Income and happiness across Europe: Do reference values matter?

Journal of Economic Psychology 30 (2009) 42–51

Contents lists available at ScienceDirect

Journal of Economic Psychology journal homepage: www.elsevier.com/locate/joep

Income and happiness across Europe: Do reference values matter? Guglielmo Maria Caporale a,*, Yannis Georgellis a, Nicholas Tsitsianis b, Ya Ping Yin b a b

Department of Economics and Finance, Brunel University, Uxbridge, Middlesex UB8 3PH, United Kingdom Department of Accounting, Finance and Economics, University of Hertfordshire, Hatfield AL10 9AB, United Kingdom

a r t i c l e

i n f o

Article history: Received 15 May 2007 Received in revised form 10 April 2008 Accepted 1 June 2008 Available online 15 September 2008

JEL classification: I31 PsycINFO classification: 3100

a b s t r a c t Using cross-sectional data from the first two rounds of the European Social Survey (ESS), we examine the relationship between income, relative income and happiness across 19 European countries. We find that a positive and statistically significant relationship between income and happiness does exist, but such a relationship is weakened by reference income. We also find that while reference income exerts a negative impact on happiness in the case of Western European countries, its effect is positive in the case of the Eastern European countries, a finding that is consistent with the ‘tunnel effect’ hypothesis. This suggests that for Eastern Europeans reference income is likely to be a source of information for forming expectations about their future economic prospects, rather than a yardstick measure for social comparisons. Ó 2008 Elsevier B.V. All rights reserved.

Keywords: Comparison income Reference groups Happiness Life satisfaction

1. Introduction The relationship between income and happiness continues to attract attention among social scientists, including economists, who, rather than using the more holistic approach adopted in other social sciences, tend to model utility as a monotonically increasing function of, and often synonymous to, absolute income. However, in recent years, such a conjecture has been subjected to the scrutiny of a substantial volume of empirical evidence, with mixed results. For instance, while cross-sectional studies tend to support the hypothesis of a causal relationship between income and happiness (a proxy for utility), there is a great variation in the strength of such a relationship across income groups. Generally, the correlation between income and happiness is weaker in cross-sectional studies that use national averages of income and happiness scores or in studies that control for education, unemployment and other moderating factors that could mitigate the strength of such a relationship (see, e.g., Diener, Sandvik, Seidlitz, & Diener, 1993; Frey & Stutzer, 1999; 2002; Heady, 1991; Oswald, 1997). In contrast, any positive correlation, as revealed in cross-sectional studies, is largely absent in time series studies. These usually find that happiness, in developed countries at least, has reached a plateau despite a continuous growth in real income. In the same vein, there is little evidence of a life-cycle trend in happiness even though income and economic circumstances change.1 * Corresponding author. Tel.: +44 0 1895 266713. E-mail address: [email protected] (G.M. Caporale). 1 The lack of any significant correlation between changes in income and changes in happiness form the basis for the ‘Easterlin Paradox’ (Easterlin, 1995, 2001), which has been the motivating hypothesis behind a substantial volume of empirical work. 0167-4870/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.joep.2008.06.004

