Economic applications of happiness data

Happiness and valuation of non-market goods

Another purely economic application of happiness data is its use as the means to value non-market goods. Reference Deaton and StoneDeaton and Stone (2013) note that different subjective well-being (SWB) measures have been used to capture various difficult-to-measure phenomena, for example, the trade-off between unemployment and inflation, the costs of air pollution or the value of environmental amenities.

The equivalent income approach is a method to assign monetary value to non-market dimensions of life. As explained by Reference Decancq, Fleurbaey, Schokkaert, Atkinson and BourguignonDecancq et al. (2015b), ‘the equivalent income is the level of income that would make the individual indifferent … between his current situation and the hypothetical reference situation where he would be at the reference values for all non-income dimensions of life’ (p. 94). The method has been used for valuation studies in health economics to value illnesses (Reference Ferrer-i-Carbonell and Van PraagFerrer-i-Carbonell and Van Praag, 2002; Reference Groot, Van den Brink and PlugGroot et al., 2004; Reference HowleyHowley, 2017), hours of provided informal care (Reference Van den Berg and Ferrer-i-CarbonellVan den Berg and Ferrer-i-Carbonell, 2007) and the death of a relative (Reference Oswald and PowdthaveeOswald and Powdthavee, 2008). In ecological economics, the equivalent income approach has been used to assess the value of climate (Reference Brereton, Clinch and FerreiraBrereton et al., 2008; Reference Frijters and Van PraagFrijters and Van Praag, 1998), airport noise (Reference Van Praag and BaarsmaVan Praag and Baarsma, 2005), air quality or pollution (Reference LevinsonLevinson, 2009; Reference LuechingerLuechinger, 2009; Reference WelschWelsch, 2006) and flood disasters (Reference Luechinger and RaschkyLuechinger and Raschky, 2009). Finally, the technique has also been used to value other non-income goods, such as family size (Reference Plug and Van PraagPlug and Van Praag, 1995), marriage (Reference Clark and OswaldClark and Oswald, 2002), social relationships (Reference PowdthaveePowdthavee, 2008) and terrorism (Reference Frey, Luechinger and StutzerFrey et al., 2009).

Below we describe in detail how the equivalent income approach can be used to value non-market goods. Following the example proposed by Reference Ferrer-i-CarbonellFerrer-i-Carbonell (2013: 52), suppose we want to estimate an individual’s satisfaction S _i as

$S_i=\alpha \log \left(y_i\right)+\beta h_i+{\sum }_k{y}_k{z}_k+\epsilon$

where y_i is the individual income, h_i is the variable we would like to value (e.g. individual health status measured by the number of chronic illnesses), γkzk are the control variables (i.e. socio-demographic characteristics: sex, age, marital status, etc.) and ε is the error term of standard properties. The relation between estimated α and β can be used to derive the monetary value of a marginal change of health. It is equal to the change in income that, in terms of satisfaction/utility, would be equivalent to a change in the initial health status.

The equivalent income method has two distinctive features relevant for economic definition of utility. One, the happiness equation includes income in logarithm terms, allowing for the decreasing marginal utility of income. In consequence, the monetary value of a non-market good will actually depend on the current level of income: richer individuals will require larger income compensation. Two, as observed by Reference Ferrer-i-CarbonellFerrer-i-Carbonell (2013),

The case of commuting time has been analysed by Reference Stutzer and FreyStutzer and Frey (2008), who note that the related time loss will only affect happiness if the housing and labour markets do not entirely compensate for commuting time. Therefore, the monetary value of time estimated with the equivalent income method includes only the costs that are not already incorporated within the house prices and wages.

Reference Van Praag and BaarsmaVan Praag and Baarsma (2005: 224) apply a mixed approach to valuation studies. They value noise damage caused by aircraft around Amsterdam Airport as ‘the sum of hedonic house price differentials and a residual cost component’, whereas ‘[t]he residual costs are assessed from a survey, including an ordinal life satisfaction scale, on which individual respondents have scored’. Their estimates show that a monthly net household income of 1500 EUR would have to be increased by roughly 34 EUR to compensate for a noise increase from 20 to 30 Ku, while an increase in noise from 20 to 40 Ku would require an income compensation of approximately 57 EUR.^{Footnote 3}

Reference Decancq, Fleurbaey and SchokkaertDecancq et al. (2015a) introduce an improved form of the equivalent income method in which they control for unobserved characteristics (individual fixed effects) influencing life satisfaction and, therefore, the estimated values. Additionally, they allow for preference variation among individuals releasing the assumption of identical preferences, represented by perfect substitutability between components of satisfaction/utility. In their work, they adopt a basic linear happiness equation with individual fixed effects α_i

