Hedonic adaptation in the non-pecuniary domain
Aart Gerritsen* August 2008
Abstract:
Many studies stress that public policy should be aimed more at non-pecuniary goals, such as family life and job security, as opposed to pecuniary goals like national income. One of the reasons for this is the observation that people adapt relatively quickly to shocks in income, so that the long term effect on happiness of income is small. This study analyzes to what extent adaptation plays a role when it comes to non-pecuniary life events. We find no evidence for adaptation to marriage or divorce. Marriage increases and divorce decreases happiness lastingly. We do find some evidence of adaptation to unemployment. The mid- and long-term unemployed seem to be less unhappy than short-term unemployed individuals. Furthermore we show that with clever survey design future research on adaptation can be done with the help of relatively easily obtainable cross-section data. This opens up the possibility for more much-needed studies on adaptation in the non-pecuniary domain.
Keywords: Life satisfaction, hedonic adaptation, pecuniary vs non-pecuniary JEL Codes: D01, I31, C81
Socrates: “And when he remembered his old habitation, and the wisdom of the den and his fellow-prisoners, do you not suppose that he would felicitate himself on the change, and pity them?”
Glaucon: “Certainly, he would.”
Plato – The Republic
1. INTRODUCTION
In recent decades many studies question the wisdom of economic growth as an important goal for public policy. The argument is that income is relative in that its value in terms of
happiness depends on a reference point determined by past experience and other persons’ income. Therefore policy aimed at economic growth would be a zero-sum game in which both individuals’ absolute income and their reference levels increase, so that relative income stays the same and the effect on aggregate happiness is nil. Instead of pecuniary goals like
economic growth, public policy should thus focus much more on non-pecuniary goals such as family life and social security.
Implicit in this argument is that the same effects that withhold pecuniary outcomes to affect happiness do not play a similar role when it comes to the non-pecuniary domain. However, there is still no convincing evidence that this is the case. The main reason for this apparent lack of empirical evidence is the fact that good quality panel survey data is per definition a time consuming and costly effort to obtain and is thus often lacking. In this study we use a relatively new and unused cross-section dataset to investigate the role adaptation plays in non-pecuniary outcomes. Specifically we test whether the effect on happiness of marriage,
divorce, children, retirement and getting unemployed diminishes over time as respondents get used to their situation. In accordance with the claims of many studies we find that adaptation does not play a major role in how marriage, divorce, children and retirement influence happiness. Our analysis indicates that adaptation does play a role when it comes to unemployment. This adaptation effect however, is counteracted by other effects such as deteriorating future prospects. Furthermore we show that with clever survey design it is possible to analyze adaptation with cross-section data. This opens up many possibilities for further research.
The outline of our paper is as follows. In section 2 we will discuss how reference levels can affect happiness and make economic growth a zero-sum game. Also we will introduce a utility function that allows for a reference level and discuss earlier empirical work on
adaptation in the non-pecuniary domain. Section 3 will present the empirical strategy that we will apply and section 4 presents the results. Finally we will conclude.
2. THEORETICAL CONSIDERATIONS
2.1 Explanations and implications of the Easterlin paradox
One of the main catalysts of research in happiness – at least when it comes to economics – has been the Easterlin paradox. Owing its name to the seminal study by Easterlin (1974), this paradox refers to the observation that average levels of happiness in Western countries have been virtually stationary since the Second World War, while per capita income has been rising sharply. This apparent lack of link between national income and happiness has been
Blanchflower and Oswald, 2004). At the same time, an extensive number of studies point to a large positive effect of individual income on individual life satisfaction, both in a cross-section setting and in a dynamic panel setting (e.g. Helliwell, 2003; Ferrer-i-Carbonell 2005). So while on individual level more income does seem to lead to more happiness, this does not hold on an aggregate level.
