Progress in Applications to Policy
Direct applications of subjective well-being to policy are at still an early stage. The years since 2009 have been immensely productive for the implementation of subjective well-being data collection in many countries, for understanding the issues in using subjective well-being data, and for the development of many techniques to cope with these issues. New research is now needed to better understand how subjective well-being measures can be transformed into a useful metric for policy-makers, and in what way they can provide meaningful information that contributes to better policy decisions. Yet another commission, from the Legatum Institute, focused on subjective well-being and governmental policy and provided a refreshingly pragmatic and thoughtful approach (O’Donnell et al., 2014). In many policy applications, some benefits and costs are recognized, but are not easily or accurately quantifiable in monetary terms, because explicit markets for these benefits and costs do not exist, and implicit valuations can only be imperfectly, if at all, placed on these factors. Thus, even if subjective well-being measures are imperfect, they have the potential to advance policy-making when compared with the imperfect measures of benefits and costs often available.
Cost-Benefit Analysis
Some work has used subjective well-being measures in cost-benefit analysis, as a method for valuing nonmarket outcomes (O’Donnell et al., 2014; Fujiwara and Campbell, 2011). The principle underlying this work is that many policies have costs and benefits that are difficult to monetize. As a result, standard cost-benefit analysis, which compares monetary costs and benefits, will lead to policy decisions that underweight those nonmonetary costs and benefits if these are not taken into account. In the case of valuing nonmarket factors for cost-benefit analysis, current methods have serious limitations, so complementing existing methods with subjective well-being-based valuations could provide additional information. The UK Treasury’s Green Book, which provides formal guidance to government agencies on the appraisal and evaluation of policy proposals, was updated in 2011 to include a section on valuation for social cost-benefit analysis, including through subjective well-being-based methods. However, additional work needs to be done to make this approach more credible. In particular, there are many problems with monetizing differences in levels of subjective well-being (Kahneman and Krueger, 2006)—not least a lack of data sets containing high-quality measures of both personal income and subjective well-being, as well as conceptual problems in identifying a unit of measure with which to convert subjective well-being into dollars or euros. As with other measurement issues, the difficulties in using subjective well-being for cost-benefit analysis must be seen in light of the difficulties in other methods for cost-benefit analysis—and subjective well-being-based methods should be seen as a complement to, rather than a replacement for, more traditional techniques.
Program and Policy Evaluation
Several policy and program evaluations have included subjective well-being as an outcome indicator, and its inclusion can help both to assess the impact of a program and to understand its mechanisms with more confidence. These studies not only support the idea that subjective well-being can be used to meaningfully measure policy impact; they also support the underlying construct of subjective well-being and its responsiveness to life circumstances. Another advantage of using subjective well-being measures in policy evaluations is that they may show that interventions have benefits that are not measured by conventional outcomes; or, conversely, to show that while conventional methods may show benefits, these benefits could be offset by lower subjective well-being (and, in both cases, providing a richer understanding of program impact).
For example, an in-work support program in the United Kingdom was found to have unexpected negative impacts on the subjective well-being of people who participated in these programs (Dorsett and Oswald, 2014). Similarly, an unconditional cash transfer program in Kenya found positive impacts on the subjective well-being of participants, but negative spill-over effects on nonparticipants (Haushofer, Reisinger, and Shapiro, 2015). In Morocco, while household connections to the municipal water supply showed no impact on health or income, they resulted in increased happiness (Devoto et al., 2012). Other studies suggest a positive impact of access to insurance on mental health (Finkelstein et al., 2012), of family leave policies on the life satisfaction of parents (D’Addio et al., 2014), and of participation in the National Citizen Service in the United Kingdom on the subjective well-being of participants (United Kingdom Cabinet Office and Ipsus MORI, 2013). Finally, Ludwig et al. (2013) found that subjective measures of mental health improved among participants in the Moving to Opportunity experiment in the United States before such results showed up in physical health. Measures of subjective well-being also have the potential to improve our understanding of people’s economic insecurity (see sidebar, “Subjective Well-Being and Economic Insecurity”), as also argued in Chapter 8 of this volume.
While work on refining the measurement and understanding of subjective well-being should continue, experimentation in policy applications should commence. As with many domains, experiments in policy applications and foundational work on measurement and understanding are likely to complement one another, in a mutually reinforcing process.