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

43

In this paper, we focus on the cross-sectional relationship between income and happiness using data for 19 European countries from the first two rounds of the European Social Survey (ESS). Previous studies are based mostly on data from a single country, those exploring this relationship in a comparative context being rather sparse. Senik (2008) is a notable exception, providing comparative evidence on the relationship between income, reference income and well-being using a variety of data sources for Western Europe, Eastern Europe and the United States, including data from the first round of the ESS survey. In the same spirit as Senik (2008), the aim of this paper is to explore the relationship between income, relative income and subjective well-being in a comparative European context. Using reported happiness and life satisfaction scores as measures of subjective well-being, we find that absolute income has indeed a strong positive effect on wellbeing.2 Marginal effects based on the estimated income coefficients in our regressions imply that an increase in income from the lowest to a middle income group increases life satisfaction scores by 0.71 points, which is comparable to an increase in life satisfaction scores caused by a marginal improvement in respondents’ health. However, we also find that the strength of the relationship between income and happiness is weakened by reference income. While reference income per se has a negative impact on life satisfaction, it also reduces the absolute income effects across all income bands. Performing separate analyses for Eastern European countries, we find evidence that the reference group’s income exerts a positive influence on individual life satisfaction, a finding which lends support to Hirschman’s (1973) ‘tunnel effect’ conjecture. This refers to the phenomenon that in uncertain and adverse situations people often interpret any positive signals that they can observe around them to predict an improvement in their own situation to occur sooner or later. Therefore, it seems that in Eastern European countries reference income does not influence individuals’ well-being through social comparisons, but rather through their informational content, which is used in order to form expectations about future income and economic situation.3 Our findings, based on an additional round of data from the ESS and using different measures of reference income, are consistent and supplement Senik’s (2008) findings, thus offering additional empirical support for the existence of an East– West divide across individuals’ perceptions of reference income as a source of social comparisons or of information about future economic prospects. The layout of the remainder of the paper is as follows. Section 2 reviews the existing empirical evidence on the relationship between happiness and income, paying particular attention to evidence based on cross-sectional studies. Section 3 describes the data and the empirical framework for the present study. Section 4 presents the main empirical findings. To check their robustness, Section 5 summarises the results of estimating life satisfaction equations using alternative specifications. Section 6 concludes. 2. Income and happiness – existing evidence and explanations In a seminal study, Easterlin (1974) provides early empirical evidence for the US showing that income growth does not lead to higher levels of happiness, a finding further supported by subsequent studies based on time series data.4 In contrast, cross-sectional studies tend to support the existence of a positive, albeit generally weak, relationship between income and happiness. At the household or individual level, a positive and statistically significant correlation is found, with estimated correlation coefficients between 0.12 in Diener et al. (1993) and about 0.2 in Easterlin (2001). The strength of the relationship between income and happiness becomes even more contentious once the sample of households is divided into income sub-groups. As Argyle (1999, p. 356) suggests, the positive relationship between happiness and income only holds for the lower end of the income distribution, whilst Glatzer (1991) finds no clear income effect on life satisfaction between the second and fifth income quintile in Germany. For cross-sections of average life satisfaction and average income across developed countries, it is found that the effect of income on happiness is weak across countries with an average annual income level above $10,000 (e.g. Diener & Suh, 1999; Frey & Stutzer, 2002). Against the background of such a controversy regarding the strength of the relationship between income and happiness, the study by Frijters, Haisken-DeNew, and Shields (2004) is particularly noteworthy in concluding that, after all, income does buy happiness, especially in Eastern European countries. A number of alternative explanations have been proposed for the apparently contradictory, or at least mixed, results in the empirical literature. The absolute income hypothesis states that the level of utility varies positively with the level of income up to a threshold level beyond which utility remains largely invariant. This characteristic of utility reflects the assumption that once a person’s basic material needs are satisfied, other aspects of life rather than further improvement in material well-being predominantly determine the person’s sense of happiness. This hypothesis is generally consistent with the 2 While many studies assume happiness and life satisfaction to be synonymous, there is considerable evidence suggesting that they are not strongly correlated (see Cummings, 1998). In general, life satisfaction refers to cognitive states of consciousness, whereas happiness is emotional and mainly concerns intimate matters of life. Indeed recent evidence (e.g., Gundelach & Kreiner, 2004) reinforces Michalos’s (1991) view that while happiness and satisfaction form part of a subjective well-being construct, it is heuristically useful to measure and analyse them separately. In this paper, we used both ‘‘Happiness” and ‘‘Life satisfaction” scores as measures of subjective well-being and as dependent variables in our estimations. The results are very similar using either measure and therefore we only report the ‘‘Life satisfaction” results. The ‘‘Happiness” regression results are available upon request. Throughout the paper we use the terms ‘‘Happiness” and ‘‘Life satisfaction” interchangeably. 3 Senik (2004b) is the first to test formally the ‘tunnel effect’ hypothesis using large-scale data. 4 For a critical evaluation of time series studies on income and happiness, see Hagerty and Veenhoven (2003). They also improve the statistical power of previous tests and find evidence of apositive link between national income and national happiness, with some evidence that such a link weakens in the longrun.

44

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

assumption of diminishing marginal utility of consumption (or income) that characterizes the neoclassical theory of utility. Rojas (2007) explains the weak relationship between income and happiness using the conceptual-referent theory of happiness (CRT). According to CRT, individuals have different notions about what a happy life is and, therefore, different evaluations of their subjective well-being. As Rojas (2007) argues, this heterogeneity in beliefs about a happy life extends to the relationship between income and happiness. A weak relationship between income and happiness may be explained partially by the fact that income might be less important for individuals with conceptual referents for happiness with an inner orientation, as opposed to an outer orientation. Recent empirical evidence shows that well-being depends on the discrepancy between actual and comparison income, questioning the assumption of utility as a monotonic function of income (for an overview, see Frey & Stutzer, 2002; Senik, 2004a). This finding leads into the realm of relative income theory, which has a distinguished lineage within economics (Veblen, 1899; Duesenberry, 1949). The relative income hypothesis states that relative, instead of, or in addition to absolute income, is what determines utility. Indeed, social norms, social comparisons, and reference values influence individuals’ subjective evaluation of their economic situation, weakening the relationship between income and happiness one could observe based only on absolute income. How individuals feel about their well-being depends on the distance between their actual income from a reference value. Such a reference or aspiration value is determined by the income enjoyed by people around them and/or the level of income that the individuals themselves have become accustomed to over time. When the economy grows, individual incomes and reference values all grow at similar rates so that the distance between the two remains relatively stable, and so do individuals’ perceptions of utility or happiness. A growing body of empirical evidence supports the relative income hypothesis.5 As Clark and Oswald (1996) show, using regression analysis and controlling for standard individual and demographic characteristics, utility depends on income relative to some reference or comparison income, based on the predicted income of ‘people like you’. Defining the reference group to include those with similar education, similar age and living in the same region, Ferrer-I-Carbonell (2005) finds that income of the reference group is as important as own income for individuals’ happiness. McBride (2001) uses all those in the same age group, within 5 years younger or older than the individual concerned, while Easterlin (1995) implicitly assumes that individuals compare themselves with all the other citizens of the same country. Exploring the link between income, relative income, and subjective well-being across Europe and the United States, Senik (2008) defines relative income as the income of one’s professional peers. In an earlier study, Van de Stadt, Kapteyn, and Van de Geer (1985) define the reference group according to education level, age, and employment status. Rizzo and Zeckhauser (2003) and Mas (2006) are notable examples of recent studies highlighting the importance of reference points as determinants of actual behavior. An alternative treatment of the concept of relative income focuses on individuals’ comparisons with their own income or economic situation in the past. As Easterlin (2001) argues, individuals adapt to their economic circumstances so that changes in income have only transitory effects on well-being. This is consistent with a large body of research in psychology providing evidence of adaptation, following Brickman and Campbell’s (1971) ‘hedonic treadmill’ hypothesis. Although Van Praag (1971) and VanPraag and Kapteyn (1973) were the first economists to explore this hypothesis, or, as they called it, the ‘preference drift’ phenomenon, the notion of adaptation was not embraced with the same enthusiasm in the economics literature. Nevertheless, there is an increasing consensus that understanding the process of adaptation and changing aspirations is important for our understanding of economic behaviour (see Kahneman & Krueger, 2006). Recent evidence by Stutzer (2004) shows that higher income aspirations, influenced by both individuals’ past income and the average income in their community, reduce utility. Interestingly, Easterlin (2005) also finds that aspirations about economic wealth and other pecuniary aspects of one’s well-being tend to change with the level of actual circumstances, suggesting almost complete adaptation.6 In summary, empirical evidence based on cross-sectional data has generally revealed a positive but weak relationship between happiness and income. However, whether or not income effects differ significantly across income groups, how such effects are mitigated by the inclusion of reference income and how absolute and relative income effects vary across different parts of Europe, remain largely contentious issues.7 Against this background, the ESS is an ideal database for the cross-sectional analysis of the relationship between income and happiness in a comparative context. 3. Data and empirical framework Our empirical analysis is based on cross-sectional data for 19 European countries from the first two rounds (2002 and 2004) of the European Social Survey (ESS). The ESS is designed to document the changing social attitudes, beliefs, and behaviour patterns across Europe. It is funded by the European Commission, the European Science Foundation and