The concept of hedonic relativism is usually offered as an explanation in order to reconcile these two at first sight contradicting observations. According to this concept people evaluate their current income in the light of a reference point. This point is generally believed to be formed by two different sorts of comparison. Firstly, there are external references like friends, colleagues, neighbors and family members who influence our judgment of how much income makes how much happiness. In the happiness literature this is usually referred to as social comparison. According to this hypothesis a raise in income of relatives, given one’s own level of income, increases the reference point and therefore diminishes the happiness derived from own income.
Secondly a reference point is affected by internal references, i.e. the individual’s past
experiences. The value in happiness terms of one’s current income depends on what level of income she is habituated to, which on its turn depends on past levels of income. Given current income, happiness will be higher, the lower past income is. That is, ceteris paribus, a person who has been receiving monthly earnings of 5,000 dollar all her working life will be less happy than someone who is currently earning 5,000 dollar but is used to earning much less. It also implies that a positive shock in income initially raises happiness but over time, as the reference point adapts to the current situation, happiness will gradually move back towards the original level. This effect is referred to as hedonic adaptation or the hedonic treadmill.
Clark, Frijters et al. (2008), by combining empirical findings from Di Tella et al. (2007) and Knight and Song (2006), suggest a utility function in which only 13 percent of the individual effect of income on happiness will survive in the long term on an aggregate level. They further argue that this will be an overestimate if new generations have higher reference levels than current generations, so that income might very well be completely relative. This has led to policy advises that deemphasize the pecuniary goals of public policy, like economic growth, and emphasize the non-pecuniary domain relating especially to factors like family life, friends, good health and employment (e.g. Layard, 2005; Easterlin 2003). However, such advises of course only make sense when the concepts undermining income’s effect on
happiness do not play a similar role in the non-pecuniary domain. There is still too little evidence on whether non-pecuniary outcomes are relative as well. In a recent study Easterlin (2005) claims to pose us with another puzzle. According to this study aspiration levels for income and consumption rise with actual income and consumption, while aspiration levels for marital status and the number of children remain constant over a person’s lifetime. These results however are highly speculative and very much in need of additional foundation.
This paper aims to contribute to this literature by empirically analyzing the scope of hedonic adaptation when it comes to the non-pecuniary domain. Specifically we will look at
unemployment, retirement, marriage, divorce and getting children and answer the question whether people’s reference levels tend to adapt as a reaction to changes in these situations. That is, when it comes for example to marriage we ask ourselves: does marriage increase happiness and does this effect dissipate over time as people habituate to their marital status?
2.2 Hedonic adaptation in the utility function
To illustrate hedonic adaptation I will make use of a utility function that exhibits reference levels. Such a utility function implies that an individual does not just derive utility from the absolute amount of consumption of a certain good, but rather derives utility from the amount of consumption relative to a reference level. If an individual consumes xi and has as reference
level xi*, then she derives utility from xi/xi*^ψ, with ψ the degree of relativity. The specific
form of the utility function is loosely based on Weinzierl (2006) and adapted to fit the
forgetting about other factors for a moment, we pose a standard utility function which we expand by including a reference component:
U(xt,xt*)= 1/(1-γ) (xt/xt*^ψ)^(1-γ) (1)
xt=exp(Mt) (2)
xt* = Πj xj^αj , Σjαj = 1, dαj/dj<0 (3)
Here Mt takes the value 1 if the individual is married at time t and zero otherwise. Equation
(1) gives the utility function and equation (2) the functional form in which marriage enters the utility function. Such a transformation is necessary because we mainly make use of dummy variables. Equation (3) equates the reference level of marriage to a Cobb-Douglas function of past levels of the marriage function exp(Mt) with shares αj decreasing with time and the sum
of which equaling 1. Notice that ψ represents the extent of adaptation: with ψ=0 past marital status does not affect current utility; with ψ=1 marital status is completely relative and there is full adaptation where dU∞/dMt = 0. We substitute (2) and (3) into (1) and take logarithms to
obtain:
lnU(Mt,Mt*) = ln1 + ln(1-γ) + (1-γ) (Mt - ψΣjαjMt-j) (4)
This can be rewritten to yield:
ln U = B0 + Σj Bj(Mt+1-j,Mt-j) = B0 + B1(Mt-Mt-1) + B2(Mt-1-Mt-2) + B3(Mt-2-Mt-3) + ... (5) B0= ln1 - ln(1-γ) B1= (1-γ) B2= (1-γ)(1 - ψ(α1)) B3= (1-γ)(1 - ψ(α2 + α2)) …
the period concerned the individual got married and 0 otherwise. In the existence of
adaptation ψ>0 and B0>B1>B2>…. In case of complete adaptation, ψ=1 and Bj goes to zeroas
j goes to infinity. How fast Bj converges to zero depends on the shares of past marital statuses
in the reference level αj. For example if there is indeed complete adaptation and the reference
level is completely determined by the marital status of last year, α1=1 and αj=0 for j≠1 and
Bj=0 for all j≠0. Before discussing our empirical methodology we will discuss the most
relevant earlier contributions to the literature on adaptation.