SUBJECTIVE WELL-BEING AND ECONOMIC INSECURITY
There are several possible threads through which one can link subjective well-being and economic insecurity. Research on both topics confronts several shared methodological issues, especially with respect to subjective economic insecurity. In addition, a primary concern about economic insecurity is that it reduces subjective well-being (even if the bad event does not actually happen). Eurostat’s “Analytical Report on Subjective Well-Being” (2016) showed, for example, that people’s inability to face unexpected expenses drastically reduces their subjective well-being, even when controlling for the impact of other variables such as their income or employment status.
A recent analysis of economic trends highlighted the differential effects of positive and negative GDP growth on people’s subjective well-being (De Neve et al., 2015), using data from the Gallup World Poll, the US Behavioral Risk Factor Surveillance System, and Eurobarometer. Motivated by inferences from prospect theory (Kahneman and Tversky, 1979), the authors found that average evaluative well-being (life satisfaction) reacted more strongly to negative GDP growth than to positive growth in all three surveys (a pattern that is consistent with loss aversion). On the other hand, experiential well-being was only impacted by negative GDP growth: daily happiness and enjoyment decreased and stress and worry increased during periods of economic declines.
These findings have implications for both economic theory and macro-economic policies, including the impact of unemployment, and nicely demonstrate the utility of taking a multi-faceted view of subjective well-being.
Continuing Issues and New Questions
Issues to Be Addressed to Gain a More Complete Understanding of Subjective Well-Being
Causality
First, as with all other types of analysis, careful attention must be paid to establish credible causality. It is difficult to reach strong conclusions about causality on the basis of much of the subjective well-being research that is currently available, which relies mainly on observational and self-reported data. In order for the field to advance, it will be important to focus less on exploratory and hypothesis-generating studies, and more on developing and testing theories. These theories, and the research designed to test them, should ideally take into account the complex interrelationships among subjective well-being correlates, in order to identify which variables are acting as mediators or moderators, and which are causing actual shifts in subjective well-being. Panel data, especially panel data that can take advantage of a discontinuity such as a policy shift, are likely to be beneficial in this respect. As mentioned earlier, this endeavor is facilitated, and indeed perhaps only possible, when explicit models and theories are used to inform analyses and interpretations.
/>
Understanding the nature of sorting with respect to preferences is also a priority for future research. For example, Krueger and Schkade (2008) provide some evidence that workers sort themselves across jobs based on their preferences, with workers who are more extroverted tending to be employed in jobs that require more social interactions.
Heterogeneity
Second, and related, analyses of subjective well-being need to go beyond the average and examine heterogeneity. Focusing on the average subjective well-being of a group or nation is misleading in the same way as focusing on the average income of a group or nation. There may be large inequalities in subjective well-being, including inequalities among people with different demographic characteristics. It is also important to consider the possibility of different levels and correlations among different groups in order to more fully understand the dynamics of subjective well-being. However, any sub-group analysis requires large sample sizes, which are not always available, although there are certainly some large-scale data sets in development, and data from NSOs will have a particularly important role to play here. In addition, data quality on the dimensions of interest for disaggregation is often poor among existing data sets, in particular income. This is a substantial problem for research on the relationship between income and subjective well-being. Again, large official data sets such as EU-SILC allow this disaggregation and analysis, with high-quality information on a range of covariates collected in a standardized way across countries. Most of these indicators and analyses are published—in the data sets and Eurostat publications—with a variety of different breakdowns such as those for sex, age, income, education level, employment status, country (for most countries also region) of residence, and degree of urbanization.
A theoretical model detailing the factors and processes underlying observed differences in subjective well-being would be invaluable for understanding and for designing possible interventions to remedy these inequalities. Building up a long time series of data across a wide range of countries will also be essential for testing these models, and to assess the dynamic relationships among drivers and outcomes.
Similarly, relatively little is known about the tails of the distribution of subjective well-being, particularly those who rate themselves as having extremely low levels of subjective well-being (including those who live with high levels of pain). These people are likely to represent a particular policy concern, and much may be learned from research focusing on them.
Finally, research efforts should continue to focus on adaptation and resilience, as these are among the most promising, if difficult, research avenues. Further investigation of the role of public goods and services in supporting people’s resilience might lead to findings that could be directly acted upon by policy-makers; in other situations, such research might highlight the resources that are needed to restore subjective well-being after a life challenge.
Data Collection and Analytic Issues
Data Collection and Availability
As described above, data collection on subjective well-being has expanded enormously. There are, however, two important priority areas where there is still a lack of data on subjective well-being, and where the inclusion of subjective well-being questions is likely be relatively low cost. The first priority is the expansion of high-quality data collection on subjective well-being to poor countries, for example, by including a life satisfaction and experiential well-being module in household surveys conducted in these countries. As well as casting important light on the societal conditions and policy environments that can influence changes in subjective well-being over time, research into differences in well-being among countries has potential for addressing persistent methodological and conceptual questions concerning the meaning of subjective well-being responses from people in different cultural and economic settings. Have people in poor countries adapted to their circumstances, and are they using the subjective well-being scales in entirely different ways than those in wealthy countries? And if so, what do country comparisons mean?