5 The publication of Kahneman and Tversky’s (1979) article on prospect theory was the catalyst for such a growth in empirical work and for the wider acceptance of the relative utility hypothesis among economists. 6 In contrast, Easterlin (2005) finds that this is not the case with marriage, number of children and other non-pecuniary aspects of one’s life. Clark, Diener, Georgellis, and Lucas (2008b) investigate patterns of adaptation for a number of life and economic events using large scale panel data. They find evidence of a rapid and complete adaptation for most life events, while this is not the case for unemployment. 7 For a comprehensive and insightful review of the main issues in the debate about the relationship between income and happiness see Clark, Frijters, and Shields (2008a).

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51 9

LIFE SATISFACTION 2002

45

LIFE SATISFACTION 2004

8

7

6

5

4

3

2

1

0 AUT BEL

CH

CZE DEU DNK ESP

FIN

GB

GR

HUN

IRL

LU

NTL NOR POL PRT SVE SLO

Fig. 1. Income and life satisfaction across Europe.

scientific funding bodies in each of the participating countries. Individual countries participate on the basis of strict guidelines and criteria determined by a central co-ordinating committee. Data are collected using face-to-face interviews lasting approximately one hour. The questionnaire consists of a ‘core’ module, covering standard socio-demographic characteristics, social and moral values, attitudes towards ethnicity and attitudes towards religion. One of the striking features of the ESS is the high degree of comparability in the data collected across nations coupled with a high response rate in all participating countries. In addition, the ‘‘principle of equality or equivalence” applies to sample selection translation of the questionnaire, and to all methods and processes. The samples are representative with comparable estimates of ‘‘eligible residential populations in each country” aged 15 or older who are resident within private households, regardless of nationality and citizenship or language. Comparability of the estimates is achieved if the transition process from gross samples to net samples is not seriously biased, that is if the response rates are high and appropriate auxiliary data are collected to aid weighting. The national research centres in each country utilize external and internal information in order to fill the non-response rates and random sub-samples are in place to be utilized when the gross response rates are low. A ‘rotating’ module includes in-depth questions on specific issues of interest and other topics asked on a rotating basis during the scheduled rounds of the survey taking place at two-year intervals. The survey covers a representative sample of approximately 2000 individuals per round per country.8 Data on the following 19 countries are analyzed: Austria, Belgium, Czech Republic, Denmark, Finland, Germany, Greece, Hungary, Ireland, Luxembourg, Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom. Our dependent variable is life satisfaction, as reported in the ESS based on responses to the question: ‘‘All things considered, how satisfied are you with your life as a whole nowadays? Please answer using this card, where 0 means extremely dissatisfied and 10 means extremely satisfied”.9 Fig. 1 shows the mean life satisfaction scores across the 19 European countries under consideration. Denmark achieved the highest score at over 8, whilst Greece, Hungary, Poland and Portugal recorded the lowest scores during the reporting periods. In general, Western European countries score higher than Eastern European ones. Due to the ordinal nature of the life satisfaction variable, we estimate ordered probit models, assuming that a latent cardinal measure of the dependent variable, a proxy for the unobserved level of utility, is related to the observed personal characteristics and income in the following way:

Si ¼ b0 zi þ ei ;

ð1Þ

where zi is a vector of explanatory variables describing individual characteristics and income measures, b is a vector of parameters to be estimated and ei is a random error term, assumed to be normally distributed. Then, the cardinal index of utility Si* is mapped into the self-reported subjective ordinal scales of life satisfaction Si

8 For a detailed description and authoritative evaluation of the sampling design and implementation of the ESS survey see Lynn, Hader, Gabler, and Laaksonen (2004). 9 As mentioned in footnote 2, we also ran regressions using ‘‘happiness” as the dependent variable. These results are available upon request.