2.3 Empirical evidence of adaptation in the non-pecuniary domain
As stated earlier there is still not much evidence whether adaptation plays a role when it comes to non-pecuniary aspects of life. Moreover, the studies that exist often contradict each other. Frederick and Loewenstein (1999) is still the most elaborate survey on hedonic
adaptation in a broad array of life domains. Specifically they indicate seven non-pecuniary domains where on the basis of empirical research it seemed “possible to draw at least tentative conclusions about the extent of hedonic adaptation” (p. 311). When it comes to undesirable experiences it seems that people do not adapt to neighborhood noise. They do however adapt entirely or almost entirely to incarceration, disability and disease and bereavement. Losing a spouse does take a very long time of adaptation of up to two decades. As for desirable experiences, Frederick and Loewenstein discuss the empirical evidence of adaptation to cosmetic surgery, sexually arousing stimuli and foods. No adaptation to cosmetic surgery is found and it thus appears to increase a person’s happiness in a lasting way. The evidence of adaptation to sexually arousing stimuli is mixed. In some studies sexual arousal diminishes with repeated presentation of stimuli like erotic slides, other studies find no reduction in arousal. In general people do seem to adapt to foods, as is evidenced by the fact that most adults usually enjoy many substances which they used to find aversive, such as coffee, beer, tobacco and chili peppers.
when it comes to big ticket consumption goods, the more of them people have, the more of them they desire. Thus Easterlin argues that people get used to the number of consumption goods and in order to be happy they will constantly need more. Contrary to this observation, the number of people that think a happy marriage is part of the good life is fairly stable, no matter if people get married or not. Also the desire for having children and the ideal number of children is relatively stable over time. Easterlin concludes from this that when it comes to the non-pecuniary domain, or at least to marriage and children, adaptation does not play such an important role as with consumption goods.
This finding has been rejected by Lucas et al. (2003) and Lucas and Clark (2006) who analyze the German Socio-Economic Panel Study (GSOEP), a nationally representative panel survey. The big advantage of using such a panel study, in which the same individuals are being followed over a certain number of years, is the possibility to control for individual fixed effects. That would allow for dropping the assumption that every individual evaluates their happiness on the same discrete scale (Weinzierl, 2006). A disadvantage of panel data as compared to a cross sectional dataset like the one we will use in our study is the fact that it is much more costly and time consuming to obtain. Also, these panels are usually very much unbalanced and attrition tends to cause significant problems. Using the GSOEP, Lucas and his colleagues find that individuals adapt quickly and entirely to marriage and only receive a short-term boost in happiness which dissolves completely after a few years. Zimmermann and Easterlin (2006), evaluating the same dataset, find that while married individuals do adapt to marriage, this adaptation is not complete and as such individuals do receive a lasting increase in happiness from a marital bond.