$S_{it}={\alpha }_i+{\mu }_t+{\gamma }_1\prime {\ell }_{it}+{\gamma }_2\prime Z_{it}+d_{it}$

where S _it is the life satisfaction expressed by individual i in period t, ℓ_it is a vector of variables capturing the life conditions of individual i (e.g. income, quality of housing and health), Z_it is a vector of personal characteristics (e.g. gender, age and marital status) which act as scaling variables, d _it is a disturbance term and (γ1,γ2) is a vector of coefficients to be estimated. In order to model the differences and changes in preferences, Reference Decancq, Fleurbaey and SchokkaertDecancq et al. (2015a: 1091) introduce interactions of the conditioning variables Z_it with income y _it and other life dimensions q_it

$S_{it}={\alpha }_i+{\mu }_t+\left(\beta +\Gamma Z_{it}\right)\ln \left(y_{it}\right)+\left(\upsilon +\wedge Z_{it}\right)\prime q_{it}+\delta \prime Z_{it}+d_{it}$

where (β, υ,δ) is a vector of direct effects, and Λ and Γ are matrices with interaction effects to be estimated. As pointed out by the authors, the vector of marginal rates of substitution (MRS) between income and the non-income dimensions of satisfaction will be individual and time-dependent

$MRSy{q}_{it}=y_{it}\left(\upsilon +\wedge Z_{it}\right)/\left(\beta +\Gamma Z_{it}\right)$

An important remark made in Reference Decancq, Fleurbaey and SchokkaertDecancq et al. (2015a) is that the preference variation introduced in their model remains limited, because it still requires the assumption that individuals characterised by the same values of the conditioning variables Z _it have identical preferences.

Happiness as a determinant of economic outcomes

Happiness, wages and labour market outcomes

The positive impact of happiness on productivity explains why some other key economic outcomes may be determined by individual well-being. According to the neoclassical approach, the value of the marginal product of labour is equal to the wage at the profit-maximising level of employment (see Reference Krugman and WellsKrugman and Wells, 2013: 536). That is to say, a worker’s productivity will strictly determine his or her earnings. In consequence, happiness, as a determinant of productivity, becomes in a natural way also a determinant of wage.

Several longitudinal studies have shown that SWB should be considered a predictor of income. Reference Marks and FlemingMarks and Fleming (1999) analysed the Australian Youth in Transition panel data, showing that, for several different cohorts, well-being increases in the preceding periods were related to future increases in earnings and to lower chances of being unemployed. In another study, Reference Diener, Nickerson and LucasDiener et al. (2002) surveyed more than 7000 students to assess the influence of dispositional affect (defined as self-rated cheerfulness at college entry) on their future economic outcomes. Nineteen years after the survey, the same participants reported their incomes: cheerful students were earning more money in their 30s compared to their less cheerful counterparts. Additionally, they also had higher job satisfaction ratings and were less likely ever to have been unemployed. These results emphasise that happier people are expected to acquire greater wealth across their lifetimes.

In a more recent contribution, Reference De Neve and OswaldDe Neve and Oswald (2012), using data from a large US representative panel study (11,000 individuals), demonstrate that adolescents and young adults who report higher life satisfaction grow up to earn significantly higher levels of income later in life. Importantly, having introduced family fixed characteristics, the authors showed that the result holds after controlling for sibling effects. Additionally, Reference De Neve and OswaldDe Neve and Oswald (2012) identify the decisive factors shaping the impact of psychological well-being on earnings. The mediators which carry the influence from happiness to income are higher probability of obtaining a college degree and getting hired or promoted, having greater degrees of optimism and extraversion and experiencing lower levels of neuroticism.

Reference Graham, Eggers and SukhtankarGraham et al. (2004), using two waves of the Russia Longitudinal Monitoring Survey (more than 5000 observations), found that people who had higher ‘residual happiness’ in 1995 had higher earnings and were in better health in a survey 5 years later. The term ‘residual happiness’ denotes happiness levels that are not explained by the usual determinants of well-being, that is, income, education and socio-demographic characteristics. The obtained results show, therefore, that happiness acts as a predictor of earnings in future periods independently of a current level of earnings. Reference Graham, Eggers and SukhtankarGraham et al. (2004) emphasise that their analysis ‘supports the evidence from the psychology literature that happier people earn more income or, more broadly speaking, perform better economically’ (p. 336). The authors conclude that ‘the same positive cognitive biases such as self-esteem, control, and optimism that affect normal happiness levels may also have positive effects on people’s performance in the labour market’.

