ArticlePDF Available

Social Exclusion and Well-Being

Authors:
Relative Social Position and Well-Being
Avralt-Od Purevjav Tauhidur Rahman
June 2017
Abstract
One of the most interesting ideas in social science is the notion that individuals are
motivated by concerns about their relative position. Using cross-sectional data from six
transition countries, Kazakhstan, Moldova, Macedonia, Serbia, Tajikistan and Ukraine,
I build on previous studies that have examined the relationship between relative position
and well-being. The main novelty is that various hypotheses are tested: the importance
of own consumption, the contribution of relative consumption, the relevance of social
exclusion, and the marginal contribution of relative social exclusion. Most importantly,
I examine the significance of “reference group” in the relationship between relative con-
sumption and well-being. First, I begin by replicating the previous studies by testing the
hypothesis that self-reported well-being (SWB) depends on relative income, with the dis-
tinction that I use relative consumption, which at a conceptual level affects well-being,
rather than relative income. The result supports the relative income hypothesis. I also
find evidence that relative consumption exerts a positive influence on SWB, a finding
which lends support to Hirschman’s “tunnel effect” conjecture. Second, I test whether
households feel worse off when they are “socially excluded” in their reference group. I
find that, accounting for a household’s own consumption and relative consumption, so-
cially excluded households are associated with lover levels of SWB. Third, I investigate
whether households feel worse off when there is greater degree of social exclusion in
their reference group, which I call the effect of “social solidarity,” where individuals feel
worse off when others around them are socially excluded. I find strong evidence for my
conjecture. Finally, I examine the question of the relevant reference group, i.e. who
belongs to the reference group of each household. Does it include all the households
living in the same region, or district, or settlement, if the reference group is defined
by geographical area? There is suggestive evidence that households compare themselves
with others in their local as well as larger regions.
KEYWORDS: Subjective well-being, social exclusion, happiness, relative position
Avralt-Od Purevjav is a doctoral student in the Charles H. Dyson School of Applied Economics and
Management, Cornell University, ap884@cornell.edu; Tauhidur Rahman is an Associate Professor in the
Department of Agricultural and Resource Economics, The University of Arizona, tauhid@ag.arizona.edu.
1
2
Contents
1 INTRODUCTION 5
2 PREVIOUS STUDIES 8
2.1 IncomeandWell-being.............................. 8
2.2 Relative Income and Well-being . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Relative Social Position and Well-being . . . . . . . . . . . . . . . . . . . . . 11
2.4 Multidimensionality of Well-being . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 GapintheLiterature............................... 13
3 Theory of Relative Concern 15
4 Empirical Strategy 19
4.1 Relative Consumption and Well-being . . . . . . . . . . . . . . . . . . . . . . 19
4.2 Relative Consumption, Social Exclusion, and Well-being . . . . . . . . . . . 19
4.3 Reference Group, and Well-being . . . . . . . . . . . . . . . . . . . . . . . . 20
4.4 Relative Consumption, Relative Social Exclusion, and Well-being . . . . . . 21
4.5 Instrument for Own Consumption . . . . . . . . . . . . . . . . . . . . . . . . 22
5 Data 24
5.1 MeasuringWell-being............................... 24
5.2 Measuring Social Exclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.3 Other Determinants of Well-being . . . . . . . . . . . . . . . . . . . . . . . . 29
6 Results 30
6.1 Relative Consumption and Well-being . . . . . . . . . . . . . . . . . . . . . . 30
6.2 Relative Consumption, Social Exclusion, and Well-being . . . . . . . . . . . 32
6.3 Reference Group and Well-being . . . . . . . . . . . . . . . . . . . . . . . . . 33
6.4 Relative Social Exclusion and Well-being . . . . . . . . . . . . . . . . . . . . 34
7 Conclusions 35
8 REFERENCES 37
List of Figures
1 Distribution of self-reported well-being . . . . . . . . . . . . . . . . . . . . . 44
List of Tables
1 Distribution of respondents by country . . . . . . . . . . . . . . . . . . . . . 43
2 Self-reported well-being across countries . . . . . . . . . . . . . . . . . . . . 44
3
3 Indicators of social exclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4 Summary statistics of indicators of social exclusion across countries . . . . . 46
5 Summary statistics of indicators of social exclusion, by country . . . . . . . . 47
6 Level of social exclusion, by country . . . . . . . . . . . . . . . . . . . . . . . 49
7 Characteristics of the households . . . . . . . . . . . . . . . . . . . . . . . . 50
8 Summary characteristics of the households across countries . . . . . . . . . . 51
9 Summary characteristics of the households, by country . . . . . . . . . . . . 52
10 Relative consumption and well-being across countries . . . . . . . . . . . . . 54
11 Relative consumption and well-being in Kazakhstan . . . . . . . . . . . . . . 55
12 Relative consumption and well-being in Moldova . . . . . . . . . . . . . . . . 56
13 Relative consumption and well-being in Macedonia . . . . . . . . . . . . . . 57
14 Relative consumption and well-being in Serbia . . . . . . . . . . . . . . . . . 58
15 Relative consumption and well-being in Tajikistan . . . . . . . . . . . . . . . 59
16 Relative consumption and well-being in Ukraine . . . . . . . . . . . . . . . . 60
17 Relative consumption, social exclusion, and well-being across countries . . . 61
18 Relative consumption, social exclusion, and well-being in Kazakhstan . . . . 62
19 Relative consumption, social exclusion, and well-being in Moldova . . . . . . 63
20 Relative consumption, social exclusion, and well-being in Macedonia . . . . . 64
21 Relative consumption, social exclusion, and well-being in Serbia . . . . . . . 65
22 Relative consumption, social exclusion, and well-being in Tajikistan . . . . . 66
23 Relative consumption, social exclusion, and well-being in Ukraine . . . . . . 67
24 Relative consumption and well-being across countries: The Role of Reference
Group ....................................... 68
25 Relative consumption and well-being in Kazakhstan: The Role of Reference
Group ....................................... 69
26 Relative consumption and well-being in Moldova: The Role of Reference Group 70
27 Relative consumption and well-being in Macedonia: The Role of Reference
Group ....................................... 71
28 Relative consumption and well-being in Serbia: The Role of Reference Group 72
29 Relative consumption and well-being in Tajikistan: The Role of Reference
Group ....................................... 73
30 Relative consumption and well-being in Ukraine: The Role of Reference Group 74
31 Relative consumption, social exclusion, and well-being across countries: The
RoleofReferenceGroup ............................. 75
32 Relative consumption, social exclusion, and well-being in Kazakhstan: The
RoleofReferenceGroup ............................. 76
33 Relative consumption, social exclusion, and well-being in Moldova: The Role
ofReferenceGroup................................ 77
34 Relative consumption, social exclusion and well-being in Macedonia: The Role
ofReferenceGroup................................ 78
35 Relative consumption, social exclusion, and well-being in Serbia: The Role of
ReferenceGroup ................................. 79
4
36 Relative consumption, social exclusion, and well-being in Tajikistan: The Role
ofReferenceGroup................................ 80
37 Relative consumption, social exclusion, and well-being in Ukraine: The Role
ofReferenceGroup................................ 81
38 Relative social exclusion and well-being across countries . . . . . . . . . . . . 82
39 Relative social exclusion and well-being in Kazakhstan . . . . . . . . . . . . 83
40 Relative social exclusion and well-being in Moldova . . . . . . . . . . . . . . 84
41 Relative social exclusion and well-being in Macedonia . . . . . . . . . . . . . 85
42 Relative social exclusion and well-being in Serbia . . . . . . . . . . . . . . . 86
43 Relative social exclusion and well-being in Tajikistan . . . . . . . . . . . . . 87
44 Relative social exclusion and well-being in Ukraine . . . . . . . . . . . . . . 88
5
1 INTRODUCTION
People are the wealth of a nation1and everyone wants to pursue happiness, which is
generally considered an ultimate goal of life.2According to the neoclassical economists, fully
informed and rational individuals pursue to maximize their expected utilities. However,
according to Mahbub ul Haq, the founder of the Human Development Report, people value
not only consumable goods and services, but also value achievements that do not show up
at all, or not immediately, in income or growth figures. For example, greater access to
knowledge, better nutrition and health services, more secure livelihoods, security against
crime and physical violence, satisfying leisure hours, political and cultural freedoms and
sense of participation in community activities. The activities and abilities that reinforce
human dignity and self-respect - what Adam Smith called the ability to mix with others
without being ‘ashamed to appear in public’.3For example, people value employment not
only because the income derived increases purchasing power, but also because it makes them
feel as worthy members of society.
Then the goal of development is to increase people’s choices such as political freedom,
human rights and self-respect. The absence of public services (e.g. law enforcement) may
increase vulnerabilities and limit people’s choices. Authoritarian regimes can violate political
and civil rights and impose restrictions on people’s freedom to engage in the social, political
and economic life of the community (Sen, 2000). According to Sen (2000), these restrictions
might limit people’s choices and, thus, their well-being. However, everyone does not have the
opportunities to engage in all that people value. Sometimes people find themselves socially
excluded, characterized by their inability to gain the opportunities and resource necessary
to participate in political processes, labor markets, social services, and cultural life activities
of the society in which they live.
Social exclusion has relevance for well-being of individuals. Not only an individual’s
economic and non-economic circumstances determine his well-being, but also his relative
position in the community (e.g. feelings of inferiority, alienation). One of the most interesting
ideas in social science is the notion that individuals are motivated by concerns about their
relative position. In economics, it dates back to classical economists such as Adam Smith
1Human Development Report: Concept and Measurement of Human Development, 1990.
2Aristotle defined happiness as the meaning and the purpose of life, the whole aim and end of human
existence.
3As quoted in Sen (2000).
6
(1759)4, Arthur Pigou (1924)5, and Duesenberry (1949)6. Classical economists view that
in some extent relative position motivates individuals. However, mainstream utility theory
states that individuals derive utility from their own consumption, U(C), rather than from a
combination of own and relative consumption, U(C, C/ ¯
C). This has changed in recent years.
There have been significant amount of empirical works, studying the relationship between
well-being and relative position. For example, by using Canadian data, Tomes (1986) relates
self-reported well-being (SWB) to own income and income in the local community. Also
by using a self-reported measure of relative position, Graham and Pettinato (2002) find
evidence in developing countries that well-being is influenced by relative income concerns.
Using German Socio-Economic Panel, Ferrer-i-Carbonell (2005) find that, controlling for
own income, SWB is decreasing in income of the reference group defined by Education ×
Age ×Region cells where region is East or West Germany. Furthermore, using panel data
for the US, Luttmer (2005) shows that relative income is a factor.
Luttmer (2005) and others (Dorn et al., 2007; Ferrer-i-Carbonell, 2005; Clark and Os-
wald, 1996; Kingdon and Knight, 2007; Knight et al., 2009) extended classical economists’
point of views by including a combination of individual’s own income and relative income
in the determination of well-being. According to them, the average income of the reference
group negatively and significantly predicts different domains of well-being. For example, “if
everybody were to drive expensive cars and living in luxury houses, one would feel unhappy
with a cheaper car and simple house.” However, there are some limitations. First, these
studies examine the influence of relative income rather than relative consumption, which, at
the conceptual level, affects utility or well-being. Second, Luttmer and other scholars con-
sidered only economic dimensions of relative position, and non-economic dimensions were
neglected. People do not compare themselves with others only in terms of income. They
also compare themselves in other aspects of life. There are at least 3 reasons for this: (i)
Recent empirical works point to comparison effects in several life domains: health problems,
labor market status, religiosity, and body shape; (ii) according to psychological studies, up-
ward and downward comparisons are common and involve large number of human domains
and outcomes; (iii) focusing on income comparisons may neglect other channels by which
individuals may improve their command over resources, such as non-cash transfers from the
government, and support from family and friends.
Therefore, in this paper, I build upon previous studies by overcoming these limitations and
extending the scope of analyses that have examined the relationship between relative position
and well-being. The main novelty is that various hypotheses are tested: the importance of
4Adam Smith (1759), for example, wrote: “Nothing is so mortifying as to be obliged to expose our distress
to the view of the public, and to feel, that though our situation is open to the eyes of all mankind, no mortal
conceives for us the half of what we suffer. Nay, it is chiefly from this regard to the sentiments of mankind,
that we pursue riches and avoid poverty.”
5Arthur Pigou (1924) approvingly quotes John Stuart Mill’s observation that “men do not desire to be
rich, but richer than other men.”
6Duesenberry (1949) empirically tested the impact of interdependent preferences on personal consumption
and savings behavior.
7
own consumption, the contribution of relative consumption, the relevance of social exclusion,
and the marginal contribution of relative social exclusion. Most importantly, I examine the
importance “reference group” in the relationship between relative consumption and well-
being.
First, I begin by replicating the previous studies by testing the hypothesis that self-
reported well-being (SWB) depends on relative income, with the distinction that I use rela-
tive consumption, which at a conceptual level affects well-being, rather than relative income.
The result supports the relative income hypothesis. I also find evidence that relative con-
sumption exerts a positive influence on SWB, a finding which lends support to Hirschman’s
“tunnel effect” conjecture.
Second, I test whether households feel worse off when they are “socially excluded” in their
reference group. I find that, accounting for a household’s own consumption and relative
consumption, socially excluded households are associated with lover levels of SWB.
Third, I investigate whether households feel worse off when there is greater degree of
social exclusion in their reference group, which I call the effect of “social solidarity,” where
individuals feel worse off when others around them are socially excluded. I find strong
evidence for my conjecture.