When it comes to divorce, the empirical evidence on adaptation is even scarcer but it
generally points at incomplete adaptation (e.g. Johnson and Wu, 2002; Lucas, 2005). A recent study by Clark, Diener et al. (2008) however find, analyzing GSOEP data, full adaptation to both marriage and divorce. They also find full adaptation to giving birth to a child and only little adaptation to unemployment. To our knowledge this is the only study that studies adaptation to giving birth to children. Adaptation to unemployment has been found to be limited by Clark et al. (2002) as well.
longitudinal panel data like the GSOEP. It is argued that such data is necessary as adaptation is by definition a dynamic process and as such needs multiple observations over the life of the individual. However, our data is essentially cross sectional as it was collected at a certain time and only contains happiness data for that point in time. But our dataset does contain detailed information on important events during the life time of every respondent, which allows us to test for adaptation. The advantage is that we can test for individuals across a wide area of 29 countries without having to take account of typical panel-data problems like attrition and unbalanced data.
3. DATA AND METHODOLOGY 3.1 Data
For our empirical analysis we rely on data from the Life in Transition Survey (LiTS), a representative cross-section survey of households in 29 countries. These are Turkey and all countries from the former socialist bloc except Turkmenistan. The survey, conducted by the European Bank of Reconstruction and Development in 2006, contains data for 1,000
observations per country on a mix of subjective perceptions and objective traits. Crucial for our analysis is the fact that while it is essentially a cross-section dataset, it contains
information on a person’s family and working life for every year between 1989 and 2006. This allows us to test for hedonic adaptation in these domains.
Life Satisfaction
The dependent variable that proxies for utility is the answer on the following five-step life satisfaction question: ‘All things considered, I am satisfied with my life now’. Possible answers are: (1) Strongly disagree, (2) Disagree, (3) Neither disagree nor agree, (4) Agree or (5) Strongly agree. Descriptive statistics per country are presented in the appendix. For further descriptive information on the LiTS we refer to EBRD (2007).
Marriage/divorce/children/retirement
The answers on these questions allow us to identify for every respondent whether they got married, divorced, got children or retired and in which year. Ceteris paribus in the case of hedonic adaptation in these domains we would expect a respondent that got married in 2006 to be happier than one who got married in 2000. We will aggregate this data over periods of three years.
Unemployment
For the respondent’s employment status there is monthly data going back 12 months. That is for each of the past 12 months we know whether the respondent was either employed, voluntarily unemployed or involuntarily unemployed. We further distinguish between the terms of (un)employment so that we end up with the following categorical dummies:
• Employed: > for more that 1 year
> for at least 4 months and up to 1 year
> for 3 months or less
• Involuntarily unemployed: > for more than 1 year
> for at least 4 months and up to 1 year
> for up to 4 months
• Voluntarily unemployed
Control variables
that most of the people can be trusted, or that you can’t be too careful in dealing with people?’.
3.2 Empirical methodology
The empirical model we estimate is based on equation (5) which we restate here for convenience:
ln Ut = B0 + Σj Bj(Mt+1-j,Mt-j)
= B0 + B1(Mt-Mt-1) + B2(Mt-1-Mt-2) + B3(Mt-2-Mt-3) + ... (5)
We will expand this utility function by including, besides marriage, similar adaptation terms for divorce, retirement, getting children and (un)employment. Other controls we include in our estimation are the individual’s gender, age, subjective health, education level, relative wealth and trust.
As a measure of utility we will take the answer on the five-step life satisfaction question.1 This is a necessary but by no means innocuous step. We will estimate the equation by ordered probit so that we do not need to assume cardinality, in that the jump in happiness associated with going from 1 to 2, does not need to be of the same size as going from 3 to 4. What we do need to assume however, is that everybody has the same interpretation of the five-step ladder. This is a strong assumption and a drawback of our analysis as compared to for example Weinzierl (2006) who uses panel data in order to control for individual fixed effects. However, it is an assumption that is conventionally made in cross-sectional analyses.
