeing.
Third, wellbeing research teaches us modesty in what we assume can be learned and known. On the one hand, the pretense of knowledge should be no issue due to the complexity arising from the organizational problems of interlinked societies. On the other, caution is advisable if one takes into account that people experience anticipatory wellbeing, adapt more or less to changes in circumstances, engage in social comparisons and form aspirations − all wellbeing-relevant aspects affected by their institutional environment.
Fourth, a perspective oriented toward the maximization of some indicator of wellbeing cannot address the normative issue; for example, of whether to tax those people more who more easily adapt to it. Following this line, there are several more issues that require a broader policy perspective. For example, should a wellbeing policy react to any of people's sensitivities or should instead processes be set up so that people reflect more on their own preferences? Or what about the manipulation of expectations, the chemical induction of wellbeing via psychotropic drugs or the brain-in-a-vat experiments whereby real outcomes are replaced by subjective impressions of them? These issues cannot be answered within a wellbeing maximization calculus, but must be decided at a more fundamental level.
Fifth, the constitutional approach provides a theoretically consistent alternative normative framework for guiding wellbeing research in the field of public policy. In particular, it acknowledges the central role that basic institutions play in forming public policy designed to raise individuals’ wellbeing. At the constitutional level, behind the veil of uncertainty, the fundamental rules and institutions are set that determine the decisions taken in the current politico-economic process. The legitimacy of political action finally rests on voluntary agreement on these fundamental rules by the citizens involved. In particular, individuals’ sovereignty includes the choice of how to best pursue their wellbeing. This holds for both the private and the collective realm. Individual sovereignty must not be reduced to a citizen's obligation to report his or her subjective wellbeing, which is then aggregated into a wellbeing indicator supposed to be maximized by government.
The political process should, rather, be institutionally structured so that people's interests become the principal controlling force in politics. Fundamental institutions, or rules of the game, have to be established that provide politicians and public bureaucrats with the motivation and information to adequately respond to people's preferences. Wellbeing research provides insights into how and to what extent institutions have systematic effects on indicators of individual subjective wellbeing. The focus is on rules and institutions, including written constitutional rules, state laws, social norms, traditions and even self-binding mechanisms.
Sixth, wellbeing research also helps to improve policy decisions within the given rules of the game. The improved measurement of subjective wellbeing strengthens political competition by allowing decision-makers to better evaluate the benefits provided by public goods and to compare various measures assessing the state of society.
In sum, the results gained from wellbeing research should be taken as inputs into the political process. These inputs have to prove themselves in political competition and in the discourse among citizens, and between citizens and politicians.
Acknowledgments
I am grateful to Patrick Balles, Bruno Frey, Kerstin Hansen, Reto Odermatt, Michaela Slotwinski and Severin Zimmermann for helpful remarks.