Finally, I examine the question of the relevant reference group, i.e. who belongs to the
reference group of each household. Does it include all the households living in the same
region, or district, or settlement, if the reference group is defined by geographical area?
There is suggestive evidence that households compare themselves with others in their local
as well as larger regions.
The remainder of the paper is organized as follows. In Chapter 2 I briefly review the
past studies on the empirical links between relative position and well-being. In Chapter 3,
theory of relative concern is discussed. I present my empirical strategy in Chapter 4. Data
is discussed in Chapter 5. The results are presented and discussed in Chapter 6. Finally,
concluding remarks are provided in Chapter 7.
8
2 PREVIOUS STUDIES
Numerous scholars have attempted to identify determinants of self-reported well-being in
the existing literature.7Until 1990s, there was not rigorous analysis about the determinants
of well-being. Prior to global measures of well-being such as happiness and life satisfaction,
the empirical literature mostly started by considering job satisfaction, reflecting wages and
labor market (Hamermesh, 1977). Since then, there have been various attempt to determine
well being in terms of economic dimensions focusing on income and well-being and relative
income and well-being. However, lately there is an increasing acceptance that the well being
of an individual is not solely determined by their economic situations but also depends
heavily on the relative social position in non-economic dimensions.
2.1 Income and Well-being
At the beginning of the 1970s, the relation between income and well-being has been one
of the widely discussed and debated topics in the literature on well-being. According to
utility theory, it is assumed that more is better and; therefore, individuals prefer or desire to
increase their income. Most of the findings and insights presented by the well-being studies
examined income and well-being correlation has come to the following conclusions. On the
one side, various researchers provide evidence that countries with higher income have higher
average levels of well-being as analyzed by Diener et al. (1995); Inglehart (1990); and Frey
and Stutzer (2002). According to them, individuals in richer countries, as well as richer
individuals in one country, are slightly happier. The few micro-panel data studies report
a positive effect of income on well-being including van Praag et al. (2003); and Ferrer-i-
Carbonell and Frijters (2004). Moreover, some scholars found evidence that within each
country at a given point in time; richer people are more satisfied with their lives. They
are for example Easterlin (1995, 2001) for the US; Frey and Stutzer (2003) for Switzerland;
MacCulloch et al. (2001) and Blanchflower and Oswald (2004) for the member countries
of the EU. In fact, these findings correlate with the utility theory premise of individuals’
interests in obtaining as higher income as possible.
On the other side, the empirical evidence based on the majority of studies employing cross-
section micro-empirical data finds a low correlation between income and well-being (see, e.g.,
Clark and Oswald (1994) for the UK; and Frey and Stutzer (2000) for Switzerland). Also,
some studies such as Blanchflower and Oswald (2004); Diener and Oishi (2000); Myers and
7For detailed literature reviews, se, e.g., Frey and Stutzer (2002); Clark et al. (2008); and Dolan et al.
(2008). Dolan, Peasgood, and White (2008) thoroughly analyzed up to date research on subjective well-
being. Based on the total of 153 papers on 19 major national and cross-national data sets that included
measures of SWB, they highlighted the following seven interaction effects that have potential influences on
well-being. These are (i) income; (ii) personal characteristics; (iii) socially developed characteristics; (iv) how
people spend time; (v) attitudes and beliefs towards self/others/life; (vi) relationships; (vii) wider economic,
social and political environment.
9
Diener (1995); Kenny (1999); Lane (1998); and Easterlin (1974, 1995) confirm that income
correlates only weakly with individual well-being. Thus, continuous income growth does
not lead to happier individuals. Easterlin (1974, 1995, 2001) reveals that income and self-
reported happiness are correlated positively across individuals within a country. However, he
argues that life satisfaction increases with average incomes but at certain point. According
to him, beyond a certain point the marginal gain in happiness declines. While real income
per capita increases twice the amount, happiness shows virtually no trend (from the General
Social Survey). This phenomenon is referred to as “Easterlin Paradox.” Not only this paradox
was seen in US but also it was apparent in Japan and European countries as well. Some argue
that once individual’s income rises above “poverty line” or the main source of increased well-
being is not income but instead friends and real family life (Lane, 2000). Thus, the studies
imply that income is only partially relevant for well-being.
There are several different reasons and explanations why higher income does not directly
translate into higher happiness. The first explanation of the Easterlin paradox depends on
individual’s perception or human behavior. Individual well-being also depends on the sub-
jective perception of whether one’s income is sufficient to fulfill one’s needs. It means that
the nature of economic competition gives individuals an incentive to make relative compar-
isons. Concern for relative position seems deep rooted in human behavior. According to
Hopkins (2008) “rivalry story,”8“information story,” 9and “perception story”10 might explain
Easterlin paradox.
The second explanation of the Easterlin paradox relies on an adaptation particularly, to
income. It is often argued that individuals get used to their new situations by changing
their expectations (Helson, 1947) so, changes in income might have temporary effects. It
implies that higher incomes correlate to the expectations that lead to “the hedonic treadmill”
(Brickman and Campbell, 1971). Thus, individuals attempt to increase their incomes even
if these bring temporary or small increase in well-being.
In summary, richer individuals in the country are only happier than their unfortunate
individuals. Some scholars assert that income poorly relates with individual well-being so
that constant increase in income does not lead to happier individuals. It can be inferred
based on these evidences that income matters but also other factors may be more significant
in explaining differences in individuals’ well-being. Differences in income can partially ex-
plain the differences in happiness among persons, which indicates that other economic and
8This idea is to attempt to explain such relative concerns are arising from the financial incentives that
arise in tournament-like situations. In other words, if life is a tournament where prizes are awarded to society
winners, it would be logical to seek high social status as noted in Hopkins (2008).
9Samuelson (2004) and Rayo and Becker (2007) explain that the unhappiness is life telling that the person
is following the wrong strategy. If these others are having success, then maybe one should be doing it too.
So this gives an incentive to gather useful information about potentially profitable activities.
10This suggests that relative concerns arise because it is a fundamental to evaluate objects, opportunities
or incomes by means of relative comparisons due to the standard utility theory that assumes preferences are
complete.
10
noneconomic factors exert strong influences beyond the indirect consequences on income
(Frey and Stutzer, 2002).
2.2 Relative Income and Well-being
Perhaps the primary explanation of Easterlin paradox rests on the ways in which income
translates into utility. According to Easterlin (1995), well-being varies directly with one’s
own income and non directly with the incomes of “others.” In economics, the interrelation
among individuals is explained in two ways. First, individuals concern the economic situation
of their peers. Second, the consumption and behavior of individuals affected by decisions of
other individuals in society (Hodgson, 1988). Studies have included relative income as one’s
relative position suggest well-being be strongly influenced by relative positions or individuals’
relative concerns. It is often referred to as the “comparison income” or “relative utility”
effect. The comparison income hypothesis suggests that the decline in relative income mean
a decline in well-being. Several scholars including Luttmer (2005) and others (Dorn et al.,
2007; Ferrer-i-Carbonell, 2005; Clark and Oswald, 1996; Kingdon and Knight, 2007; Knight
et al., 2009) supported this concept. According to them, the average income of the reference
group negatively and significantly predicts different domains of well-being. For example, “if
everybody were to drive expensive cars and living in luxury houses, one would feel unhappy
with a cheaper car and simple house.” Thus, individual’s happiness and welfare depend not
only on the material achievements and income in absolute terms but also on one’s relative
position income wise. Following this line of thought, it is usually assumed that individual
well-being depends on the individual’s personal income as well as on the income of a reference
group. The reference group comprises of members of a community or only a subgroup, such
as individuals living in the same place or having the same education level.11
The first economist to estimate subjective well-being equations using both income of own
and income of others particularly “people like me” (comparison income) was Hamermesh
(1977) and followed by Clark and Oswald (1996); Sloane and Williams (2000); and L´evy-
Garboua and Montmarquette (2004). They estimated coefficients on income and comparison
income in a job satisfaction and found a negative correlation between them. In fact, Clark
and Oswald (1996) found evidence of the negative influence of others’ income on an in-
dividual’s personal job satisfaction. Thus, they analyze the comparison income effect on
job-utility.
Some other scholars claim that there is a negative correlation between an individual’s
own well-being or welfare and others’ incomes. For example, Kapteyn and Van Herwaarden
11Current literature indicates that the reference group has two definitions. One is “people like me” as
defined in van de Stadt et al. (1985), Clark and Oswald (1996) , McBride (2001), Ferrer-i-Carbonell (2005).
The other is “people living in the same region, city, or country” as determined in Persky and Tam (1990),
Easterlin (1995), Blanchflower and Oswald (2004), Luttmer (2005), Graham and Felton (2006), Helliwell
and Huang (2010), Knight et al. (2009).
11
(1980), Kapteyn et al. (1978), Kapteyn et al. (1997), van Praag et al. (1979), and ?present
an empirical analysis of the importance for individuals’ utility of their perception about where
they are in the income distribution. They find that an individual utility depends negatively
on the income of the reference group. They call this phenomenon the “reference drift effect.”
In terms of individual happiness, McBride (2001) presents an empirical analysis of the effect
of an individual’s personal income, past financial situation (whether they were better-off or
worse-off than their own parents) and cohort (reference) income on individuals’ well-being.
His study, as in the present case, is based on self-reported happiness. McBride (2001) finds
a negative correlation between well-being and the average income of the individual’s cohort
and the financial situation of the parents. In a nutshell, the higher the income of the peers,
the less satisfied is the individual.
In addition, by using panel data from the U.S. National Survey of Families and House-
holds, matched with local earnings data from Public Use Micro-data Areas, Luttmer (2005)
explored the effects of inequality on welfare. He stated that there is a negative correlation
with respondents’ life satisfaction, conditional on their own income. His findings highlighted
the importance of relative income differences as people assess the adequacy of their personal
income compared to those around them. In other words, higher earnings of neighbors are
associated with lower levels of self reported financial satisfaction. Following Luttmer, Gra-
ham and Felton (2006) across eighteen Latin American countries; Heliwell and Huang (2005)
in Canada and Knight et al. (2009) in China replicated this finding and suggested that life
satisfaction be relative in income.
These studies have shown that income is not the only factor that people consider when
they compare themselves with others in the community. They also care about their relative
positions in other domains (e.g. social services and participation in civic and social life and
networks). As indicated, Luttmer and other scholars have taken only economic dimensions
(consumption of own and relative consumption) into consideration. However, non-economic
dimensions in terms of social position and relative social position matters in the deterimi-
nation of well-being.
2.3 Relative Social Position and Well-being
In fact, the idea that people compare themselves with others has started earlier. Clas-
sical economists view that in some extent relative position motivates individuals (Luttmer,
2005). According to Fisher, introduction of the consumption of other individuals in indi-
vidual utility is important to analyze. He argued that the purchase of precious stones such
as diamond, for example, depends not only on the diamond itself but also on the position
given to it by society at large (Stigler, 1950). Knight (1922) and Clark (1918) highlighted
the interdependent nature of wants. Later Duesenberry (1949) empirically tested the impact
of interdependent preferences on personal consumption and savings behavior. Leibenstein
(1950) came to the conclusion that other factors related to the consumption of the good
12
(nonfunctional demand)12 might affect one’s satisfaction. Following Duesenberry, other re-
cent studies find that others’ consumption partly drive personal consumption. According to
them, consumption decisions are viewed, as a result, of imitating others and pursuing social
standards. 13
2.4 Multidimensionality of Well-being
It can be inferred based on the interdependence of preferences that individual’s happiness
and satisfaction will depend on what one achieves in comparison with others. Since well-
being of a person is intrinsically multidimensional, it is unlikely, however, that when people
compare themselves with their societal peers rely solely on comparison income information
and disregards other non-economical dimensions of life. Moreover, income is not significant
per se but it should be a measure of an individual command over financial resources. People
care not only about other’s income, but also on their relative social position in a number of
other domains, including economic status, access to social services, and political and cultural
participations. For instance, “if person has a strong preference for social status, then a high-
ranking person in an impoverished country could be happier than a low-ranked person in a
rich country even if they have an equal pay.”
In addition, there is an agreement among economists that income or related measures
of income are substantially insufficient measures of well-being as it captures only economic
domains of the quality of living or individual’s well-being. Historically, well-being of peo-
ple have been measured by the Human Development Index (HDI). However, some scholars
including Rahman et al. (2011) argue that the HDI ignores other major domains such as
contact with family, emotional well-being, work efficiency, safety, and the quality of the envi-
ronment. Their studies suggest that several other factors may effect individual’s well-being
such as (i) contact with family and friends; (ii) emotional well-being; (iii) health; (iv) work
and productive activity; (v) material well-being; (vi) feeling part of one’s local community;
(vii) own safety; (viii) quality of environment.
With the Laeken European Council in December 2001 it was established that, apart from
income, other indicators of quality of life of an individual are necessary to evaluate person’s
well-being. In fact, the shift from the concept of “poverty” to “social exclusion” reflects the
need for a multidimensional approach to study social disadvantage. The evidence from the
existing literature on well-being suggests that determinants or the factors that affect on
individual’s well-being be factors associated with social exclusion. In other words, different
12This concept is described as the distinction between intrinsic value and the subjective value. According
to non-functional demand, individuals consume a good because a large proportion of the community also
uses it. Thus, the good expresses the social belonging of an individual. This is known as the “Bandwagon
effect”.
13Furthermore, interdependence of preferences was also analyzed by, among others, Frank (1985a);
Kapteyn et al. (1978); Holl¨ander (2001); Childers and Rao (1992); Bearden and Etzel (1982); Falk and
Knell (2004); and Frank (1985b)
13
socio-economic factors that have been impacting on the well-beings of an individual are
crucial to identify whether individuals have excluded socially or not.