As a baseline model we will first estimate a happiness equation in which there is no hedonic adaptation, so that in terms of equation (1) ψ=0. The estimated equation is then as follows:
LS = β0+β1married+β2divorced+β3children+β4retired+β5invol.unemployed
+β6vol.unemployed+Xβ7+ε (6)
For matters of consistency childrenindicates whether the person has got any children or not. As an alternative we also estimated the equation using the number of children but this did not alter the results. X is a vector of individual specific variables and country dummies. When incorporating hedonic adaptation as in equation (5) we need to determine the unit of measurement of time t and the number of lags to include. Due to limitation in the data we make different choices for marriage, divorce, children and retirement on the one hand and employment status on the other. For the first group of variables we include three-yearly lags up until the period 1989-1991. That is, for marriage we include variables that indicate whether the respondent got married between 2004 and 2006, between 2001 and 2003 etcetera. For unemployed we take shorter intervals as is discussed in the previous section.
There are a number of ways in which the length of employment/unemployment can have an influence on happiness. Focusing for a moment on unemployment, the theory of adaptation would predict a lower level of happiness at the moment a person gets unemployed, which would return to initial values over time. On the basis of adaptation we would thus expect short term unemployment to have a negative impact on happiness and mid-term and long-term unemployment a less negative or no impact. On the other hand one would predict that uncertainty about future employment and income increases over time: after one year of unemployment the hope of ever getting a new job will usually be much lower than after just a month of unemployment. This implies a negative impact on happiness which gets more negative over time. Other factors like deteriorating self esteem may also play a significant role. For employment the opposite would hold. In the case of adaptive reference levels a person who just recently got a job would be much happier, while this happiness fades away over time. But also in the case of employment one could expect an influence of job security: the longer one would have a stable job, the more secure the person would feel.
The estimated equation incorporating the possibility of hedonic adaptation will be the following:
LS = β0+
β1mar0406+β2mar0103+β3mar9800+β4mar9597+β5mar9294+β6mar8991+
γ1div0406+γ2div0103+γ3div9800+γ4div9597+γ5div9294+γ6div8991+
δ1child0406+δ2child0103+δ3child9800+ δ4child9597+ δ5child9294+ δ6child8991+
θ1unemp0003+θ2unemp0412+θ1unemp12up+θ1emp0003+θ1emp0412+θ1volunemp+
Xλ+ε (7)
Here the second line gives the variables for marriage. Mar0406 takes the value 1 if the person got married between 2004 and 2006 and 0 otherwise; mar0103 takes the value 1 if the person got married between 2001 and 2003 and 0 otherwise etcetera. For the third, fourth and fifth line the same holds but for respectively divorce, getting children and getting retired. The sixth line includes the employment variables. Here unemp0003 for example stands for the short term (up to three months) unemployed, while emp0412 stands for middle term (four to twelve months) employed. The long term employed, emp12up, is taken as control group so that for example θ1 indicates how much happier short term unemployed people are as compared to
long term employed people. If there is adaptation we would expect coefficients to decrease as the period that the variable refers to gets further away. So with adaptation to marriage we would expect β1>β2>β3>….