46

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

8 0 > > > > > 1 > > > > > <2

if if if Si ¼ : > > > : > > > > > : > > : 10 if

1 6 Si 6 l1 l1 < Si 6 l2 l2 < Si 6 l3 ð1aÞ

l10 < Si 6 1

where li represents thresholds to be estimated (along with the parameter vector b).10 The probabilities of Si taking different scale values are determined as: Prob(Si = 0|zi) = U(l1b0 zi), Prob(Si = 1|zi) = U(l2b0 zi)U(l1b0 zi), . . ., Prob(Si = J|zi) = 1U(l 0 J1b zi), where U(..) is the cumulative density function. It is worth noting that the parameters b do not measure marginal effects on life satisfaction. A positive sign for a particular b indicates that the entire distribution of Si* is shifted to the right as the value of the associated variable increases. Such a change will have different impacts on the probabilities of life satisfaction taking different scores: whilst the probability for the lowest score (i.e. Prob(Si = 0|zi)) is definitely reduced and the probability for the highest score (Prob(Si = 10|zi)) is definitely increased, the probabilities for the intermediate cases can only be determined numerically. To assess the quantitative significance of the income and relative income effects on life satisfaction we calculate marginal effects based on the estimated regressions coefficients.11 Income is reported in banded categories in Euros, with respondents being presented with a card of eleven income bands with from below 70€ to over 2310 Euros per week. The exact income question asks: ‘‘Using this card, if you add up the income from all sources, which letter describes your household’s total net income?” The ESS data also provide information on a rich set of standard demographic and labour market characteristics that we use as controls in our life satisfaction regressions. Such controls include personal characteristics, education, and health. Information on past unemployment experience is also used to evaluate whether individuals’ perceptions about their current economic situation is influenced by past income shocks, usually associated with unemployment. Our proxy for reference income is based on McBride’s (2001) definition of the reference group that includes all individuals who are in the age range of 5 years younger and 5 years older than the individual concerned. For example, the reference income category for a 40-year old male worker living in Germany in 2002 will be the income category that male workers in Germany during the same year who are between 35 and 45 years of age are expected (i.e. on average) to belong to.12 The definitions and sample means of all variables used in our analysis are in Appendix 1. We limit our sample to full-time salaried employees, which yields 30,285 observations fairly equally split between 2002 and 2004. Thus, for the great majority of respondents most of their income is labour income, which allows us to control for heterogeneity in life satisfaction responses due to unemployment and other labour market states as well as for variation in responses due to differences in the generosity of unemployment and welfare benefit systems across Europe. 4. Empirical findings Table 1 reports the results for the life satisfaction regressions. Column (1) shows the estimated coefficients with reference income being excluded from the list of explanatory variables. Consistent with the findings of previous studies, the estimated coefficients reveal that men tend to report lower satisfaction than women, while life satisfaction exhibits a U-shaped relationship with age. This is a pattern which is well documented in the literature and reflecting life-cycle aspects of individuals’ social, family and economic circumstances (e.g. Alesina, DiTella, & MacCulloch, 2004; Blanchflower & Oswald,2004, 2006). Being married has a positive effect on life satisfaction, while the opposite is true for divorce, separation and widowhood. The results also reveal a negative effect of the presence of children on life satisfaction. As expected, good health has a significant positive effect. There is some weak evidence that higher education qualifications tend to exert a negative impact on life satisfaction, with the estimated coefficient of ‘Post tertiary’ education being negative and statistically significant. Fernandez and Kulik (1981) report a similar effect, which could be explained by the fact that education raises aspirations that are often not easily fulfilled.13 10

For a discussion of the ordered probit model, see McKelvey and Zavoina (1975). The procedure for estimating marginal effects is briefly described here. The lowest income band (i.e., below 70 Euros) is used as the base group in the regression. For the remaining ten income bands, the estimated bj (j = 1,. . .,10) measures how the entire distribution of Si* is shifted (to the right if bj > 0 or to the left if bj < 0) if an individual moves up from the base group to a higher income band j. To calculate the marginal effect on life satisfaction of this upward movement from the base band to a higher income band j we follow the following three steps: (i) obtain the values of Si* assuming that the categorical variable for income band j takes the value of 0 and then 1 successively whilst all the other variables are given their mean values in both cases; (ii) obtain the probabilities Prob(Si = k|zi), (k = 0,. . .,10 being the life satisfaction scores) in the above two scenarios; (iii) the marginal effects are obtained by taking the differences between the two sets of probabilities. Since some marginal effects are positive and the others are negative, we also calculate the marginal effect in terms of the difference in the expected value of life satisfaction scores. 12 We also defined the reference group to contain all individuals with a similar education level, inside the same age bracket, and living in the same country, as suggested by Ferrer-i-Carbonell (2005). Our main results, especially regarding the effect of income and reference income, remain largely unaffected. For a summary of the various methods to calculate reference income in the literature, see Clark et al. (2008a). 13 When re-estimating the regression using the Ferrer-i-Carbonell (2005) definition of reference income, the education coefficient becomes positive and significant. This is partially due to the high correlation between educational dummies and the reference income measure, defined as the average income of individuals with similar educational qualifications. However, other estimated coefficients and especially those of income and reference income are largely unaffected by the use of this alternative measure of reference income. Multicollinearity among the remaining explanatory variables is not a cause for concern either. A matrix of sample correlation coefficients for all explanatory variables is available upon request. 11