2.5 Gap in the Literature
Current literature review indicates that relationship between well-being and relative so-
cial position has not studied widely. Most of the available studies explain that relative
position in terms of income matters in the determination of well-being (Clark and Oswald,
1996; Luttmer, 2005; Ferrer-i-Carbonell, 2005; Kingdon and Knight, 2007; Dorn et al., 2007;
Weinzierl, 2005; Knight et al., 2009; Clark et al., 2008). However, besides income people
consider other elements when they compare themselves with others in the community. They
also compare themselves in other aspects of life. There are at least 3 reasons for this.
First, recent empirical works point to comparison effects in several life domains: health
problems (Powdthavee, 2008) and labor market status (Clark et al., 2008), religiosity (Clark
and Lelkes, 2009) and body shape (Clark and Etil´e, 2011).
Second, according to psychological studies, upward and downward comparisons are com-
mon and involve large number of human domains and outcomes (Lyubomirsky, 2001).
Third, focusing on income comparisons may neglect other channels by which individuals
may improve their command over resources, such as non-cash transfers from the government,
and support from family and friends.
In addition, people care about their relative positions in other domains (e.g. social ser-
vices and participation in civic and social life and networks). Several scholars have explicitly
addressed these issues by developing various indices for the measurement of phenomeno
(Chakravarty and D’Ambrosio, 2006; Bossert et al., 2007; D’Ambrosio and Frick, 2007) in
different life domains (Bellani and D’Ambrosio, 2011; Cuesta and Budr´ıa, 2012)14 . However,
it is still unclear how social position in other dimensions of life shape individual well-being.
Thus, the relative position in non-economic dimension is an important in explaining differ-
ences in well-being and independent factor in households’ well-being.
Based on the current studies, it can be concluded that studies focusing on non-income
comparisons are lacking. To my best knowledge, two exceptions are Bellani and D’Ambrosio
(2011) and Cuesta and Budr´ıa (2012). Bellani and D’Ambrosio (2011) noted that well-being
depends on negatively on composite index of non-monetary deprivations. As for Cuesta and
Budria (2012), tried top distinguish their work by disaggregating the impacts of composite
14Among the recent studies, Cuesta and Budr´ıa (2012) made preliminary attempt to estimate the impact
of individual’s deprivation on subjective well being. Their work suggested that policies and practices and
initiatives aimed at improving well-being require a better understanding of individuals’ sensitiveness to
others’ income.
14
index and by driving equivalence scales between income and deprivation in other domains.
However, these two studies still suffer from two limitations. First, like other studies they use
income rather than consumption. Second, these studies are unable to show the contribution
of non-economic deprivations after controlling for both own income and income in the refer-
ence group. For instance, Bellani and D’Ambrosio used mean of country income as reference
income whereas Cuesta and Budria did not control for reference income.
15
3 Theory of Relative Concern
Consider an individual iwith consumption ciwho has utility of the form
U(ci, ci)(1)
where cirepresents the consumptions of others, for example, given a population of nindi-
viduals it would be a vector (c1, . . . , ci1, ci+1, . . . , cn). However, the standard neoclassical
assumption views that an individual’s utility depends only on his or her own consumption.
It is assumed that one’s own consumption cicontributes positively to one’s utility. In
general, it is assumed that the effect of an increase in consumption of others richer than
the individual is negative, which is called "envy" effect. However, there is no consensus
on the effect of changes in consumption of those who are poorer. Some assume that any
improvement for others, who relatively poorer has an adverse effect, which is called "pride"
effect. However, others assume that an improvement for those below you has a positive
effect, which is called "compassion" effect (Friedman and Ostrov, 2008).
Envy effect ∂U (ci, ci)
∂cj
<0for cj> ci
Pride effect ∂U (ci, ci)
∂cj
<0for cj< ci
Compassion effect ∂U (ci, ci)
∂cj
>0for cj< ci
The models of relative concerns with both pride and envy can be divided into two cat-
egories on the basis of functional form: (i) the mean-dependent models, often called the
"keeping up with the Joneses",15 (ii) the rank-based models.16
The first group of models assume that utility is increasing in one’s own absolute consump-
tion cibut there is also a relative component where one’s personal consumption is compared
with the average consumption of others ¯c. For instance:17
U(ci, ci) = U(ci, ci/¯c)(2)
15As Clark and Oswald (1998) point out a model that is mean dependent may not imply a desire to "keep
up" with others.
16These models were pioneered by Layard (1980), Frank (1985b) and Robson (1992).
17This formulation goes back to Duesenberry (1949) and has been used by many authors including Boskin
and Sheshinski (1978), Abel (1990), Gali (1994), Harbaugh (1996), Clark and Oswald (1996, 1998) and
Futagami and Shibata (1998). An alternative formulation of U(ci, ci¯c)is also popular.
16
The second group of models assumes that utility takes the following form:
U(ci, ci) = U(ci, F (ci)) (3)
where ciis one’s own consumption and F(·)is a distribution of consumption F(·). One’s
utility or well-being is increasing in own consumption cibut also in the rank F(ci)one holds
in consumption. This formulation has pride in the sense that, if a group of persons who are
currently richer than you had their consumptions reduced to a level below yours, your rank
and hence your utility would increase.
This form of the utility function can also potentially explain the Easterlin paradox. For a
fixed distribution of consumption, the utility of an individual is increasing in consumption ci.
In this case, both the direct effect U/∂ciand through the effect on rank U/∂F ·f(c), where
f(c)is the density of F(c), are positive. An increase in consumption for a single person,
keeping other consumption constant, raises his rank. Thus, well-being would be increasing
in cross-section. However, when society as a whole becomes richer, the average rank must
remain constant.
In contrast to the above models, it is assumed that individuals have "compassion." The
inequity aversion model of Fehr and Schmidt (1999) is perhaps the best known. It assumes
that utility depends positively on one’s own consumption, but negatively on the difference
between one’s own consumption and that of others. For an individual with consumption ci
comparing herself with n other people with consumption cithis has the simple form
U(ci, ci) = ciα
n1X
cj>ci
(cjci)β
n1X
cj<ci
(cicj)(4)
where αis a weight on the average of consumptions that are above yours and βis a weight
on the average of consumptions below yours.
The model assumes that αβand that βsatisfies 1> β 0. Given αis positive we have
what we called envy, a dislike of others having more. If βis positive, then low consumptions
for others reduce one’s own utility, that is, there is compassion. However, if, contrary to the
assumptions of the model, βwere negative, then we have pride as then lower consumptions
for others raise an individual’s utility. The relation between the inequity aversion model and
a mean-dependent model is as follows by using manipulation of Equation (4).
U(ci, ci) = ciβ(ci¯ci)α+β
n1X
cj>ci
(cjci)(5)
where ¯ciis the average of ci, incomes held by others apart from individual i. One can
see that if βis negative and equal to α, the inequity aversion model of Fehr and Schmidt
(1999) model reduces to a mean-dependent model such as Equation (2). In other words,
17
typical mean-dependent models are the particular case of the model of Fehr and Schmidt
model where pride is as intense as envy, and there is no compassion.
Since its original formulation shown in Equation (4) is linear in own absolute consumption,
it is less successful at explaining the Easterlin paradox.18 Therefore, we simply assume that
the utility is strictly concave rather than linear in own absolute consumption.
U(ci, ci) = u(ci)α
n1X
cj>ci
(cjci)β
n1X
cj<ci
(cicj)(6)
where u(·)is an increasing but strictly concave function. Now, if the level of consumption
is high enough, general increases in consumption will have less than one-for-one effect on
average happiness. Specifically, it is easy to verify that the relative part of the above utility
function (the terms in αand β) is unchanged if everyone’s consumption increases by the
same amount. So, again we have the familiar story that an increase in personal consumption
keeping others’ consumptions constant will have a greater effect than from raising all con-
sumptions. Further, as the own consumption part of the utility function u(·)is concave, the
effect of a general increase in consumption on utility could be quite small if consumptions
are already large. That is; economic growth in rich countries would have a smaller effect on
happiness than a similar increase in incomes in a poor country.
Models of social preferences can imply that an increase in inequality can have a direct
negative effect on individuals who see no change in their own material circumstances. The
Fehr–Schmidt model assumes a utility function in the form of Equation (4). The equivalent
in a large population with consumption distribution F(·)is for the an individual to have
utility
U(ci, ci) = ciαZ
ci
(tci)dF (t)βZci
0
(cit)dF (t) = ci+S(ci, ci)(7)
Further, as Deaton (2003) notes, one can rewrite Equation (7) above as
U(ci, ci) = ciβ(ci¯c)(α+β)Z
ci
(tci)dF (t) = ciβ(ci¯c)(α+β)R(ci)(8)
where again ¯cis an average income and R(ci) = R
ci(tci)dF (t)is the measure of "relative
deprivation” introduced by Yitzhaki (1979).
18Suppose all consumptions are increased by the addition of $c. Then, it is easy to calculate that the
relative component of utility, the terms in αand βin Equation (4), will be unchanged, and consequently
a change in utility will be determined by the first term in the utility function (4) which is simply ci. So,
utility for each would rise by c, the same amount as the increase in consumptions. Thus, in contrast to the
data on happiness, substantial rises in real consumption should lead to substantial increases in happiness.
18
This implies that the Fehr–Schmidt model has the additional property that the utility can
be increasing in the degree of equality. It shows that if there are two distributions F(c)and
G(c)that have the same mean and the same support and if Fis more equal in the sense of
second-order stochastic dominance (equivalently generalised Lorenz dominance) then R(c)
is lower at all consumption levels under Fthan under G. Thus, if as F S assume α > β,
then, even keeping her own consumption constant, an individual will have higher utility in
a more equal society. With the Fehr–Schmidt model, it is possible for utility to fall at every
level of consumption if consumption becomes less equally distributed around an unchanged
mean. In summary, the Fehr–Schmidt model predicts a negative relation between happiness
and inequality at a given level of own consumption. Further, this is something that is not
present in rank based or mean-dependent models of relative concerns introduced earlier.
19
4 Empirical Strategy
4.1 Relative Consumption and Well-being
The first specification we assume is that utility is increasing in one’s own absolute con-
sumption but there is also a relative component where one’s personal consumption is com-
pared with the average consumption of others, which is expressed by the following form:
SW B =f(own comsumption,
average consumption in locality,
controls)(9)
where SW B is self-reported well-being.
A standard assumption in economics is that household consumption (or income) is posi-
tively related to well-being. In cross-section analysis, the income coefficient has been always
found to be positive although not very large. Following the discussion Equations (2) and
(3) the utility or individual well-being function is assumed to be concave in expenditure
and, consequently, consumption and average consumption will be introduced in logarithmic
forms.
The specification assumes that subjective well-being depends on only economic relative
concerns of an individual, which is measured by the average consumption in locality. It
is expected to have a negative correlation with individual well-being. In other words, the
higher the consumption of the reference group, the less satisfied individuals are with their own
expenditure. We define the reference consumption as the average household consumption
of the individuals who belong to the same reference group or those who live in the same
community, which we call it “locality.” The empirical studies on subjective well-being have
found a negative coefficient on the average consumption of the reference group19, which we
expect as well.
Consistently with previous works on this topic, a list of socio-demographic variables will
be used as control variables such as sex, age (age squared), marital status, education, number
of persons living in the household etc.20
4.2 Relative Consumption, Social Exclusion, and Well-being
19See, e.g., Clark and Oswald (1996); Kapteyn and Van Herwaarden (1980); Kapteyn et al. (1997); McBride
(2001); and Luttmer (2005).
20Full list of control variables is presented in the next chapter.
20
The next specification assumes that well-being depends on the non-economic relative
concerns of an individual in addition to economic relative concerns. One’s utility or well-
being is increasing not only in own consumption but also in the social status one holds
in the community. It can be inferred based on the interdependence of preferences that
individual’s well-being and satisfaction will depend on what one achieves in comparison
with others. Since well-being of a person is intrinsically multidimensional, it is unlikely,
however, that people rely solely on comparison income information and disregard other non-
economical dimensions of life when comparing themselves with their societal peers. Income
is not significant per se, but it should be a measure of an individual command over financial
resources. People care not only about other’s income in the community, but also on their
relative social position in a number of other domains, including economic status, access to
social services, and political and cultural participations. If a group of persons has higher
social status (more socially integrated) than you, then your satisfaction with life would be
lower than theirs. The following specification is analyzed to examine if lower social status is
correlated with lower well-being:
SW B =f(own comsumption,
average consumption in locality,
social exclusion,
controls)(10)
A binary indicator on individual’s status of being “socially excluded” is used as a proxy
variable for one’s social status (rank) in the locality, particularly, state of being “socially
disadvantaged” (next chapter discusses the construction of this variable). We expect that
“socially excluded” people are less satisfied with their lives than others, controlling for con-
sumption and other factors. For instance, “if person has a strong preference for social status,
then a high-ranking person in an impoverished country could be happier than a low-ranked
person in a rich country even if they have an equal pay.” It indicates that the lack of opportu-
nities for participation in economic, social and civic processes affects individual’s well-being
negatively.
As discussed in the previous section, the utility of an individual is increasing in con-
sumption, as both through the direct effect and through the effect on social status, where
consumption can have a positive impact on one’s social status. An increase in consumption
for a single person, keeping other factors constant, raises one’s social status. However, not
only economic but also non-economic factors determine invidual’s social status or rank in
the community. So, when society as a whole becomes richer, keeping non-economic factors
constant, everyone’s social rank may not change.