Numerous psychological studies highlighted the role of emotional well-being in explaining success in employment. Reference Burger and CaldwellBurger and Caldwell (2000) demonstrate that even before finding a job, a happier person is more likely to be invited to a job interview. A study of college students showed that the higher their positive affect prior to graduation, the greater their probability of receiving a job interview 3 months later. In this manner, positive emotions are a distinct advantage when searching for a job. In addition, compared to less happy peers, happy individuals are less likely to lose their jobs (Reference Diener, Nickerson and LucasDiener et al., 2002) and more likely to find a new job when unemployed (Reference Marks and FlemingMarks and Fleming, 1999).

In an interesting contribution to the happiness–employment debate, Reference Lucas, Clark and GeorgellisLucas et al. (2004) studied 24,000 German citizens over a 15-year time span and showed that a significant drop in life satisfaction can be observed as early as 2 years prior to the job loss. Reference CornelißenCornelißen (2009) identifies job satisfaction to be an important determinant of the self-reported probability of job search, which, in turn, predicts actual job changes. These results highlight the predictive role of SWB in understanding an individual’s labour market choices and outcomes: unhappiness may cause lower performance at work and, therefore, it may lead to a job loss; unhappiness, caused by job dissatisfaction, may force a job change decision.

Reference KrauseKrause (2013: 1) in another work supporting the influence of psychological well-being on economic performance demonstrates that happiness plays an important role in finding reemployment. The author examines the job search results of 2500 unemployed Germans and finds a ‘significant inverted U-shaped effect of residual happiness on an unemployed individual’s future reemployment probability and re-entry wage, even after controlling for demographic and socio-economic characteristics, labour market histories and future job prospects’. An additional finding was that happiness is a strong predictor for an individual to move from unemployment into self-employment rather than regular employment. Reference Gielen and Van OursGielen and Van Ours (2014) using data from the German Socio-Economic Panel (SOEP, 1994–2007) found that an unemployed individual who experienced a sharp drop in happiness after a job loss searched more actively for a new job; however, it did not speed up his or her job finding. This conclusion contradicts with the results of two earlier works using data from the British Household Panel Survey (BHPS): Reference ClarkClark (2003) and Reference MavridisMavridis (2010) indicate that people who experience greater drops in well-being upon becoming unemployed have a lower probability of remaining unemployed 1 year later. Reference MavridisMavridis (2015) explains that these findings ‘confirm the theoretical prediction from job search theory: search effort and unemployment duration are affected by the utility differential between having a job and being unemployed’ (p. 1).

In the most recent contribution on the influence of happiness on labour market performance, Reference O’ConnorO’Connor (2017) uses data from SOEP (1996–2013) to show that happier people have a lower probability of being unemployed. This relation is most likely to be causal as it holds in an instrumental variable setting and is confirmed within the residual happiness framework. The author shows, additionally, that residual SWB is comparable to cognitive ability in explaining wages.

Economics of personality traits

In order to fully understand the relation between happiness and economic outcomes, it is necessary to assess the role of individual differences in personality. Reference BoyceBoyce (2010) demonstrates that when explaining SWB, personality traits account for roughly 20% of the unobserved heterogeneity across individuals. In psychology, and recently in economics, personality is assessed through the Big-Five traits defined as follows (Reference Becker, Deckers and DohmenBecker et al., 2012):

1. 1. Openness. Individual differences in the tendency to be open to new aesthetic, cultural and intellectual experiences;

2. 2. Conscientiousness: The tendency to be organised, responsible and hardworking;

3. 3. Extroversion: An orientation of one’s interests and energies towards the outer world of people and things rather than the inner world of subjective experience;

4. 4. Agreeableness. The tendency to act in a cooperative, unselfish manner;

5. 5. Neuroticism. A chronic level of emotional instability and proneness to psychological distress.

Several earlier contributions have demonstrated the impact of personality traits on economic outcomes. Reference Mueller and PlugMueller and Plug (2006) find that openness and conscientiousness are likely to be rewarded in the labour market, while agreeableness and neuroticism are likely to be penalised with lower wages. Reference Nyhus and PonsNyhus and Pons (2005) arrive at similar conclusions and show that earnings may be additionally determined by an individual’s degree of control. Reference GrovesGroves (2005) explains that psychological traits, such as autonomy, social withdrawal and aggression, play an important role in determining female earnings. Reference Borghans, Duckworth and HeckmanBorghans et al. (2008) note that personality assessments offer a considerable benefit to designing motivation schemes for employees. They argue that economic incentives may influence individuals differently according to the differences in personality. Reference Uysal and PohlmeierUysal and Pohlmeier (2011) claim that psychological attitudes, such as propensity to motivate and control oneself, influence job search intensity. Using SOEP data (1984–2007), the authors find evidence that the instantaneous likelihood of finding a job is positively associated with conscientiousness and negatively associated with neuroticism.