47

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51 Table 1 Life satisfaction regressions (ordered probit) (1)

(2)

Coeff.

|t-ratio|

Coeff.

|t-ratio|

Male Age Age2 Married Separated Divorced Widoweda Children Good health

0.123 0.042 0.050 0.212 0.289 0.068 0.067 0.123 0.358

10.23 13.11 14.89 11.37 5.96 2.62 2.01 10.23 46.21

0.124 0.028 0.034 0.213 0.288 0.066 0.067 0.124 0.358

10.25 3.66 3.80 11.43 5.93 2.55 2.01 10.25 46.18

Education Low secondary High secondary Post secondary Tertiary post Tertiaryb

0.003 0.027 0.009 0.020 0.055

0.14 1.16 0.31 0.76 1.70

0.003 0.027 0.009 0.020 0.056

0.11 1.17 0.30 0.79 1.72

Unemployment In the last 12 months In the last 5 years

0.023 0.009

3.42 1.37

0.023 0.009

3.42 1.37

Income [weekly] 70–120€ 120–230€ 230–350€ 350–460€ 460–580€ 580–690€ 690–1150€ 1150–1730€ 1730–2310€ > 2310€c

0.137 0.213 0.301 0.363 0.406 0.481 0.516 0.572 0.551 0.583

3.68 6.05 8.30 9.79 10.68 12.44 13.58 13.59 10.17 8.83

0.138 0.213 0.301 0.364 0.407 0.483 0.519 0.575 0.553 0.586

3.69 6.05 8.32 9.80 10.71 12.47 13.63 13.65 10.21 8.88

0.096

1.98

Year dummy 2004 Country dummies

0.018 Yes

1.46

0.015 Yes

1.17

Log-likelihood Likelihood-ratio v2[Prob] Number of observations

55885.764 8275.21 [0.000] 30285

Reference income

55883.904 8278.93[0.000] 30285

Notes: Excluded categories. a Never married. b Primary or less education. c Less than 70€.

As the estimated coefficients in column (1) show, past unemployment has a positive effect on life satisfaction, with such an effect being stronger for unemployment experienced more recently (in the last twelve months) as opposed to unemployment in the more distant past (in the last five years). The most plausible explanation for this positive effect is that it captures the increase in income usually associated with the transition from unemployment back to paid employment. Interestingly, this positive effect is stronger for the Scandinavian countries but it disappears altogether when estimating the model separately for the Southern Mediterranean countries or Eastern Europe, possibly reflecting differences in social welfare systems and to what extent unemployment is a viewed as social norm.14 The results provide clear evidence of a strong relationship between income and life satisfaction. Moving from the lowest to the highest income group, the estimated coefficients increase monotonically and are all statistically significant. The associated marginal effects show these effects are remarkably strong. An increase in income from the base band (70–120€) to the next income band raises the expected value of life satisfaction score by 0.27 points, whilst the movement to the middle income band (350–460€) increases life satisfaction by 0.71 points. A move to the highest band (> 2310€) improves the score by 1.08 points. In fact, as a respondent moves to successively higher income bands, the expected value of life satisfaction scores increases almost monotonically. In column (2), we re-estimate the model by controlling for relative income. As the results show, reference income has a negative and significant coefficient, suggesting that comparison effects in life satisfaction are present, thus lending support 14 According to Clark, Georgellis, and Sanfey’s (2001), ‘scarring hypothesis’, past unemployment exerts a significant negative effect on life satisfaction even for those currently employed (i.e. unemployment ‘scars’ psychologically). Other studies using large-scale panel data find that adaptation to unemployment is generally slow and incomplete (see Clark et al. 2008b; Lucas, Clark, Georgellis, & Diener, 2004).

48

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

Table 2 Life satisfaction regressions: Eastern Europe (ordered probit) (1)

Income (weekly) 70–120€ 120–230€ 230–350€ 350–460€ 460–580€ 580–690€ 690–1150€ 1150–1730€ 1730–2310€ > 2310€a

(2)

Coeff.

|t-ratio|

Coeff.

|t-ratio|

0.226 0.328 0.461 0.500 0.479 0.560 0.485 0.124 0.909 0.156

4.79 6.60 7.74 6.97 5.44 4.68 3.83 0.54 2.68 0.27

0.227 0.329 0.463 0.501 0.481 0.563 0.484 0.117 0.918 0.166

4.82 6.62 7.77 6.98 5.45 4.70 3.82 0.51 2.70 0.28

0.084

1.61

Reference income Log-likelihood Likelihood-Ratio v2[Prob] Number of observations

10220.492 969.55[0.000] 4913

10220.124 970.28[0.000] 4913

Notes: Other regressors as in Table 1 and the same excluded categories. a Excluded category: Less than Euros 70.