4. RESULTS
4.1 Marriage, divorce, children and retirement
We will first discuss the results on marriage, divorce, getting children and retirement before turning our attention to employment and unemployment. Table 1 shows the result of
Table 1 – Ordered probit estimation without adaptation Coef. P Married 0.088 0.000 Divorced -0.182 0.000 Retired 0.009 0.696 Children -0.004 0.836 Vol. Unemployed 0.052 0.004 Invol. Unemployed -0.232 0.000 Gender 0.018 0.229 Age 25-34 -0.137 0.000 Age 35-44 -0.144 0.000 Age 45-54 -0.138 0.000 Age 55-64 -0.142 0.000 Age 64 up -0.053 0.130 Health 0.189 0.000 Educ.: vocational 0.010 0.715 Educ.: sec 0.026 0.302 Educ.: high 0.127 0.000 Wealth 0.215 0.000
Trust: compl. distrust -0.290 0.000 Trust: some distrust -0.055 0.020
Trust: some trust 0.153 0.000
Trust: compl. trust 0.269 0.000
N=26846, p-levels based on clustered standard errors, country dummies suppressed
When we take account of the dynamic aspect and test for adaptation, the results are depicted in table 2. Results significant at the five percent level are expressed in bold. As we can see, getting married appears to have a very significantly positive effect on happiness initially in the first three years. The next three years, the effect is still positive and of virtually the size as for the first three years. People who have been married for seven to nine years do not seem to be happier than non-married, non-divorced persons at all. What is striking from our estimation is that people who are married for more than nine years are again generally happier. As such, there is no evidence of any adaptation to being married: marriage seems to have a lasting positive effect on happiness. In our eyes, there can be two explanations for the observed insignificant effect of being married for seven to nine years. First it can simply be a data oddity. Second, it might indicate the so-called ‘seven-year itch’: named after the 1955 Billy Wilder picture this term refers to the observation in psychological research that on average after seven or eight years marital satisfaction decreases, e.g. Kurdek (1999). Kurdek studies only ten years of marriage; our analysis shows that married couples surviving the seven-year itch are again happier.
Table 2 – Orderd probit estimation with adaptation Coef. P Married 04-06 0.129 0.005 Married 01-03 0.125 0.000 Married 98-00 0.066 0.141 Married 95-97 0.124 0.020 Married 92-94 0.118 0.002 Married 89-91 0.080 0.015 Divorced 04-06 -0.165 0.007 Divorced 01-03 -0.231 0.001 Divorced 98-00 -0.090 0.209 Divorced 95-97 -0.288 0.002 Divorced 92-94 -0.248 0.014 Divorced 89-91 -0.035 0.727 Got child 04-06 -0.015 0.633 Got child 01-03 -0.006 0.842 Got child 98-00 -0.009 0.711 Got child 95-97 -0.025 0.369 Got child 92-94 0.005 0.803 Got child 89-91 -0.016 0.545 Retired 04-06 0.007 0.870 Retired 01-03 -0.004 0.909 Retired 98-00 -0.027 0.565 Retired 95-97 0.021 0.571 Retired 92-94 0.050 0.176 Retired 89-91 0.005 0.905 Vol. unemployed 0.046 0.013 Invol. unemployed 1-4m -0.347 0.000 Invol. unemployed 4-12m -0.163 0.076 Invol. unemployed 12m up -0.236 0.000 Employed 1-3m -0.045 0.178 Employed 4-12m -0.050 0.098
N=26846, p-levels based on clustered standard errors, country dummies and other controls suppressed
A similar pattern as that for marriage holds for divorce. The first six years of divorce seem to have a significantly negative impact on happiness, while people who are seven to nine years divorced do not seem to have a different level of happiness than non-married, non-divorced persons. People who have been divorced for more than nine years again seem to have a lower happiness score. We see this as evidence against the hypothesis of adaptation to marriage and divorce. The impulse response functions for getting married and divorced, including
Figure 1: Impulse Responses Marriage -0.050 0.000 0.050 0.100 0.150 0.200 0.250
1-3y 4-6y 7-9y 10-12y 13-15y 16-18y
Divorce -0.500 -0.400 -0.300 -0.200 -0.100 0.000 0.100 0.200
1-3y 4-6y 7-9y 10-12y 13-15y 16-18y
4.2 Unemployment
From table 1 we see that involuntarily unemployed people are significantly less happy than employed people. Voluntarily unemployed people however, seem to be slightly happier. This effect of voluntary unemployment seems to still hold in table 2 where we take into account adaptation terms. For short term involuntarily unemployed people we see a strong negative impact, while mid-term unemployed people are not significantly less happy then long-term employed people. Long term unemployed people again are significantly less happy. The negative effect of long term unemployment however, is not as large as the effect of short term unemployment. Thus there seems to be a U-shape in the effect of unemployment length on happiness. The length of employment on the other hand does not seem to have an impact on happiness: it does not matter whether someone is employed for up to three months, four to twelve months, or longer than a year.