4.3 Reference Group, and Well-being
21
Current literature indicates that the reference group has two definitions. One is “people
like me” as defined in ?, Clark and Oswald (1996), McBride (2001), Ferrer-i-Carbonell (2005).
The other is “people living in the same region, city, or country” as determined in Persky and
Tam (1990), Easterlin (1995), Blanchflower and Oswald (2004), Luttmer (2005), Graham
and Felton (2006), Helliwell and Huang (2010), Knight et al. (2009). For example, Easterlin
(1995) implicitly assumes that individuals compare themselves with all the other citizens of
the same country. Persky and Tam (1990) assume that all individuals living in the same
area are part of the same reference group. Luttmer (2005) also takes a geographic approach
to reference groups, and calculates average income by local area (Public Use Micro Area)
in the U.S. Knight et al. (2009) analyzed cross-sectional information of 9200 households in
China and confirmed that 70 percent of individuals indeed see their village as their reference
group.
Following Luttmer (2005) and Knight et al. (2009), the reference groups are defined as
those living in the same community or neighborhood or local area. The reference group
contains all the individuals living in the (i) same region; (ii) same district; and (iii) same
settlement type (village, a small town, regional or economic center and capital). Then, will
examine the first two specification with alternative definitions of the reference groups.
SW B =f(own comsumption,
average consumption in locality,
where locality is alternatively defined,
controls)(11)
4.4 Relative Consumption, Relative Social Exclusion, and Well-
being
On the basis of Fehr–Schmidt model prediction, the last specification hypothesizes that
there is a negative relation between well-being and inequality at a given level of household
consumption (or income). According to the Fehr–Schmidt model, utility can be increasing
in the degree of equality. In other words, there is a negative relation between well-being and
inequality at a given level of own consumption. Keeping her own consumption constant,
an individual will have higher utility in a more equal society. The following specification
is analysed to investigate the relationship between the incidence and the intensity of social
22
exclusion in locality and well-being.
SW B =f(own comsumption,
average consumption in locality,
social exclusion,
relative social position,
controls)(12)
The adjusted headcount ratio, or Multidimensional Social Exclusion Index (see the next
chapter for the construction of the index) is used to measure one’s relative social position in
the locality. If the majority of people are excluded socially in the locality, then they might
feel sad. The reason they feel sad is that people care not only about their happiness, but
also they wish others to be happy.
4.5 Instrument for Own Consumption
The empirical results suggest positive but provide diminishing returns to consumption
or expenditure. It means higher consumption does not necessarily make people happier.
Instead, it means that happier people earn higher income, e.g., because they might have
a passion for working harder, or they might have tended to spend much and might have
more active social life (Frey and Stutzer, 2002). The empirical results suggest that the
positive relation between well-being and consumption (or income) is likely to be due to
reserve causation21 or unobserved individual characteristics, such as personality factors.22
In order to address the concerns about reserve causation and unobserved individual char-
acteristics driving the result, household consumption was instrumented. The predicted
household consumption is on the respondent’s industry ×occupation composition of the
locality at a point in time if the respondent is employed or based on the household head’s
job status at a point in time if the respondent is unemployed. These variables are not likely
to influence well-being directly, but indirectly through household consumption. To predict
household consumption, we regress log of household consumption on a full set of industry
and occupation dummies interacted with employment dummy, a complete set of job status
dummies of household’s head interacted with employment dummy, and all control variables
(See Table ?? for the results from the first stage estimation).
Rather than actual household consumption, a predicted measure of household consump-
21Some studies show that higher well-being leads to higher incomes in the future (Diener et al., 2002;
Graham et al., 2004; Marks and Fleming, 1999; Schyns, 2001)
22Some studies find a reduced income effect after controlling for personal effects (Ferrer-i-Carbonell and
Frijters, 2004; Luttmer, 2005).
23
tion is used. All specifications will be estimated to address the concern of reserve causation
and unobserved individual characteristics.
24
5 Data
The data on subjective well-being as well as the indicators of the social exclusion are used
from the Social Exclusion Survey (SES).23 The survey generated a data set on the magnitude
and determinants of social exclusion based on the hypothesis that the social exclusion results
from inequalities in access to economic resources, education and employment, as well as in
access to social services, social networks, and political, cultural and civic participation. The
survey was taken in 2009 and carried out in six countries: Kazakhstan, Serbia, the Republic
of Moldova, Tajikistan, the former Yugoslav Republic of Macedonia, and Ukraine.
The unit of observation was the individual. In each country, 2,700 interviews were con-
ducted.24 The sample is a multi-stage random sample divided into 450 clusters and drawn
is representative by age, gender and territorial distribution. First the territory of a country
divided into regions on the basis of the similarity in social, economic, historical, geographical
characteristics and administrative divisions. Then each region was stratified into urban and
rural area. The sampling population was selected in accordance with demographics structure
of the settlement and administrative sectors for the population aged 15 and more. In each
district (sector) number of streets (routes) was randomly selected proportionally to popula-
tion living in these sectors. No more than 6 interviews are allowed in each route. Households
within the selected route were chosen by using random route method.25 Also, the nearest
birthday method26 was used for selecting respondents in each household (Table 1).
5.1 Measuring Well-being
The main dependent variable is self-reported well-being (SWB)27, which is the answer to
23The questionnaire, the raw survey dataset, the technical report, the frequency report and focus group
reports are in the following link: http://europeandcis.undp.org/poverty/socialinclusion. The questionnaire
that was for face-to-face interviews comprised 136 questions reflecting 500 variables. Questionnaire was
identical in all the surveyed countries (adjusted to accommodate the different currencies). It is available
in a number of local languages (Serbian, Macedonian, Albanian, Moldovan, Ukrainian, Kazakh, Tajik and
Russian).
24In Serbia, 3,001 interviews were made (2,401 with members of the general population, plus two boosters
with 300 Roma, and 300 internally displaced persons (IDPs), which was not available).
25Households within the selected primary sampling unit (route) were selected by using random route
method with a statistical level. Interviews were given starting points for each course, and the direction in
which to move. Following the direction, households were selected by pre-determined step factor, according
to the instructions.
26Only one person who qualifies the following criteria is available for an interview. These are (i) age 15
and over, (ii) participation approval, (iii) the closest date of birthday to the date of interview among all the
members of the family, if in the household were more than one person of 15 years and over.
27The analysis of satisfaction with life or happiness also referred to as, sub jective well-being (SWB), is
relatively new but rapidly growing topic for economists. In short, it refers to how people experience and
evaluate their lives and specific domains and activities in their lives (Stone et al., 2014) and it is often used by
psychologists to find out how we think and feel about our lives (Diener et al., 1999). According to Frey and
25
the question: "Are you satisfied or dissatisfied with your standard of living? ”. The answer to
this question takes discrete values from 1 to 5 and will be referred to as subjective well-being.
Respondents answer on a 5-point scale where 1 is defined as “Completely Dissatisfied,” 2 is
defined as “Dissatisfied,” 3 is defined as “Neither satisfied nor dissatisfied,” 4 is defined as
“Satisfied,” and 5 is defined as “Completely Satisfied.”28 Figure (1) shows the distribution
of responses to this question from the main respondents in the whole sample. The average
SWB across individuals and over countries is 3.04 (s.d. = 1.04). Well-being answers are
skewed; individuals tend to be either “neither satisfied nor dissatisfied” or “satisfied” with
their lives, with almost 31.52% of the sample reporting a SWB score below 3 and only 5.8%
reporting 5.
Table 2 shows the distribution of the responses in the six countries. Except Serbia and
Ukraine, more than one-third the population were either “satisfied” or “completely satisfied”
with their standard of living. However, “neither satisfied nor dissatisfied” was the most
frequent satisfaction assessment in Moldova, Macedonia, and Serbia while “satisfied” is the
most frequent in Kazakhstan and Tajikistan. Nevertheless, some considerable differences
between countries can be observed; for example, 10.6% of the Kazakhstanian “completely
satisfied” with their standard of living, whereas this figure is as low as 2.3% for Ukraine and
3.9% for Serbia.
5.2 Measuring Social Exclusion
We employ the social exclusion measure presented in the regional human development
report (Ananiev et al., 2011) that captures social exclusion along various dimensions in a
single, methodologically robust figure.29 The measure has two components. First, the social
exclusion headcount, which is the percentage of people facing a number of deprivations above
a certain threshold (the share of people who are identified “socially excluded”). Second, the
multidimensional social exclusion index, which is the headcount weighted by the intensity of
Stutzer (2003), this approximation permits a direct analysis of what people really value. The measurement
of utility has made great progress based on the extensive work by numerous psychologists (Diener et al.,
1999; Kahnemann et al., 1999; Kahneman and Krueger, 2006). With the help of a single question, or
several questions on global self-reports, it is possible to get indications of individuals’ evaluation of their life
satisfaction or happiness. Economists who have worked with happiness or subjective well-being data all agree
with importance of happiness data in economic analysis for several reasons. Frey and Stutzer (2002) in their
work highlighted several important reasons why happiness is of relevance to economists in terms of providing
information on economic policy decisions, measuring effects of institutional conditions, understanding the
formation of subjective well-being and helping to find solution on some paradoxes. Gruber and Mullainathan
(2005) inferred that by using subjective well-being data economists would be able to directly assess the
impacts of public policy on well-being. According to Kimball and Willis (2006), happiness data provides
significant information about preferences that is appropriate subject for economic policy analysis.
28The variable is reverse coded in the analyses.
29The methodology uses the Alkire and Foster (2011) methodology of multidimensional poverty monitoring
which has been applied to 104 countries in the 2010 UNDP Human Development Report. The measure has
been adapted to account for the diversity of the Europe and Central Asia region.
26
exclusion (the average number of deprivations each socially excluded household experiences).
The deprivations are expressed in terms of 24 indicators – eight indicators for each of
the three dimensions of social exclusion: (i) exclusion from economic life; (ii) social services;
and (iii) civic and social participation (Table 3).30 Each of the three dimensions of social
exclusion has equal weight, as does each indicator.
In the first dimension – economic exclusion – indicators reflect deprivation in incomes and
basic needs; employment, financial services and material assets; amenities that households
need but cannot afford, and dwelling size. Economic exclusion marginalizes individuals in
the distribution of financial resources. From a human development perspective, this hinders
the development of people’s capabilities, which help them to satisfy their needs and exercise
their rights, enabling them to make choices to attain the living standards and quality of life
that they value. Economic exclusion limits people’s access to the labor, financial and housing
markets, as well as to goods and services. It leads not only to income poverty, but also to
reduced access to services such as education, health care and social insurance – ultimately
resulting in a loss of capabilities.
The second dimension – exclusion from social services – encompasses education and health
services, as well as public services.
The third dimension – exclusion from civic and social life — covers deprivation in political,
cultural and social networks, as well as reflects diminished opportunities for social and civic
participation. The indicators chosen for this social exclusion measure are objective: they
indicate status, rather than perceptions.
The identification of “socially excluded person has two steps. First, we need to determine
whether a person is deprived in the single indicator as presented in (Table 3). Being deprived
means households need such items but can not afford them. All deprivation indicators are
dummy variables defined as follows:
dj=(1,if a person is deprived in the individual indicator j
0,otherwise (13)
Second, a person is identified as “socially excluded” by counting the number of indicators
30The indicators were selected based on the analysis of the three dimensions of exclusion, informed by
findings from focus group discussions, from national consultations, as well as from relevant international
literature. For example, indicators have been selected (and in some cases modified) from EU surveys, the
European Quality of Life survey, social capital studies and the ‘missing dimensions of poverty’ surveys
piloted by the Oxford Poverty and Human Development Initiative. Robustness checks have been carried out
to ensure that the individual indicators are not correlated, and that each indicator is relevant for explaining
social exclusion in the six surveyed countries (Ananiev et al., 2011).
27
across the three dimensions in which he or she is deprived.
SEi=(1,if P24
j=1 dj> k
0,otherwise (14)
where SEiis a dummy for a person who is considered “socially excluded” and k= 0,1,2,...,24
is the cut-off value. It is a variable used in the specification 2, 3 and 4. Setting kreflects a
judgement regarding the maximally acceptable multiplicity of deprivations. A person with
a greater multiplicity of deprivations is given higher priority than someone with only one
or two deprivations. According to Alkire and Foster (2011), the choice of k could be a nor-
mative one, reflecting the minimum deprivation count required to be considered “socially
excluded” in a particular context under consideration. The choice of the cut-off value could
also be chosen to reflect specific policy goals and priorities. Taking into account all these
considerations, the cut-off k at the level of 9 was chosen for the social exclusion index, which
is close to 3 deprivations per dimension.31 There are two main reasons behind this choice
of threshold. One is to apply a conservative threshold that does not inflate the multiple
deprivation headcounts. Nine indicators also reflect the minimum number that is necessary
for an individual to be socially excluded in more than one dimension, since one dimension
contains only eight indicators. “Being excluded” means “facing an unacceptable number of
deprivations,” rather than “belonging to a minority isolated from the majority.” Table 4 and
Table 5 contain summary statistics for these indicators.
The social exclusion index is built using the 24 indicators. Three measures were con-
structed: First, the social exclusion headcount ratio (SEHR) is defined as the share of
people who are deprived in at least k indicators for any given k(in this report k= 9). It
indicates the incidence of social exclusion.
SEHRjc =1
njc
njc
X
i
SEi(15)
where SEHRjc is the social exclusion headcount ratio and njc is the number of people in
the area jin the country c.