to the relative utility hypothesis.15 Senik (2008) finds the same positive effect of relative income on subjective well-being across Western European countries, using a variety of data sources. However, when using the first wave of the ESS data the coefficient of relative income for Western Europe is not statistically significant. As she explains, this is partly due to the crude measure of reference income used in her ESS analysis. The inclusion of reference income in the list of explanatory variables tends also to mitigate the strength of the estimated income effects. In comparison with the case where reference income is excluded, the expected values of life satisfaction scores are reduced across the board. For example, an increase in income from the lower band (70–120€) to a middle income band (350–460€) is associated with an increase in the life satisfaction score by 0.61 points. When repeating the analysis by limiting our sample to the Eastern European countries (see Table 2) any evidence of social comparison effects disappears. In contrast, we find that reference income exerts a positive and significant effect on life satisfaction, consistent with the ‘‘tunnel effect” hypothesis (see Senik, 2004b). The same positive effect of relative income in the case of Eastern European countries is found by Senik (2008). The fact that we confirm such an effect by using additional ESS data, different measures for relative income, and different specifications/methods from those used by Senik (2008) provides further confidence in the robustness of our findings and the validity of the ‘tunnel effect’ hypothesis in the case of transition economies. The rapid growth of income that certain segments of the population experienced during the period of economic transition increased the expectations of the remainder of the population for higher incomes in the future. In a sense, pockets of high income and prosperity in the economy offer an optimistic outlook for those who are yet to catch up. As Hayo and Seifert (2003) highlight, during the early 1990s, there was a general climate of optimism among Eastern Europeans that their economic situation would improve, or at least not deteriorate, in the next five years. During these early years of reform, catching-up with the well-being levels of industrialized countries would dominate any relative income effects. Therefore, one should expect that such ‘‘tunnel effects” might be short-lived as those at the lower end of the income distribution realise that the gap between their economic position and that of the high earners widens without any prospects of ever catching up with them. If this conjecture is valid, then in the economies of transition in Eastern Europe we should expect ‘tunnel effects’ to be more prominent during the early years of economic reform and start to weaken as time passes by in a non-monotonic fashion. Given that our sample is based on data almost ten years after the ex-communist Eastern European countries embarked on a programme of economic reforms towards free market economies, evidence of ‘tunnel effects’ might not be as strong as in the earlier years of economic transition. 5. Further results Table 3 summarises the results of re-estimating various specifications of the life satisfaction regressions, treating income as a continuous variable by using the midpoints of each banded income category. In general, the results confirm the positive influence of income on life satisfactions. With the exception of the Eastern European countries, reference income has a negative coefficient, reinforcing the ever-growing literature supporting the view that this variable should be part of a conventional utility function.

15

The same effect is found when using Ferrer-i-Carbonell’s (2005) proxy for relative income.

49

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51 Table 3 Life satisfaction regressions: Further results INCOME (weekly)

Reference income

Coeff.

|t-ratio|

Coeff.

|t-ratio|

Log-likelihood

Number of obs.

[1] All countries

0.118

3.85

0.043

[2] Eastern Europe (Czech Republic, Poland, Hungary, Slovenia)

0.147

4.89

0.039

2.34

51334.34

30285

1.96

8232.02

4913

[3] Southern Europe (Greece, Spain, Portugal)

0.113

2.04

[4] Scandinavia (Denmark, Sweden, Finland, Norway)

0.129

2.24

0.060

1.88

6345.04

2725

0.045

2.81

8668.03

[5] Central Europe (Austria, Switzerland, Germany, Britain, Ireland, Luxembourg, Holland, Belgium)

0.131

8453

2.22

0.052

2.32

24737.07

14194

[6] All countries (Age < = 40) [7] All countries (Age > 40)

0.099

2.23

0.051

1.74

38909.88

11044

0.123

2.21

0.065

2.16

43261.62

19241

[8] All countries (education: primary, low secondary, high secondary)

0.077

1.77

0.037

1.68

39738.56

15142

[9] All countries (education: post secondary, tertiary, post tertiary)

0.087

2.04

0.055

1.95

25326.35

15142

Notes: Other regressors as in Table 1.

As shown in models [3–5] in Table 3, Southern Europeans are the most ‘‘affected” by the negative influence of the reference income and the Scandinavians as the least ‘‘affected”. In models [6] and [7], we investigated whether the effect persists for different age groups. By running separate regression for those below and those above 40 years of age, we find that the reference income effect is more pronounced for the older group. In models [8] and [9], we split the sample into two groups by educational attainment. Workers with higher levels of education seem to derive higher levels of life satisfaction as their absolute income increases than those with lower levels of education. However, those with higher levels of education are associated with a more pronounced negative reference income effect. The results in Table 3 offer additional reassurance about the robustness of the causal relationship between income, reference income and subjective wellbeing.

6. Conclusions In recent years, support for the notion that reference values are important in affecting individuals’ behaviour has become widespread both in the psychology and the economics literature. Economists, in particular, tend to agree that decision makers evaluate the options available to them not on the basis of the absolute values of wealth or welfare but rather of relative values, implying that utility is relative in nature, a hypothesis supported by an increasing number of empirical studies in recent years. In this paper, we have re-examined the link between income and subjective well-being for a number of European countries, paying particular attention to whether relative income is indeed an important determinant of subjective well-being. Our results support both the absolute and relative income hypotheses. Focusing on the latter, there is clear evidence that the income of a reference group exerts a negative effect on well-being, even after controlling for absolute income and other personal and demographic characteristics. More intriguing, perhaps, is the fact that such social comparison effects tend to disappear when we limit our analysis to the Eastern European countries. In the case of Eastern Europe, reference income has a positive effect on happiness, consistently with the presence of a ‘tunnel effect’. To the extent that the ‘pursuit of happiness’ enters the political agenda, our results highlight the existence of a clear wedge between Western and Easter European countries that can have important implications for the design of welfare reforms and income redistribution policies. If, as our results seem to imply, an increasing income gap between the rich and poor reduces well-being due to social comparisons, alleviating income inequality moves higher up in the policy agenda. In contrast, if higher inequality raises the expectations of the poor that they are to enjoy higher incomes in the future (i.e. ‘tunnel effect’), then increased income inequality during rapid growth at the early stages of reforms becomes socially and politically more acceptable. Acknowledgements We thank four anonymous reviewers and participants at the 65th International Atlantic Economic Conference for helpful comments and suggestions on an earlier draft.