Table 3 – Ordered probit estimation with prospect variable
Coef. P Vol. unemployed 0.045 0.014 Invol. unemployed 1-4m -0.232 0.007 Invol. unemployed 4-12m -0.061 0.515 Invol. unemployed 12m up -0.094 0.047 Employed 1-3m -0.044 0.197 Employed 4-12m -0.049 0.109 Bad prospect -0.189 0.000
N=26846, p-levels based on clustered standard errors, country dummies, adaptation terms not related to employment and other controls suppressed
The effect of the time length of unemployment is in line with the expectations. People who get unemployed receive a negative shock to happiness, which fades away over time as people habituate to there status of unemployed. Besides this initial effect which fades away,
affects happiness negatively. Furthermore, we see that after including this variable, the coefficients on unemployment got less negative, especially the coefficient on long term unemployment. This coefficient is now only marginally significant indicating that the
negative effect of long term unemployment can be explained for a large part by bad future job prospects.
5. CONCLUDING REMARKS
In this study we analyzed empirically to what extent hedonic adaptation affects happiness when it comes to getting married, divorced, retired, unemployed and getting children. In accordance to the claim that hedonic adaptation does not play a big role in non-pecuniary life events we find that there is no adaptation to marriage or divorce. Retirement and getting children does not seem to affect happiness at all. Contrary to earlier research we do find evidence of adaptation to unemployment. Recently unemployed people seem to be much unhappier than employed people but over the first year of unemployment this negative effect on happiness seems to dissolve entirely. Over the long term the negative effect appears again, which we argue is largely due to deteriorating future prospects.
6. APPENDIX
Table A1 – Descriptive statistics life satisfaction
‘All things considered, I am satisfied with my life now’
country Strongly disagree Disagree Neither agree nor disagree Agree Strongly agree Average
Albania 7% 18% 29% 35% 11% 3.26 Armenia 18% 34% 22% 24% 1% 2.57 Azerbaijan 13% 34% 25% 23% 5% 2.72 Belarus 1% 11% 20% 58% 10% 3.64 Bosnia 20% 25% 27% 25% 3% 2.65 Bulgaria 13% 33% 26% 23% 6% 2.77 Croatia 10% 16% 19% 41% 14% 3.34 Czech Republic 3% 13% 28% 44% 13% 3.51 Estionia 3% 15% 15% 53% 14% 3.60 F.Y.R.O. Macedonia 21% 26% 25% 24% 4% 2.64 Georgia 17% 36% 24% 19% 3% 2.55 Hungary 20% 27% 27% 22% 4% 2.63 Kazakhstan 4% 18% 24% 44% 10% 3.37 Kyrgyzstan 4% 20% 17% 53% 6% 3.37 Latvia 8% 20% 17% 46% 10% 3.30 Lithuania 6% 22% 19% 40% 12% 3.31 Moldova 14% 30% 29% 25% 3% 2.73 Mongolia 6% 21% 34% 32% 7% 3.15 Montenegro 18% 28% 25% 25% 5% 2.72 Poland 6% 17% 25% 42% 11% 3.36 Romania 13% 25% 29% 28% 5% 2.87 Russia 9% 21% 24% 34% 11% 3.17 Serbia 23% 29% 22% 22% 5% 2.58 Slovakia 4% 17% 20% 51% 8% 3.42 Slovenia 1% 8% 19% 54% 18% 3.80 Tajikistan 6% 12% 13% 49% 21% 3.68 Turkey 16% 16% 24% 32% 12% 3.10 Ukraine 11% 27% 24% 32% 6% 2.94 Uzbekistan 2% 10% 15% 57% 16% 3.73 Total 10% 22% 23% 36% 9% 3.12 REFERENCES
Blanchflower, David G. and Andrew J. Oswald (2004). Well-being over time in Britain and the USA. Journal of Public Economics, 88, 1359-1386.