Second, the average deprivation share (ADS) across the “socially excluded” is calculated as
the average number of deprivations divided by the maximum possible number of deprivations
(24 in our case). The average deprivation share indicates the fraction of possible indicators
in which the average "socially excluded" person endures deprivation. In other words, it
provides additional information on the intensity of social exclusion.
31See Ananiev et al. (2011) for the details of the selection thresholds and robustness checks of indicators
for the social exclusion index.
28
ADSjc =1
24 1
mjc
mjc
X
i 24
X
j
dj!! if SEi= 1 (16)
where ADSjc is average deprivation share of the "socially excluded" and mj c is the number
of the "socially excluded" in the area jin the country c.
Third, the adjusted headcount ratio, or Multidimensional Social Exclusion Index, M SEI ,
are calculated to solve the issue of violation of a ‘strict dimensional monotonicity’. The
adjusted headcount ratio, MSEI combines information on the incidence of social exclusion
and the average intensity of a socially excluded person’s deprivation. As a simple product
of the two partial indices SEHRj c and ADSjc , the measure MSEIjc is sensitive both to the
incidence and the intensity of social exclusion.
MSEIjc =SEHRj c ·ADSjc (17)
Thus, the MSEIjc measure satisfies the property of dimensional monotonicity: if a “socially
excluded” person becomes deprived in an additional indicator, ADSjc rises and so does
MSEIjc . All three multidimensional social exclusion measures were calculated on the basis
of household members.
Table 6 captures social exclusion in terms of headcount, and intensity. It also presents
the multidimensional Social Exclusion Index, which integrates the headcount and intensity
of social exclusion. The data show that more than one out of three persons in the region
is socially excluded, with a wide range of variation across countries. Social exclusion in
Tajikistan is the most acute, with 64.2 percent of the population found to experience nine
or more deprivations. While the share of people considered to be socially excluded varies
significantly across the six countries, the intensity of their exclusion is found to be quite
similar despite variations among countries in terms of population size, GDP and human
development levels. The intensity of social exclusion ranges from 47 percent in Kazakhstan
and Ukraine and (where socially excluded people face on average about 11 deprivations out
of 24) to 49 percent in Moldova, Serbia, and Tajikistan (where socially excluded people face
on average about 12 deprivations out of 24).
The Social Exclusion Index can be disaggregated by dimension, which provides infor-
mation about the contribution of each dimension to the overall social exclusion index. It
creates opportunities for policy makers to see the composition of social exclusion in their
area of interest. The data clearly indicate that economic factors alone do not determine
social exclusion. In five out of six countries all three dimensions contribute roughly equally,
while access to social services contributes slightly more than the other two. It reinforces
the message that in order to tackle social exclusion, all three dimensions must be addressed
29
equally: focusing solely on poverty reduction or economic inclusion is not sufficient.
5.3 Other Determinants of Well-being
The set of variables that influence individual’s subjective well-being has been discussed
in the economic and psychological literature (see, e.g., (Frey and Stutzer, 2002; Dolan et al.,
2008). In the present study, the decision of which variables have to be included is based on
the literature and data availability. A fairly extensive body of literature has found consistent
links between subjective well-being and a number of demographic and socioeconomic vari-
ables. These include income or household consumption, gender, age, age-squared, marital
status, race or ethnicity, education, household size, religion and country- or region-specific
variables. These variables have fairly consistent effects on well-being across societies and
across time in both the developed and developing economies for which there is data and
chosen to make clear comparison with the previous studies.32 Table 7 contain detailed defi-
nitions and Table 8 (for pooled sample of all countries) and Table 9 (by-country sub-samples)
present summary statistics for these control variables including social exclusion index and
as well as the dependent variable, subjective well-being.
32For detailed literature reviews, se, e.g., Frey and Stutzer (2002), Clark et al. (2008), and Dolan et al.
(2008). Dolan et al. (2008) thoroughly analyzed up to date literature on sub jective well-being. Based on the
total of 153 papers on 19 major national and cross-national data sets that included measures of SWB, they
highlighted the following seven interaction effects that have potential influences on well-being: (i) income;
(ii) personal characteristics; (iii) socially developed characteristics; (iv) how people spend time; (v) attitudes
and beliefs towards self/others/life; (vi) relationships; (vii) wider economic, social and political environment.
30
6 Results
6.1 Relative Consumption and Well-being
We first estimate, as a baseline model, Equation (9) including only household consumption
and all the contributing factors, mentioned in the previous section, to make a clear compar-
ison with the specification with average household consumption in locality, which controls
for the relative consumption effect on subjective well-being. All models specifications are
estimated on the pooled sample of six countries (Table 10) and by-country sub-samples (Ta-
ble 10 - 16) with self-reported well-being as the dependent variable using the ordered probit
model.
Table 10 - 16 present ordered probit models estimating the determinants of self-reported
well-being. The first two columns provide results of the necessary specifications containing
the variables that are commonly used in the determinants of self-reported well-being except
for average household consumption in locality whereas Column 3 and 4 provide results of
the specification with average household consumption in locality as a reference consumption.
Columns 1 and 3 include a set of region dummy variables to show the relationships that
hold within regions and to control region-specific fixed effects, which also capture time-
invariant individual characteristics. In column 5 and 6, we show the same specifications
as those in 1 and 3 respectively expect that household consumption is instrumented by
the predicted household consumption to address the concern about reserve causation and
unobserved individual characteristics driving the result. The prediction is based on the
respondent’s industry ×occupation composition of the locality at a point in time if the
respondent is employed or based on the household head’s job status at a point in time if the
respondent is unemployed.
In line with many empirical findings in the literature, the results suggest that well-being
is significantly associated with household consumption as well as household wealth, labor
market status, age, ethnicity, marital status, education, household size, and religion. The
results are fairly robust to model specifications (see Table 10 - 16). The discussion focuses
on the household consumption coefficients. The coefficients of the other variables do not
present surprises for the connoisseur of the subjective well-being literature (Dolan et al.,
2008).
As we expected, the results for the first, most simple, specification, in which only house-
hold consumption and the control variables without average household consumption are
included, is presented in Column 1 and 2 of Table 10. It shows that the household consump-
tion coefficient is significant and positively related to self-reported well-being for the pooled
sample and all 6 by-country sub-samples, i.e. Kazakhstan, Moldova, Macedonia, Serbia,
31
Tajikistan, and Ukraine.33 In the simple ordered probit regressions, both pooled and by
country, the level of household consumption is always significantly positively correlated with
well-being while the effect of it varies by country. This result is in accordance with the usual
findings: namely, that richer individuals are, ceteris paribus, happier than poorer ones.
The individual relative position has a significant and positive impact on well-being, be-
sides the level of household consumption. The relative position is measured by the average
household consumption in the locality, where locality is defined as a settlement. Column
3 and 4 of Table 10 - 11 presents the results for the specification in the form of Equation
(9), in which, besides household consumption, the average consumption in the locality (or
of the reference group) is introduced to the model. The inclusion of the average household
consumption in locality does not change the household consumption coefficient significantly,
suggesting that our previous findings hold for this specification as well. As we expected,
there is always negative, and statistically significant coefficient of average household con-
sumption (see the second row of Table 10). In the by-country regressions, the level of the
average consumption in locality is mostly negatively correlated with well-being, while the ef-
fect and significance of it vary by country (see the second row of Table 11 - 16). For the most
part, both household consumption coefficients are very similar (e.g. Macedonia, Serbia, and
Tajikistan). For the pooled sample of six countries, the coefficient of the average household
consumption in locality is higher than the coefficient of the household’s own consumption.
The results imply that if all individuals of the same neighbors enjoy the same magnitude
increase in their household consumption, then their well-being is unlikely to change. Our
estimated coefficients on the variables of absolute and average household expenditure are
consistent with the idea that well-being is a positive function of consumption and a nega-
tive function of aspirations. For example, aspirations tend to be governed by the standards
and norms of the community or in the locality. This makes a person’s relative position in
the community relevant to his or her well-being. Thus, aspirations adjust to the income
of the community, so creating a “hedonic treadmill” which makes self-reported well-being
insensitive to absolute consumption while being sensitive to relative consumption.
Comparing the results with actual household consumption in Column 1 and 3 of Table
10 and the results with the predicted household consumption in Column 5 and 6 of the
same table, the estimated coefficients on the variables are fairly similar except for that
on the household consumption variables. The coefficient on actual household consumption
was consistently positive and significant, ranging from 0.157 to 0.177. With the predicted
household consumption, the coefficient is always significant and positive, ranging from 0.458
to 0.505. Instrumenting yields an estimated effect of predicted household consumption on
self-reported well-being that is about three times as large as the estimate in the specification
with actual household consumption. It is expected that the coefficient on actual household
consumption to be biased upwards. In other words, happier people earn a higher income
because they like to work harder and are more productive. Also, unobserved characteristics
33See the first rows of Table 11 - 16 for by-country results.
32
such as active social networks (like spending time with friends or colleagues) or ability might
raise both household consumption and well-being. However, the higher impact of predicted
household consumption suggests that there are some characteristics which increase household
consumption, but decrease subjective well-being, or that instrumenting corrects for an error
in the measurement of household consumption, so reducing downward attenuation bias. It
is notable that this finding of downward bias is not unique (Luttmer, 2005; Knight et al.,
2009).
6.2 Relative Consumption, Social Exclusion, and Well-being
The next specification (Equation 9) hypothesizes that self-reported well-being depends on
the non-economic relative concerns of an individual in addition to economic relative concerns.
Particularly, it is aimed to empirically investigate the independent role of social exclusion
(or status of being “socially excluded”) in the determination of self-reported well-being.
Table 17 - 23 explore the relationship between the state of being “socially excluded”
and self-reported well-being and present the results of ordered probit models estimating
the determinants of self-reported well-being with the introduction of the variable “socially
excluded.” The organization of the tables is the same as the previous table. Column 1, 2,
and 5 replicate the baseline regression in the form of Equation (9) (same as in Column 3, 4
and 6 of Table 10 - 11). Other columns show regressions that are identical to the baseline
regression, except that the dummy variable for the status of being “socially excluded” are
added as a control variable for relative position in non-economic dimensions.
More relevant for our research question is a comparison between the specifications with
just relative consumption (Equation (9)) and the specification with not only relative con-
sumption, but also social exclusion (Equation (10)), where an alternative measure of relative
position is considered. If the hypothesis that individual’s relative position measured by only
consumption (or income) and the one proxied by status of being “socially excluded” (the
multidimensional concepts of deprivation) are not capturing the same phenomenon holds,
we would expect that even after the introduction of a variable for non-economic relative
position as a control, both household consumption and average household consumption in
locality remain significantly correlated with self-reported well-being. As shown in Column 3,
4, and 6 of Table 17 - 23, in addition to relative consumption (household consumption and
average household expenditure in locality), being “socially excluded” has a highly significant
and negative impact on self-reported well-being. The result is strongly robust to all specifica-
tions and it suggests that controlling for other factors including household consumption and
economic relative position, “socially excluded” people are less satisfied with their lives than
those who are not. In other words, having lack of opportunities for participation in economic,
social and civic processes strongly affects individual’s well-being. It supports the idea that
economic positions do not determine people’s comparison to neighbors, colleagues, or more
to a reference group but also by non-economic areas such as exclusion from social services
33
and civic and social life. Also, it reinforces the idea of the importance of incorporating the
multi-dimensionality in measuring individual well-being.
After comparisin of the results from the specification in the form of Equation (9) with the
results from the specification in the form of Equation (10), the estimated coefficients show
small differences on the variables. The only exception was for few variables including house-
hold consumption, unemployed, not in the labor force and education. Since unemployment
is one of the deprivation indicators, being unemployed has both direct and indirect impact
on individual’s well-being. Moreover, the impact on well-being of being “socially excluded”
is higher than on those who are unemployed, which suggests that being unemployed leads
not only to income poverty, but also to reduced access to services such as education, health
care and social insurance – ultimately resulting in a loss of capabilities.
Substituting the actual household consumption with the predicted household expenditure
hardly change the effect of being “socially excluded” on self-reported well- being. The results
show that persistence in the state of being “socially excluded” are negatively associated with
self-reported well-being. It has greater effects than any other determinants. Interestingly,
the impact of average household consumption is weakened by the introduction of the social
exclusion variable. It implies that relative consumption does not seem to have a strong
association with self-reported well-being when one’s social status measure based on the
multidimensional concepts of deprivation indicators are introduced as a control variable.
6.3 Reference Group and Well-being
We evaluated the impacts of relative consumption and social exclusion in three different
specifications, which vary in terms of the reference group. Particularly, average household
consumption is calculated in three different localities: (i) region, (ii) district, and (ii) settle-
ment type. The results are shown in Table 24 - 37.
Surprisingly, the impact of average household consumption in the locality on self-reported
well-being increases in magnitude as the scope of the locality decreases. When the reference
group is defined as all individuals in the same area, the average household consumption of the
region has a negative impact on the self-reported well-beings of those who live in that region,
but the result is often not statistically significant. However, the impact is highly significant
and negative when the locality is defined as district or settlement. There is a slightly higher
impact for settlement than for district, which suggests that relative household expenditure
within settlement (village, or small town, or regional center, or capital) might be more
important for individuals when they make comparison with others. For the pooled sample,
the point estimates indicate that the self-reported well-being of an individual declines faster
with an increase in average consumption in settlement or district than with an increase in
average consumption in the region (Table 24 and 31). This result is consistent with the
findings that the reference group is likely to be determined by information sets and by social
34
interactions and most rural people confine their reference groups to the village: their orbits
of comparison are narrow (Knight et al., 2009).