Appendix 1 Table A1.

50

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

Table A1 Variables definitions and sample means Definition

Mean 2002

Life satisfaction Male Age Married Separated Divorced Widowed Never married Children Good health EDUCATION Up to primary Low secondary High secondary Post secondary Tertiary Post tertiary

All things considered, how satisfied are you with your life as a whole nowadays? Please answer using this card, where 0 means completely unsatisfied and 10 means completely satisfied Dummy variable: 1 = Male; 0 Female Age in years Dummy variable: 1 = Married; 0 otherwise Dummy variable: 1 = Separated; 0 otherwise Dummy variable: 1 = Divorced; 0 otherwise Dummy variable: 1 = Widowed; 0 otherwise Dummy variable: 1 = Never Married; 0 otherwise Dummy variable: 1 = Children in household; 0 otherwise Subjective general health (physical and mental), Ordinal variable: 1 = Very Bad, 2 = Bad, 3 = Fair, 4 = Good, 5 = Very good

2004

7.153

7.037

0.499 45.853 0.616 0.016 0.087 0.051 0.227 0.466 3.874

0.493 48.238 0.611 0.016 0.090 0.051 0.229 0.457 3.888

0.116 0.204 0.377 0.086 0.160 0.061

0.132 0.177 0.392 0.062 0.193 0.056

Dummy variable: 1 = whether the respondent has any periods of unemployment over the last 12 months; 0 = otherwise Dummy variable: 1 = whether the respondent has any periods of unemployment over the last 05 years; 0 = otherwise

0.102

0.099

0.097

.098

Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy Dummy

0.064 0.107 0.125 0.125 0.121 0.111 0.184 0.072 0.020 0.010

0.054 0.106 0.113 0.124 0.113 0.122 0.208 0.086 0.024 0.013

Dummy Dummy Dummy Dummy Dummy Dummy

variable: variable: variable: variable: variable: variable:

1 = Up to primary; 0 otherwise 1 = Low secondary; 0 otherwise 1 = High secondary; 0 otherwise 1 = Post secondary; 0 otherwise 1 = Tertiary; 0 otherwise 1 = Post tertiary; 0 otherwise

UNEMPLOYMENT In the last 12 months In the last 5 years INCOME [weekly] (Household’s total net income, all sources) 70–120€ 120–230€ 230–350€ 350–460€ 460–580€ 580–690€ 690–1150€ 1150–1730€ 1730–2310€ > 2310€

variable: variable: variable: variable: variable: variable: variable: variable: variable: variable:

1 = between 70 and 20€; 0 otherwise 1 = between 120 and 230€; 0 otherwise 1 = between 230 and 350€; 0 otherwise 1 = between 350 and 460€; 0 otherwise 1 = between 460 and 580€; 0 otherwise 1 = between 580 and 690€; 0 otherwise 1 = between 690 and 1150€; 0 otherwise 1 = between 1150 and 1730€; 0 otherwise 1 = between 1730 and 2310€; 0 otherwise 1 = more than 2310€; 0 otherwise

References Alesina, A., DiTella, R., & MacCulloch, R. (2004). Inequality and happiness: Are Europeans and Americans different? Journal of Public Economics, 88, 2009–2042. Argyle, M. (1999). Causes and correlates of happiness. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 353–373). New York: Russell Sage Foundation. Blanchflower, D., & Oswald, A. (2004). Well-being over time in Britain and the USA. Journal of Public Economics, 88, 1359–1386. Blanchflower, D., & Oswald, A. (2006). Is well-being U-shaped over the life cycle? Mimeo: University of Warwick. Brickman, P., & Campbell, D. (1971). Hedonic relativism and planning the good society. In M. H. Appley (Ed.), Adaptation-level theory: A symposium. New York: Academic Press. Clark, A., & Oswald, A. (1996). Satisfaction and comparison income. Journal of Public Economics, 61, 359–381. Clark, A., Frijters, P., & Shields, M. (2008a). Relative income, happiness and utility: An explanation of the Easterlin paradox and other puzzles. Journal of Economic Literature, 46, 95–144. Clark, A., Diener, E., Georgellis, Y., & Lucas, R. (2008b). Lags and leads in life satisfaction: A test of the baseline hypothesis. Economic Journal, 118, F222–F243. Clark, A., Georgellis, Y., & Sanfey, P. (2001). Scarring: The psychological impact of past unemployment. Economica, 68, 221–242. Cummings, R. (1998). The second approximation to an international standard for life satisfaction. Social Indicators Research, 43, 307–334. Diener, E., Sandvik, E., Seidlitz, L., & Diener, M. (1993). The relationship between income and subjective well-being: Relative or absolute? Social Indicators Research, 28, 195–223. Diener, E., & Suh, E. M. (1999). National differences in subjective well-being. In D. Kahneman, E. Diener, & N. Schwarz (Eds.), Well-being: The foundations of hedonic psychology (pp. 434–450). New York: Russell Sage Foundation. Duesenberry, J. S. (1949). Income, saving and the theory of consumer behavior. Harvard: University of Harvard Press. Easterlin, R. (1974). Does economic growth improve the human lot? Some empirical evidence. In P. A. David & M. W. Reder (Eds.), Nations and households in economic growth (pp. 89–125). New York: Academic Press. Easterlin, R. (1995). Will raising the income of all increase the happiness of all? Journal of Economic Behavior and Organization, 27, 35–47. Easterlin, R. (2001). Income and happiness: Toward a unified theory. Economic Journal, 111, 464–484. Easterlin, R. (2005). A puzzle for adaptive theory. Journal of Economic Behavior and Organization, 56, 513–521. Fernandez, R., & Kulik, J. (1981). A multilevel model of life satisfaction: Effects of individual characteristics and neighborhood composition. American Sociological Review, 46, 840–850. Ferrer-i-Carbonell, A. (2005). Income and well being: An empirical analysis of the comparison income effect. Journal of Public Economics, 89, 997–1019.