Brickman, Philip and Donald T. Campbell (1971). Hedonic relativism and planning the good society. In: Mortimer H. Appley (ed.), Adaptation-level Theory. New York: Academic Press. Clark, Andrew E., Paul Frijters and Michael A. Shields (2008). Relative Income, Happiness and Utility: An Explanation for the Easterlin Paradox and Other Puzzles. Journal of Economic
Literature, 46:1, 95-144.
Clark, Andrew E., Yannis Georgellis, Richard E. Lucas and Ed Diener (2002). Unemployment Alters the Set-Point for Life Satisfaction. Unpublished.
Di Tella, Rafael, John Haisken-De New and Robert MacCulloch (2007). Happiness
Adaptation to Income and to Status in an Individual panel. NBER Working Paper, no. 13159. Diener Ed and Robert Biswas-Diener (2002). Will money increase subjective well-being?
Social Indicators Research, 57, 119-169.
Easterlin, Richard A. (1974). Does Economic Growth Improve the Human Lot? Some
Empirical Evidence. In: David, Paul A. and Melvin W. Reder (eds.), Nations and Households
in Economic Growth: Essays in Honour of Moses Abramowitz. New York: Academic Press.
Easterlin, Richard A. (1995). Will raising the incomes of all increase the happiness of all?
Journal of Economic Behavior and Organization, 27, 35-47.
Easterlin, Richard A. (2003). Explaining happiness. Proceedings of the National Academy of
Sciences of the United States of America, 100, 11176-11183.
Easterlin, Richard A. (2005). A Puzzle for adaptive theory. Journal of Economic Behavior &
Organization, 56, 513-521.
EBRD (2007). Life in Transition. A survey of people’s experiences and attitudes. Available from: http://www.ebrd.com/pubs/econo/lits.pdf.
Ferrer-i-Carbonell, Ada (2005). Income and Well-Being: An Empirical Analysis of the Comparison Income Effect. Journal of Public Economics, 89, 997-1019.
Frederick, Shane and George F. Loewenstein (1999). Hedonic Adaptation. In: Daniel Kahneman, Ed Diener and Norbert Schwarz. Well-Being: The Foundation of Hedonic
Psychology. New York: Russell Sage Foundation, 302-329.
Helliwell, John F. (2003). How’s life? Combining individual and national variables to explain subjective well-being. Economic Modelling, 20, 331-360.
Johnson, David R. and Jian Wu (2002). An empirical test of crisis, social selection, and role explanations of the relationship between marital disruption and psychological distress: A pooled time-series analysis of four-wave panel data. Journal of Marriage and the Family, 64, 211-224.
Knight, John, and Lina Song (2006). Subjective Well-Being and Its Determinants in Rural China. Unpublished.
Kurdek, Lawrence A. (1999). The Nature and Predictors of the Trajectory of Change in Marital Quality for Husbands and Wives Over the First 10 Years of Marriage. Development
Psychology, 35, 1283-1296.
Lucas, Richard E. (2005). Time Does Not Heal All Wounds, A Longitudinal Study of Reaction and Adaptation to Divorce. Psychological Science, 16, 945-950.
Lucas, Richard E. and Andrew E. Clark (2006). Do People Really Adapt to Marriage?
Journal of Happiness Studies, 7, 405-426.
Lucas, Richard E., Andrew E. Clark, Yannis Georgellis and Ed Diener (2003). Reexamining Adaptation and the Set Point Model of Happiness: Reactions to Changes in Marital Status.
Journal of Personality and Social Psychology, 84, 527-539.
Weinzierl, Matthew (2006). Estimating a Relative Utility Function. Unpublished.