6.4 Relative Social Exclusion and Well-being
Status of being “socially excluded” is not purely relative concepts because one’s being
“social excluded” does not affect other’s status of being “socially excluded.” Instead, it can
be interpreted as that an individual “facing an unacceptable number of deprivations” or “not
having things that others have” or “not being part of society – not socially integrated” or
“socially disadvantaged.” In this sense, if someone is socially disadvantaged, it does not have
to influence others’ well-being directly. However, there can be an indirect effect through the
head count of those who are “socially excluded” in the community (incidence) or through
the intensity of social exclusion.
As discussed in the data section, the adjusted headcount ratio, (or Multidimensional So-
cial Exclusion Index), combines information on the incidence of social exclusion and the
average intensity of “socially excluded” person’s deprivation. This measure is sensitive both
to the incidence and the intensity of social exclusion. If a “socially excluded” person be-
comes deprived in an additional indicator, intensity of social exclusion rises and so does the
adjusted headcount ratio. This index is a proxy for relative social position (or inequality )
in the community. The prediction derived from the Fehr–Schmidt model hypothesizes that
there is a negative relation between well-being and inequality at a given level of household
consumption. We analyzed hypothesis from Table 38 through 44. Column 1 and 2 replicate
the baseline regression in the form of Equation (9) and the specification with social exclu-
sion respectively and Column 3 show regressions that are identical to Equation (10), except
that the multidimensional social exclusion index is a proxy for relative social position in the
locality.
The estimated coefficients on relative social position are consistently negative and statis-
tically significant. The result that individual’s well-being will be decreasing with the degree
of the incidence and the intensity of social exclusion. In other words, there is a negative
relation between well-being and inequality. It can be measured by the multidimensional
social exclusion index. Control for one’s consumption, relative consumption, and social sta-
tus, an individual will have lower utility in the community, where many people are “socially
excluded” or those who are “socially disadvantaged.”
35
7 Conclusions
This this presents an empirical test of hypotheses about the importance of relative posi-
tion in terms of non-economic dimension in the determination of well-being. The empirical
analysis has taken national representative household survey data on Social Exclusion to ex-
plore the relationship between the relative position and well-being in six transition countries:
Kazakhstan, Moldova, Macedonia, Serbia, Tajikistan, and Ukraine. The estimation results
confirm previous research by showing that not only relative position in the economic dimen-
sion but also the relative position in non-economic dimension is strongly significant in the
determination of well-being.
However, the relevance of the present study lies in four features. First, it contributes to the
current literature that the relative position in non-economic dimension is an important and
independent factor in households’ well-being. Rather than taking only economic dimension
into consideration like most of the previous scholars did, we took non-economic dimensions
into consideration. We identify whether an individual is being “socially excluded” or not as
a proxy measure for one’s social status in the locality or a measure of “socially disadvan-
taged” person. Our contribution has been to provide a new outlook of the determination
of well-being. Second, it differs from other studies, as it investigates the determination
of self-reported well-being in six transition countries since most of the self-reported well-
being studies were applied to developed countries and neglected transition and developing
countries. Third, we tested the impact of alternatively defined reference groups/localities
on self-reported well-being. Finally, we measure household’s relative social position in the
community by constructing a multidimensional index of social exclusion.
The main conclusions are as follows: (i) The paper finds that well-being is significantly
associated with household consumption as well as the other control variables including house-
hold wealth, labor market status, age, ethnicity, marital status, education, household size,
and religion. The study provides strong evidence that the richer family is, keeping others con-
stant, happier than poorer ones. In other words, the household consumption is strongly and
positively associated with self-reported well-being in the surveyed countries: Kazakhstan,
Moldova, Macedonia, Serbia, Tajikistan, and Ukraine. However, the results imply that if
all individuals of the same neighbors enjoy the same magnitude increase in their household
expenditure, then their well-being is not expected to change; (ii) A person’s status of being
“socially disadvantaged” or “socially excluded” is critical and independent factor in his or her
well-being. The finding is very robust to all specifications, which suggests that controlling
for other factors including household consumption and economic relative position, those who
are “socially excluded” are less satisfied with their lives than those who are not. People’s
comparison to the reference group is not only determined by their economic positions but also
by non-economic positions. It also reinforces the idea of the importance of incorporating the
multi-dimensionality in measuring individual well-being; (iii) The result shows that people
likely to compare themselves with their immediate neighbors. Smaller the reference groups
36
are, higher the relative comparisons are. (iv) Having lack of opportunities for participation
in economic, social and civic processes strongly affect individual’s well-being both directly
and indirectly.
37
8 REFERENCES
Abel, A. B. (1990). Asset prices under habit formation and catching up with the joneses.
The American Economic Review, pages 38–42.
Alkire, S. and Foster, J. (2011). Counting and multidimensional poverty measurement.
Journal of Public Economics, 95(7):476–487.
Ananiev, J., Atanasov, P., Gerovska-Mitev, M., and Shukarov, M. (2011). Beyond transition
towards inclusive societies.
Bearden, W. O. and Etzel, M. J. (1982). Reference group influence on product and brand
purchase decisions. Journal of Consumer research, pages 183–194.
Bellani, L. and D’Ambrosio, C. (2011). Deprivation, social exclusion and subjective well-
being. Social Indicators Research, 104(1):67–86.
Blanchflower, D. G. and Oswald, A. J. (2004). Well-being over time in britain and the usa.
Journal of Public Economics, 88(7-8):1359–1386.
Boskin, M. J. and Sheshinski, E. (1978). Optimal redistributive taxation when individual
welfare depends upon relative income. The Quarterly Journal of Economics, pages 589–
601.
Bossert, W., D’Ambrosio, C., and Peragine, V. (2007). Deprivation and social exclusion.
Economica, 74(296):777–803.
Brickman, P. and Campbell, D. T. (1971). Hedonic relativism and planning the good so-
ciety. In Fagerberg, J., Mowery, D., and Nelson, R., editors, Adaptation-level Theory: A
Symposium, pages 287–302. Academic Press, New York.
Chakravarty, S. R. and D’Ambrosio, C. (2006). The measurement of social exclusion. Review
of income and wealth, 52(3):377–398.
Childers, T. L. and Rao, A. R. (1992). The influence of familial and peer-based reference
groups on consumer decisions. Journal of Consumer research, 19(2):198.
Clark, A. and Lelkes, O. (2009). Let us pray: Religious interactions in life satisfaction. paris
school of economics. Technical report, Working Paper 2009-01.
Clark, A. E. and Etil´e, F. (2011). Happy house: Spousal weight and individual well-being.
Journal of Health Economics, 30(5):1124–1136.
Clark, A. E., Frijters, P., and Shields, M. A. (2008). Relative income, happiness, and
utility: An explanation for the easterlin paradox and other puzzles. Journal of Economic
Literature, 46(1):pp. 95–144.
38
Clark, A. E. and Oswald, A. J. (1994). Unhappiness and unemployment. Economic Journal,
104(424):648–59.
Clark, A. E. and Oswald, A. J. (1996). Satisfaction and comparison income. Journal of
Public Economics, 61(3):359–381.
Clark, A. E. and Oswald, A. J. (1998). Comparison-concave utility and following behaviour
in social and economic settings. Journal of Public Economics, 70(1):133–155.
Clark, J. M. (1918). Economics and modern psycholoy: I. The Journal of Political Economy,
pages 1–30.
Cuesta, M. B. and Budr´ıa, S. (2012). Deprivation and subjective well-being: Evidence from
panel data. Review of Income and Wealth.
D’Ambrosio, C. and Frick, J. (2007). Income satisfaction and relative deprivation: An
empirical link. Social Indicators Research, 81(3):497–519.
Deaton, A. (2003). Health, inequality, and economic development. Journal of Economic
Literature, 41(1):113–158.
Diener, E., Diener, M., and Diener, C. (1995). Factors predicting the subjective well-being
of nations. Journal of Personality and Social Psychology, 69(5):851 – 864.
Diener, E., Lucas, R. E., Oishi, S., and Suh, E. M. (2002). Looking up and looking down:
Weighting good and bad information in life satisfaction judgments. Personality and Social
Psychology Bulletin, 28(4):437–445.
Diener, E. and Oishi, S. (2000). Money and happiness: Income and subjective well-being
across nations. Culture and subjective well-being, pages 185–218.
Diener, E., Suh, E. M., Lucas, R. E., and Smith, H. L. (1999). Subjective well-being: Three
decades of progress. Psychological bulletin, 125(2):276.
Dolan, P., Peasgood, T., and White, M. (2008). Do we really know what makes us happy?
a review of the economic literature on the factors associated with subjective well-being.
Journal of Economic Psychology, 29(1):94 – 122.
Dorn, D., Fischer, J., Kirchg¨assner, G., and Sousa-Poza, A. (2007). Is it culture or democ-
racy? the impact of democracy and culture on happiness. Social Indicators Research,
82(3):505–526.
Duesenberry, J. S. (1949). Income, saving, and the theory of consumer behavior.
Easterlin, R. A. (1974). Does economic growth improve the human lot? some empirical
evidence. Nations and households in economic growth, 89.
39
Easterlin, R. A. (1995). Will raising the incomes of all increase the happiness of all? Journal
of Economic Behavior &amp; Organization, 27(1):35–47.
Easterlin, R. A. (2001). Income and happiness: Towards an unified theory. Economic
Journal, 111(473):465–84.
Falk, A. and Knell, M. (2004). Choosing the joneses: Endogenous goals and reference
standards. The Scandinavian Journal of Economics, 106(3):417–435.
Fehr, E. and Schmidt, K. M. (1999). A Theory Of Fairness, Competition, And Cooperation.
The Quarterly Journal of Economics, 114(3):817–868.
Ferrer-i-Carbonell, A. (2005). Income and well-being: an empirical analysis of the comparison
income effect. Journal of Public Economics, 89(5-6):997–1019.
Ferrer-i-Carbonell, A. and Frijters, P. (2004). How important is methodology for the esti-
mates of the determinants of happiness? Economic Journal, 114(497):641–659.
Frank, R. H. (1985a). Choosing the right pond: Human behavior and the quest for status.
Oxford University Press.
Frank, R. H. (1985b). The Demand for Unobservable and Other Nonpositional Goods.
American Economic Review, 75(1):101–16.
Frey, B. S. and Stutzer, A. (2000). Happiness, economy and institutions. Economic Journal,
110(466):918–38.
Frey, B. S. and Stutzer, A. (2002). What can economists learn from happiness research?
Journal of Economic Literature, 40(2):402–435.
Frey, B. S. and Stutzer, A. (2003). Testing Theories of Happiness. IEW - Working Papers
147, Institute for Empirical Research in Economics - University of Zurich.
Friedman, D. and Ostrov, D. N. (2008). Conspicuous consumption dynamics. Games and
Economic Behavior, 64(1):121–145.
Futagami, K. and Shibata, A. (1998). Keeping one step ahead of the joneses: status, the dis-
tribution of wealth, and long run growth. Journal of Economic Behavior & Organization,
36(1):109–126.
Gali, J. (1994). Keeping up with the joneses: Consumption externalities, portfolio choice,
and asset prices. Journal of Money, Credit and Banking, pages 1–8.
Graham, C., Eggers, A., and Sukhtankar, S. (2004). Does happiness pay?: An exploration
based on panel data from russia. Journal of Economic Behavior &amp; Organization,
55(3):319–342.
40
Graham, C. and Felton, A. (2006). Inequality and happiness: Insights from latin america.
Journal of Economic Inequality, 4(1):107–122.
Graham, C. and Pettinato, S. (2002). Frustrated achievers: Winners, losers and subjective
well-being in new market economies. The Journal of Development Studies, 38(4):100–140.
Gruber, J. H. and Mullainathan, S. (2005). Do cigarette taxes make smokers happier. The
B.E. Journal of Economic Analysis &amp; Policy, 5(1):1–45.
Hamermesh, D. S. (1977). Economic aspects of job satisfaction. Essays in Labor Market
Analysis, pages 53–72.
Harbaugh, R. (1996). Falling behind the joneses: relative consumption and the growth-
savings paradox. Economics Letters, 53(3):297–304.
Heliwell, J. F. and Huang, H. (2005). How’s the job? well-being and social capital in the
workplace. Technical report, National Bureau of Economic Research.
Helliwell, J. F. and Huang, H. (2010). How’s the job? well-being and social capital in the
workplace. Industrial & Labor Relations Review, 63(2):205–227.
Helson, H. (1947). Adaptation-level as frame of reference for prediction of psychophysical
data. The American journal of psychology, pages 1–29.
Hodgson, G. M. (1988). Economics and institutions. In Journal of Economic Issues. Citeseer.
Holl¨ander, H. (2001). On the validity of utility statements: standard theory versus duesen-
berry’s. Journal of Economic Behavior & Organization, 45(3):227–249.
Hopkins, E. (2008). Inequality, happiness and relative concerns: What actually is their
relationship? The Journal of Economic Inequality, 6(4):351–372.
Inglehart, R. (1990). Culture shift in advanced industrial society. Princeton University Press.
Kahneman, D. and Krueger, A. B. (2006). Developments in the measurement of subjective
well-being. The journal of economic perspectives, 20(1):3–24.
Kahnemann, D., Diener, E., and Schwarz, N. (1999). Well-Being: The Foundations of
Hedonic Psychology. Russell Sage Foundation.
Kapteyn, A., Van de Geer, S., Van de Stadt, H., and Wansbeek, T. (1997). Interdependent
preferences: an econometric analysis. Journal of Applied Econometrics, 12(6):665–686.
Kapteyn, A. and Van Herwaarden, F. G. (1980). Interdependent welfare functions and
optimal income distribution. Journal of Public Economics, 14(3):375–397.