G.M. Caporale et al. / Journal of Economic Psychology 30 (2009) 42–51

51

Frey, B., & Stutzer, A. (1999). Measuring preferences by subjective well-being. Journal of Institutional and Theoretical Economics, 155, 1–24. Frey, B., & Stutzer, A. (2002). Happiness, economy and institutions. Economic Journal, 110, 918–938. Frijters, P., Haisken-DeNew, J. P., & Shields, M. A. (2004). Money does matter! Evidence from increasing real income and life satisfaction in East Germany following reunification. American Economic Review, 94, 730–740. Glatzer, W. (1991). Quality of life in advanced industrialized countries: The case of West Germany. In F. Strack, M. Argyle, & N. Schwarz (Eds.), Subjective well-being: An interdisciplinary perspective (pp. 261–279). Oxford: Pergamon. Gundelach, P., & Kreiner, S. (2004). Happiness and life satisfaction in advanced European countries. Cross-Cultural Research, 38, 359–386. Hagerty, M., & Veenhoven, R. (2003). Wealth and happiness revisited – Growing national income does go with greater happiness. Social Indicators Research, 64, 1–27. Hayo, B., & Seifert, W. (2003). Subjective economic well-being in Eastern Europe. Journal of Economic Psychology, 24, 329–348. Heady, B. (1991). An economic model of subjective well-being: Integrating economic and psychological theories. Social Indicators Research, 28, 97–116. Hirschman, A. (1973). The changing tolerance for income inequality in the course of economic development. Quarterly Journal of Economics, 87, 544–566. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291. Kahneman, D., & Krueger, A. (2006). Developments in the measurement of subjective well-being. Journal of Economic Perspectives, 20, 3–24. Lucas, R., Clark, A., Georgellis, Y., & Diener, E. (2004). Unemployment alters the set point of life satisfaction. Psychological Science, 15, 8–13. Lynn, P., Hader, S., Gabler, S., & Laaksonen, S. (2004). Methods for achieving equivalence of samples in cross-national surveys: The European Social Survey Experience. ISER Working Paper 2004-09. Colchester: University of Essex. Mas, A. (2006). Pay, reference points, and police performance. Quarterly Journal of Economics, 121, 783–821. McBride, M. (2001). Relative-income effects on subjective well-being in the cross-section. Journal of Economic Behavior and Organization, 45, 251–278. McKelvey, R., & Zavoina, W. (1975). A statistical model for the analysis of ordinal level dependent variables. Journal of Mathematical Sociology, 4, 103–120. Michalos, A. C. (1991). Global report on student well-being. Life satisfaction and happiness (Vol. 1). New York: Springer-Verlag. Oswald, A. J. (1997). Happiness and economic performance. Economic Journal, 107, 1815–1831. Rizzo, J., & Zeckhauser, R. (2003). Reference incomes, loss aversion, and physician behavior. Review of Economics and Statistics, 85, 909–922. Rojas, M. (2007). Heterogeneity in the relationship between income and happiness: A conceptual-referent-theory explanation. Journal of Economic Psychology, 28, 1–14. Senik, C. (2004a). Income distribution and well-being: What can we learn from subjective data? Journal of Economic Surveys, 19, 43–63. Senik, C. (2004b). When information dominates comparison: Learning from Russian subjective panel data. Journal of Public Economics, 88, 2099–2123. Senik, C. (2008). Ambition and jealousy: Income interactions in the ‘old’ Europe versus the ‘new’ Europe and the United States. Economica, 75, 495–513. Stutzer, A. (2004). The role of income aspirations in individual happiness. Journal of Economic Behavior and Organization, 54, 89–109. Van de Stadt, H., Kapteyn, A., & Van de Geer, S. (1985). The relativity of utility: Evidence from panel data. Review of Economics and Statistics, 67, 179–187. Van Praag, B. M. S. (1971). The welfare function of income in Belgium: An empirical investigation. European Economic Review, 2, 337–369. Van Praag, B. M. S., & Kapteyn, A. (1973). Further evidence on the individual welfare function of income: An empirical investigation in the Netherlands. European Economic Review, 4, 33–62. Veblen, T. (1899). The theory of the leisure class. New York: Macmillan.