The analysis of personality traits helps understand the relation between SWB and changes in income. Reference Boyce, Wood and BrownBoyce et al. (2010) analyse the role of conscientiousness in explaining the well-being decrease following a negative income shock due to job loss. In a longitudinal study of 9000 respondents, the authors document that after 3 years of unemployment individuals with high conscientiousness (i.e. one standard deviation above the mean) experience a 120% higher decrease in life satisfaction than those with low levels of conscientiousness. These findings emphasise that conscientiousness, which itself has a positive impact on well-being, could turn out detrimental when failure is experienced. Reference Boyce, Wood and FergusonBoyce et al. (2016) examine how personality traits relate to aversion to income losses. Their results show that individuals high in conscientiousness have the strongest reactions to income losses, suggesting an elevated loss aversion effect, whereas no such effect is observed for those moderately unconscientious. Reference Boyce and WoodBoyce and Wood (2011) give an example of how the study of personality traits can provide new insights into standard microeconomic analysis. Using data for more than 13,000 German adults over 4 years, they demonstrate that marginal utility of money changes with personality: individuals with higher levels of conscientiousness obtain more satisfaction in their lives from increases to their household income. Reference Soto and LuhmannSoto and Luhmann (2013) perform a similar analysis using data from three different large-sample longitudinal studies (BHPS, SOEP and Household, Income and Labour Dynamics in Australia (HILDA)). They show that neuroticism consistently moderates the life satisfaction effects of between-person income differences and within-person income fluctuations. In particular, highly neurotic individuals are characterised by greater vulnerability to unfavourable income comparisons (with their own past and with others) and by stronger reactions to negative changes in individual income.

Reference Proto and RustichiniProto and Rustichini (2015) introduce a quadratic specification in the model linking individual income to life satisfaction. Using data from SOEP and BHPS, they find that neuroticism increases the usually observed concavity of the happiness–income relationship. More neurotic individuals enjoy extra income more if they are poorer and enjoy extra income less if they are richer. Reference Friehe, Mechtel and PannenbergFriehe et al. (2014) employ data from the pretest modules of SOEP (2008–2010) to determine which personality traits relate to stronger positional income concerns. The authors provide evidence of a statistically significant link between personality and the self-reported importance of income comparisons (comparison intensity). They estimate a negative association of comparison intensity with agreeableness, which means that individuals characterised by the ‘tendency to act in a cooperative, unselfish manner’ are considerably less concerned about positional competition with others. Reference PiekalkiewiczPiekalkiewicz (2016) obtains similar results, showing that the well-being of trustful and sociable individuals is less affected by income comparisons with their peers.

Importantly for economists and policy makers, psychological traits influencing SWB and economic outcomes are not stable across lifetime and may vary depending on external circumstances. Reference Heckman and KautzHeckman and Kautz (2013) review the literature on cognitive and non-cognitive skills, and propose to consider personality as a variable component of human capital in equations explaining wage and job search outcomes. According to the authors, an individual’s personality should be considered as a set of ‘character skills’, defined as ‘personality traits, goals, motivations, and preferences that are valued in the labour market, in school, and in many other domains’. Reference Heckman and KautzHeckman and Kautz (2013) explain that

The authors claim that character skills can be fostered and describe in detail numerous examples of interventions from different parts of the world: early-life interventions that begin before formal schooling, education in kindergarten and elementary school, interventions targeted towards adolescents and young adults, apprenticeship programmes for workers and interventions applied to multiple age groups.

Reference Boyce, Wood and PowdthaveeBoyce et al. (2013: 302) provide empirical evidence on the variability of individual psychological traits and examine changes in personality of 8600 individuals surveyed in two waves of the HILDA study. Their results show that ‘personality varies at least as much as socio-economic factors that are typically considered as variable, such as income, unemployment and marital status’, and that personality variations are a strong predictor of changes in life satisfaction. Reference Boyce, Wood and PowdthaveeBoyce et al. (2013) emphasise that these findings may help inform policy debate over how best to help individuals and nations improve their well-being. Reference Boyce, Wood and DalyBoyce et al. (2015) identify which life events have the strongest impact on the Big-Five personality traits. They investigate data for 6700 Germans using a latent change model and find significant personality changes following a job loss. These results are consistent with the view that personality may be shaped by contextual and environmental factors (Reference Heckman and KautzHeckman and Kautz, 2013), suggesting that ‘public policy can play a key role in enabling psychological growth’ (Reference Boyce, Wood and DalyBoyce et al., 2015: 1007).