Kapteyn, A., Van Praag, B., and Van Herwaarden, F. G. (1978). Individual welfare functions
and social reference spaces. Economics Letters, 1(2):173–177.
41
Kenny, C. (1999). Does growth cause happiness, or does happiness cause growth? Kyklos,
52(1):3–25.
Kimball, M. and Willis, R. (2006). Utility and happiness.
Kingdon, G. G. and Knight, J. (2007). Community, comparisons and subjective well-being
in a divided society. Journal of Economic Behavior & Organization, 64(1):69–90.
Knight, F. H. (1922). Ethics and the economic interpretation. The Quarterly Journal of
Economics, pages 454–481.
Knight, J., Song, L., and Gunatilaka, R. (2009). Subjective well-being and its determinants
in rural china. China Economic Review, 20(4):635–649.
Lane, R. E. (1998). The joyless market economy. Economics, values, and organization, 1:998.
Lane, R. E. (2000). The loss of happiness in market democracies. Yale University Press.
Layard, R. (1980). Human satisfactions and public policy. The Economic Journal, pages
737–750.
Leibenstein, H. (1950). Bandwagon, snob, and veblen effects in the theory of consumers’
demand. The Quarterly Journal of Economics, 64(2):183–207.
evy-Garboua, L. and Montmarquette, C. (2004). Reported job satisfaction: what does it
mean? The Journal of Socio-Economics, 33(2):135–151.
Luttmer, E. F. P. (2005). Neighbors as negatives: Relative earnings and well-being. The
Quarterly Journal of Economics, 120(3):963–1002.
Lyubomirsky, S. (2001). Why are some people happier than others? the role of cognitive
and motivational processes in well-being. American Psychologist, 56(3):239.
MacCulloch, R. J., Tella, R. D., and Oswald, A. J. (2001). Preferences over inflation and un-
employment: Evidence from surveys of happiness. American Economic Review, 91(1):335–
341.
Marks, G. and Fleming, N. (1999). Influences and consequences of well-being among aus-
tralian young people: 1980–1995. Social Indicators Research, 46(3):301–323.
McBride, M. (2001). Relative-income effects on subjective well-being in the cross-section.
Journal of Economic Behavior &amp; Organization, 45(3):251–278.
Myers, D. G. and Diener, E. (1995). Who is happy? Psychological Science, 6(1):10–19.
Persky, J. and Tam, M.-Y. (1990). Local status and national social welfare. Journal of
Regional Science, 30(2):229–238.
Pigou, A. C. (1924). The economics of welfare. Transaction Publishers.
42
Powdthavee, N. (2008). Putting a price tag on friends, relatives, and neighbours: Using
surveys of life satisfaction to value social relationships. The Journal of Socio-Economics,
37(4):1459–1480.
Rahman, T., Mittelhammer, R. C., and Wandschneider, P. R. (2011). Measuring quality of
life across countries: A multiple indicators and multiple causes approach. The Journal of
Socio-Economics, 40(1):43 – 52.
Rayo, L. and Becker, G. S. (2007). Evolutionary efficiency and happiness. Journal of Political
Economy, 115(2):302–337.
Robson, A. (1992). Status, the distribution of wealth, private and social attitudes to risk.
Econometrica, 60(4):837–857.
Samuelson, L. (2004). Information-based relative consumption effects. Econometrica,
72(1):93–118.
Schyns, P. (2001). Income and satisfaction in russia. Journal of Happiness Studies, 2(2):173–
204.
Sen, A. K. (2000). Social exclusion: Concept, application, and scrutiny. Number 1. Office of
Environment and Social Development, Asian Development Bank Manila.
Sloane, P. J. and Williams, H. (2000). Job satisfaction, comparison earnings, and gender.
Labour, 14(3):473–502.
Smith, A. (1759). The theory ofmoral sentiments. New York: Garland, 1971.
Stigler, G. J. (1950). The development of utility theory. ii. The Journal of Political Economy,
58:373–396.
Stone, A. A., Mackie, C., et al. (2014). Subjective well-being: Measuring happiness, suffering,
and other dimensions of experience. National Academies Press.
Tomes, N. (1986). Income distribution, happiness and satisfaction: A direct test of the
interdependent preferences model. Journal of Economic Psychology, 7(4):425–446.
van de Stadt, H., Kapteyn, A., and van de Geer, S. (1985). The relativity of utility: Evidence
from panel data. The Review of Economics and Statistics, 67(2):179–87.
van Praag, B. M., Kapteyn, A., and Van Herwaarden, F. (1979). The definition and mea-
surement of social reference spaces. The Netherlands’ Journal of Sociology, 15:13–25.
van Praag, B. M. S., Frijters, P., and Ferrer-i-Carbonell, A. (2003). The anatomy of subjec-
tive well-being. Journal of Economic Behavior &amp; Organization, 51(1):29–49.
Weinzierl, M. (2005). Estimating a relative utility function. Harvard University.
Yitzhaki, S. (1979). Relative deprivation and the gini coefficient. The Quarterly Journal of
Economics, 93(2):321–24.
43
Table 1: Distribution of respondents by country*
KZH MLD FYRM SER TJK UKR Total
Male 1,316 1,145 1,284 1,209 1,164 1,196 7,314
% -49% -42% 48% 50% 43% 44% 46%
Female 1,384 1,555 1,416 1,192 1,536 1,504 8,587
% -51% -58% 52% 50% 57% 56% 54%
Total 2,700 2,700 2,700 2,401 2,700 2,700 15,901
Number of Region 14 4 8 6 5 11 48
Average number of respondents
in each region 193 675 338 450 540 245 331
Number of District 161 33 41 25 69 26 355
Average number of respondents
in each district 17 82 66 108 39 104 45
Number of Settlement 302 116 119 300 348 110 1295
Average number of respondents
in each settlement 9 23 23 9 8 25 12
Number of Settlement Types ** 4 4 4 4 4 4 4
Average number of respondents
in each settlement type 40 8 10 6 17 7 11
Number of Cluster 450 450 485 400 450 450 2685
Average number of respondents
in each cluster 6 6 6 6 6 6 6
*KZH=Kazakhstan, MLD=Moldova, FYRM=Macedonia, SER=Serbia, TJK=Tajikistan, and
UKR=Ukraine.
** There are four different settlements: village, small towns, regional or economic center, and capital.
44
Figure 1: Distribution of self-reported well-being
Table 2: Self-reported well-being across countries
Country Completely Dissatisfied*Neither Satisfied*Completely Mean S.D. N
dissatisfied*satisfied nor Satisfied*
dissatisfied*
Kazakhstan 4.8 22.2 26.4 38.3 8.3 3.23 1.04 2689
Moldova 6.7 22.7 36.3 29.2 5.2 3.03 1.00 2696
Macedonia 6.6 22.5 36.4 30.4 4.1 3.03 0.98 2679
Serbia 12.6 30.0 31.3 22.2 3.9 2.75 1.06 2393
Tajikistan 3.2 12.0 29.9 44.3 10.6 3.47 0.95 2684
Ukraine 13.0 34.1 28.5 22.1 2.3 2.67 1.03 2684
TOTAL 7.7 23.8 31.5 31.3 5.8 3.04 1.04 15825
*All figures are in percentage
45
Table 3: Indicators of social exclusion
Variable Definition
Economic Exclusion
At risk of income poverty 1 if per-capita household expenditure is less than 60 percent of median
equivalent expenditures in the country, 0 otherwise (inequality).
Unmet basic needs 1 if in the past 12 months the household has not been able to afford three
meals a day, or pay bills regularly, or keep the home adequately warm, or
buy new clothes and shoes, 0 otherwise.
Unemployment 1 if person is unemployed or a discouraged worker,
0 otherwise.
Financial services 1 if person has a lack of access to a bank account on one’s own name, 0
otherwise.
Housing 1 if household cannot afford a bed for every member of the household, 0
otherwise (material deprivation) .
Amenities 1 if household needs a washing machine, freezer or microwave but cannot
afford one, 0 otherwise (material deprivation).
ICT 1 if household needs a computer or internet but cannot afford one, 0 other-
wise (material deprivation).
Overcrowding 1 if household’s accommodation is less than 6m2 per person, 0 otherwise
(material deprivation).
Exclusion from Social Services
Water 1 if household has no running water or sewerage system, 0 otherwise (public
utilities).
Heating 1 if household heats with wood or with no heating device, 0 otherwise (public
utilities).
Low education 1 if person has low educational achievements (basic schooling) or early school
leavers, 0 otherwise (education).
School materials 1 if household could not afford to buy school materials for every child in the
past 12 months, 0 otherwise (education).
School drop out 1 if household has young children not in school or pre-school, 0 otherwise
(education).
Medication 1 if household could not afford medication or dental checks for every child
in the past 12 months, 0 otherwise (health care).
Health care 1 if household’s medical needs not being met by the health care system, 0
otherwise (health care) .
Transportation 1 if person has lack of opportunities to attend events due to distance, 0
otherwise (lack of transportation or social infrastructure).
Exclusion from Participation in Civic & Social life and Networks
Social ties with family 1 if person has rare or infrequent social contact with family or relatives, 0
otherwise (social capital).
Social ties with friends 1 if person has rare social contact with friends, 0 otherwise (social capital).
Support network 1 if person has lack of support networks that could help in the event of
emergency, 0 otherwise (social capital).
Social participation (pri-
vate)
1 if in the past 12 months the household has not been able to afford inviting
friends or family for a meal or drink at least once a month, 0 otherwise
(social participation).
Social participation (cul-
ture)
1 if the household has not been able to afford to buy books, cinema or
theater tickets in the past 12 months, 0 otherwise (social participation).
Political participation 1 if person has inability to vote due to lack of eligibility or distance to polling
station, 0 otherwise (civic participation).
Social participation (clubs) 1 if person has no participation/membership in associations, teams or clubs,
0 otherwise (civic participation).
Civic participation 1 if person has no participation in political/civic activities, 0 otherwise (civic
participation).
46
Table 4: Summary statistics of indicators of social exclusion across countries
Indicators N*Mean STD Min Max
Economic Exclusion 11552 2.00 1.55 0 8
At risk of income poverty 14174 0.30 0.46 0 1
Unmet basic needs 15483 0.10 0.30 0 1
Unemployment 15236 0.13 0.33 0 1
Financial services 15704 0.65 0.48 0 1
Housing 15824 0.14 0.34 0 1
Amenities 15705 0.38 0.48 0 1
ICT 15620 0.38 0.49 0 1
Overcrowding 13828 0.03 0.17 0 1
Exclusion from Social Services 10172 2.65 1.68 0 8
Water 15829 0.43 0.50 0 1
Heating 15826 0.38 0.49 0 1
Low education 15901 0.23 0.42 0 1
School materials 12041 0.31 0.46 0 1
School drop out 15901 0.06 0.24 0 1
Medication 13160 0.27 0.44 0 1
Health care 13906 0.52 0.50 0 1
Transportation 15719 0.40 0.49 0 1
Exclusion from Participation
in Civic & Social life and Networks 12106 2.39 1.30 0 8
Social ties with family 15808 0.10 0.30 0 1
Social ties with friends 14693 0.05 0.23 0 1
Support network 15007 0.18 0.39 0 1
Social participation (private) 15564 0.12 0.32 0 1
Social participation (culture) 15259 0.49 0.50 0 1
Political participation 14668 0.03 0.17 0 1
Social participation (clubs) 15725 0.60 0.49 0 1
Civic participation 15710 0.86 0.35 0 1
"Socially Excluded" 6239 0.24 0.43 0 1
*Information was collected on the level of the household or respondent only, the respondent’s answers
were assumed to be valid for all household members. Other characteristics (age group, gender, education)
were available for all household members. Respondents with missed observations on the 24 indicators were
excluded. The final dataset for the construction of the multidimensional Social Exclusion Index therefore
includes in total 6239 respondents, who represent 23166 household members.
47
Table 5: Summary statistics of indicators of social exclusion, by country
Kazakhstan Moldova Macedonia
Indicators N Mean STD N Mean STD N Mean STD
Economic Exclusion 2223 2.10 1.30 2103 2.33 1.35 1912 1.13 1.26
At risk of income poverty 2517 0.31 0.46 2568 0.31 0.46 2392 0.35 0.48
Unmet basic needs 2640 0.06 0.24 2665 0.12 0.32 2663 0.03 0.17
Unemployment 2679 0.10 0.30 2673 0.09 0.29 2440 0.23 0.42
Financial services 2673 0.77 0.42 2677 0.82 0.38 2637 0.31 0.46
Housing 2683 0.07 0.25 2691 0.07 0.26 2694 0.02 0.15
Amenities 2654 0.36 0.48 2670 0.48 0.50 2667 0.14 0.35
ICT 2634 0.45 0.50 2684 0.43 0.50 2664 0.14 0.34
Overcrowding 2486 0.02 0.15 2281 0.01 0.12 2434 0.00 0.07
Exclusion from Social Services 1732 2.47 1.51 1807 3.15 1.78 1522 1.89 1.43
Water 2698 0.57 0.50 2696 0.53 0.50 2680 0.15 0.35
Heating 2694 0.08 0.27 2693 0.51 0.50 2688 0.64 0.48
Low education 2700 0.18 0.39 2700 0.29 0.46 2700 0.19 0.39
School materials 2084 0.34 0.48 1954 0.35 0.48 1733 0.18 0.38
School drop out 2700 0.10 0.29 2700 0.03 0.16 2700 0.02 0.13
Medication 2289 0.27 0.44 2073 0.26 0.44 2138 0.09 0.29
Health care 2225 0.47 0.50 2586 0.63 0.48 2419 0.33 0.47
Transportation 2679 0.44 0.50 2692 0.49 0.50 2642 0.26 0.44
Exclusion from Participation in
Civic & Social life and Networks 2117 2.68 1.19 1840 2.51 1.42 2205 1.97 1.24
Social ties with family 2679 0.08 0.26 2696 0.21 0.41 2683 0.06 0.24
Social ties with friends 2583 0.05 0.22 2071 0.09 0.29 2651 0.01 0.11
Support network 2580 0.25 0.43 2636 0.20 0.40 2511 0.11 0.32
Social participation (private) 2640 0.09 0.28 2670 0.13 0.34 2657 0.08 0.27
Social participation (culture) 2528 0.52 0.50 2656 0.53 0.50 2580 0.41 0.49
Political participation 2492 0.02 0.14 2528 0.03 0.17 2586 0.01 0.11
Social participation (clubs) 2663 0.75 0.44 2690 0.59 0.49 2665 0.52 0.50
Civic participation 2668 0.92 0.27 2697 0.85 0.36 2649 0.80 0.40
’"Socially Excluded" 1251 0.23 0.42 988 0.33 0.47 974 0.09 0.29
48
Table 5: Summary statistics of indicators of social exclusion, by country (Cont.)