The second priority is the inclusion of subjective well-being measures in official time-use surveys in order to increase our understanding of experiential well-being, as recommended in the National Academy of Science (NAS) report mentioned above. Such efforts could be supported by more guidance on which approach is best for this purpose, and by increasing the research output linked to existing efforts. The Guidelines on Harmonising Time Use Surveys (UNECE, 2013) have been useful in this area, but additional analysis of different approaches to collecting experiential well-being data are necessary in order to provide empirical guidance on best practice in this area. As mentioned above, the inclusion of subjective well-being items in ongoing time use surveys, such as the American Time Use Survey, is an efficient way to achieve this goal. Collecting time-use data and experiential well-being in poor countries would also be particularly useful, as little is known about the daily activities or experiential well-being of rural populations.
Finally, timely access to these data is critical, as is responsiveness by researchers to the new data. Measurement initiatives will push forward our knowledge and understanding of subjective well-being, but researchers must have access to these data in order to achieve this outcome. In turn, researchers must demonstrate the usefulness of these measures, or the measures risk being dropped. Increased cooperation among various actors would improve the quality and usage of data on subjective well-being, and networks can play an important role in this respect.
Data Analysis and Interpretation
As discussed above, one of the most important issues inadequately addressed by current research is that of systematic differences in question interpretation and response styles between population groups. Is there conclusive evidence that this is a problem? And, if so, are there ways to adjust for it? Information is needed about which types of group comparisons are affected, about the magnitude of the problem, and about the psychological mechanisms underlying these systematic differences.
While some of the methodological issues that have been associated with subjective well-being measures are, to varying degrees, minimized through the use of standardized questionnaires, this issue is not resolved through standardization alone. So far, only limited analysis has been performed for the assessment of the 2013 EU-SILC ad hoc module regarding the viability of cross-national comparisons (Eurostat, 2016). The analysis that exists has been mostly undertaken through internal and external validation (correlations with certain variables of interest), and the results are encouraging.
Some progress has also been made on using vignettes to address this issue, but this approach has limitations. For example, people’s responses to the circumstances described in vignettes are very likely to be shaped by the policy environment in which they live—because the implications of, for example, living with a chronic health condition depend on factors such as health care cost and availability, as well as disability benefits. This means that “correcting” subjective well-being self-reports of own conditions according to international differences in how people rate the same vignettes could ultimately remove the most policy-relevant part of the international variation in well-being, effectively throwing the baby out with the bathwater.
The U-index, which calculates the share of time that individuals spend in an unpleasant state—defined as a period when the strongest reported emotion is a negative one (Kahneman and Krueger, 2006)—is a promising method for neutralizing differences in the extent to which response scales are used differentially across countries or groups of people. While some work is available supporting its utility, its application so far has been limited to measures of experiential well-being. More work is needed to see actually how large is the problem of inter-personal differences in the use of response scale, and to find new cost-effective solutions.
Subjective well-being questions are generic, but there are times when they should be tailored to the application at hand. For example, in the Moving to Opportunity project
in the United States, feelings of security and anxiety were especially relevant for program participants and were specifically targeted. In many medical studies, pain may be a particularly relevant type of experiential well-being, whereas misery may be especially relevant in studies of refugees.
More population-level work is needed on subjective well-being beyond life satisfaction, i.e., extending measurements to eudaemonic and experiential well-being. At present, a variety of different approaches are being used across OECD statistical offices to collect this type of information (see Exton, Siegerink, and Smith, forthcoming, for further details). As described in the sidebar above, measures of subjective well-being belong to three different categories. Life satisfaction is generally the most widely used measure of subjective well-being, and the one for which there has been the most research, for example, on adaptation. A focus on a single indicator is, on the one hand, beneficial and pragmatic, as it ensures a wider evidence base for at least one indicator. On the other hand, there are surely missed opportunities, as there is likely much to be learned from data on the other dimensions of subjective well-being.
We concur with the conclusion of the NAS report (Stone and Mackie, 2015) that research on subjective well-being should be explicit about the types of measure used, and should ideally include more than one type of measure. More effort should be given to understanding eudaemonia and experiential well-being, to describing the relationship between the different measures, and to ensuring that these outcomes are not being neglected.
For Good Measure Page 30