Serbia Tajikistan Ukraine
Indicators N Mean STD N Mean STD N Mean STD
Economic Exclusion 1760 1.22 1.41 1581 3.77 1.28 1973 1.69 1.23
At risk of income poverty 2047 0.24 0.43 2157 0.27 0.45 2493 0.30 0.46
Unmet basic needs 2314 0.15 0.36 2587 0.15 0.36 2614 0.10 0.29
Unemployment 2297 0.16 0.37 2510 0.14 0.34 2637 0.07 0.25
Financial services 2378 0.37 0.48 2672 0.97 0.17 2667 0.65 0.48
Housing 2391 0.02 0.14 2675 0.60 0.49 2690 0.01 0.11
Amenities 2363 0.17 0.37 2682 0.82 0.39 2669 0.27 0.44
ICT 2356 0.21 0.40 2661 0.73 0.45 2621 0.33 0.47
Overcrowding 2215 0.00 0.05 2135 0.14 0.35 2277 0.01 0.08
Exclusion from Social Services 1442 2.20 1.62 1856 3.58 1.56 1813 2.35 1.50
Water 2379 0.22 0.42 2682 0.78 0.41 2694 0.32 0.47
Heating 2392 0.46 0.50 2665 0.55 0.50 2694 0.05 0.22
Low education 2401 0.23 0.42 2700 0.34 0.48 2700 0.12 0.32
School materials 1674 0.39 0.49 2580 0.18 0.39 2016 0.42 0.49
School drop out 2401 0.03 0.17 2700 0.18 0.39 2700 0.03 0.16
Medication 1957 0.18 0.38 2624 0.40 0.49 2079 0.36 0.48
Health care 2129 0.40 0.49 2025 0.68 0.47 2522 0.63 0.48
Transportation 2374 0.32 0.47 2658 0.45 0.50 2674 0.44 0.50
Exclusion from Participation in
Civic & Social life and Networks 2000 2.07 1.17 1863 2.65 1.22 2081 2.52 1.39
Social ties with family 2383 0.10 0.30 2684 0.05 0.22 2683 0.11 0.31
Social ties with friends 2342 0.01 0.11 2537 0.07 0.26 2509 0.09 0.29
Support network 2278 0.17 0.38 2477 0.12 0.33 2525 0.24 0.43
Social participation (private) 2358 0.04 0.20 2620 0.23 0.42 2619 0.13 0.34
Social participation (culture) 2305 0.44 0.50 2595 0.64 0.48 2595 0.39 0.49
Political participation 2257 0.02 0.13 2286 0.05 0.22 2519 0.05 0.21
Social participation (clubs) 2371 0.43 0.50 2674 0.63 0.48 2662 0.68 0.47
Civic participation 2384 0.88 0.32 2625 0.84 0.37 2687 0.86 0.34
’"Socially Excluded" 994 0.11 0.32 849 0.58 0.49 1183 0.19 0.39
49
Table 6: Level of social exclusion, by country*
KZH MLD FYRM SER TJK UKR
Magnitude of Social Exclusion
Social exclusion headcount, SEHR, (%) 23 30 8 10 64 15
Average number of deprivations
among the socially excluded (intensity) 11 12 12 12 12 11
Average share of deprivations, ADS, (intensity)
(the number of deprivations as a percentage of the 24) 47 49 48 49 49 47
Social Exclusion Index (SEHR)×(ADS) 11 15 4 5 32 7
Contribution of Dimensions to the Social Exclusion Index
Economic exclusion 30 30 24 22 37 26
Exclusion from social services 35 39 37 40 37 36
Exclusion from participation in civic and
social life and networks 35 31 39 38 25 38
*KZH=Kazakhstan, MLD=Moldova, FYRM=Macedonia, SER=Serbia, TJK=Tajikistan, and
UKR=Ukraine. The table is constructed based on the 23166 household members.
50
Table 7: Characteristics of the households
Characteristics Definition
Self-reported well-being The answer to the question: Are you satisfied or dissatisfied with
your standard of living? where 5 is defined as Completely Satis-
fied, 4 is defined as Satisfied, 3 is defined as Neither satisfied nor
dissatisfied, 2 is defined as Dissatisfied, 1 is defined as Completely
Dissatisfied.
Socially Excluded Dummy for being socially excluded: 1 if respondent is deprived in at
least nine indicators out of the selected 24 indicators, 0 otherwise.
ln Household consumption Log of respondent’s actual household expenditure (PPP adjusted).
Answer to the question “How much money did your household
spend last month in total [local currency]?”
ln Mean consumption in locality Log of average of household expenditures in the locality or the ref-
erence group. The locality or the reference is defined as same
country; region; distric; and settlement. There are four types of
settlement: village; small town; regional or economic center; and
capital.
House owner Dummy for owning one’s house: 1 if respondent’s household owns
the house (with or without mortgage), 0 otherwise.
Land owner Dummy for owning one’s land: 1 if respondent’s household owns
land, 0 otherwise.
Unemployed Unemployment dummy: 1 if respondent, during the last month, did
not work for payment at least for one day and were registered with
the employment services, 0 otherwise.
Not in the labor force Dummy for those neither currently employed nor unemployed: 1 if
respondent is not employed and is not looking for work during
the past 4 weeks, 0 otherwise.
Gender Gender dummy: 1 if respondent is female, 0 otherwise.
Age Age of respondent at the time of survey (in years).
Age Squared /100 Age in years squared and divided by 100.
Ethnicity in Settlement Ethnicity dummy: 1 if percentage of the respondent’s ethnic group
in the settlement is less than 10% [between 10-20%; between 20-
40%; between 60-80%; greater than 80%], 0 otherwise.
Marital Status Marrital status dummy: 1 for those either married or not married
but living with a partner, 0 otherwise. If divorsed, widowed,
single or not living with a partner, those are coded into the neither
married nor cohabiting category.
Education Years of completed education at the time of the interview (in years).
ln Household size Log of number of people in the household.
Religion: Religion dummy: 1 if respondent is an Orthodox [is a Catholic; is
a Muslim; has no religion; has other religion or missing or n.a.],
0 otherwise.
Settlement type: Settlement dummy: 1 if settlement type is village [small town; re-
gional or economic center; capital], 0 otherwise.
ln population in settlement Log of population in settlement.
Country: Country dummy: 1 if country is Kazakhstan [Moldova; Macedonia;
Serbia; Tajikistan; Ukraine], 0 otherwise.
51
Table 8: Summary characteristics of the households across countries
Characteristics N Mean STD Min Max
Self-reported well-being 15825 3.04 1.04 1.00 5.00
Socially Excluded 6239 0.24 0.43 0.00 1.00
ln Household consumption (actual) 14157 6.23 0.92 2.76 10.13
ln Mean consumption (country) 15901 6.60 0.34 6.13 7.01
ln Mean consumption (region) 15901 6.60 0.37 5.74 7.26
ln Mean consumption (district) 15900 6.56 0.49 4.83 8.13
ln Mean consumption (village) 7698 6.37 0.55 4.70 8.40
ln Mean consumption (small town) 3042 6.64 0.50 3.97 7.54
ln Mean consumption (regional center) 3144 6.68 0.44 4.37 7.73
ln Mean consumption (capital) 2011 6.83 0.35 5.80 7.33
House owner 15901 0.90 0.30 0.00 1.00
Land owner 15901 0.57 0.50 0.00 1.00
Unemployed 15236 0.12 0.32 0.00 1.00
Not in the labor force 15901 0.40 0.49 0.00 1.00
Gender 15901 0.54 0.50 0.00 1.00
Age 15901 42.24 17.38 15.00 105.00
Age squared 15901 20.86 16.15 2.25 110.25
Ethnicity in Settlement (<10%) 15901 0.06 0.23 0.00 1.00
Ethnicity in Settlement (10-20%) 15901 0.04 0.19 0.00 1.00
Ethnicity in Settlement (20-40%) 15901 0.07 0.25 0.00 1.00
Ethnicity in Settlement (40-60%) 15901 0.15 0.35 0.00 1.00
Ethnicity in Settlement (60-80%) 15901 0.16 0.37 0.00 1.00
Ethnicity in Settlement (>80%) 15901 0.52 0.50 0.00 1.00
Marital Status 15901 0.59 0.49 0.00 1.00
Education 15901 11.81 3.11 0.00 35.00
ln Household size 15901 1.13 0.56 0.00 2.30
Religion: Orthodox 15901 0.59 0.49 0.00 1.00
Religion: Catholic 15901 0.07 0.26 0.00 1.00
Religion: Islam 15901 0.27 0.44 0.00 1.00
Religion: Atheist 15901 0.04 0.20 0.00 1.00
Religion: Other 15901 0.03 0.17 0.00 1.00
Settlement type: Village 15901 0.48 0.50 0.00 1.00
Settlement type: Small town 15901 0.19 0.39 0.00 1.00
Settlement type: Regional or economic center 15901 0.20 0.40 0.00 1.00
Settlement type: Capital 15901 0.13 0.33 0.00 1.00
ln population in settlement 15901 9.80 2.31 6.21 14.22
52
Table 9: Summary characteristics of the households, by country
Kazakhstan Moldova Macedonia
Characteristics N Mean STD N Mean STD N Mean STD
Self-reported well-being 2689 3.23 1.04 2696 3.03 1.00 2679 3.03 0.98
Socially Excluded 1251 0.23 0.42 988 0.33 0.47 974 0.09 0.29
ln Household consumption (actual) 2512 5.87 0.73 2568 5.81 0.95 2391 6.71 0.80
ln Mean consumption (country) 2700 6.13 0.00 2700 6.27 0.00 2700 6.97 0.00
ln Mean consumption (region) 2700 6.12 0.21 2700 6.26 0.10 2700 6.97 0.10
ln Mean consumption (district) 2700 6.07 0.38 2700 6.25 0.22 2699 6.94 0.31
ln Mean consumption (village) 1403 5.91 0.38 1621 6.08 0.23 824 6.83 0.41
ln Mean consumption (small town) 410 6.02 0.48 82 6.28 0.15 617 6.84 0.42
ln Mean consumption (reg. center) 765 6.31 0.29 506 6.41 0.41 643 7.05 0.16
ln Mean consumption (capital) 122 6.47 0.07 491 6.47 0.00 615 7.02 0.20
House owner 2700 0.92 0.28 2700 0.91 0.29 2700 0.83 0.38
Land owner 2700 0.56 0.50 2700 0.74 0.44 2700 0.48 0.50
Unemployed 2679 0.08 0.27 2673 0.07 0.26 2440 0.22 0.41
Not in the labor force 2700 0.41 0.49 2700 0.48 0.50 2700 0.29 0.45
Gender 2700 0.51 0.50 2700 0.58 0.49 2700 0.52 0.50
Age 2700 42.43 17.14 2700 43.24 18.04 2700 42.99 16.73
Age squared 2700 20.94 16.01 2700 21.95 16.69 2700 21.28 15.76
Ethnicity in Settlement (<10%) 2700 0.09 0.29 2700 0.04 0.19 2700 0.04 0.20
Ethnicity in Settlement (10-20%) 2700 0.10 0.30 2700 0.05 0.23 2700 0.00 0.02
Ethnicity in Settlement (20-40%) 2700 0.21 0.41 2700 0.03 0.16 2700 0.05 0.22
Ethnicity in Settlement (40-60%) 2700 0.22 0.41 2700 0.07 0.26 2700 0.14 0.35
Ethnicity in Settlement (60-80%) 2700 0.15 0.36 2700 0.23 0.42 2700 0.04 0.19
Ethnicity in Settlement (>80%) 2700 0.23 0.42 2700 0.58 0.49 2700 0.67 0.47
Marital Status 2700 0.55 0.50 2700 0.58 0.49 2700 0.63 0.48
Education 2700 12.10 2.92 2700 11.37 3.20 2700 12.40 3.46
ln Household size 2700 1.07 0.59 2700 0.90 0.51 2700 1.19 0.46
Religion: Orthodox 2700 0.33 0.47 2700 0.94 0.23 2700 0.71 0.45
Religion: Catholic 2700 0.00 0.07 2700 0.03 0.18 2700 0.23 0.42
Religion: Islam 2700 0.58 0.49 2700 0.00 0.04 2700 0.00 0.00
Religion: Atheist 2700 0.07 0.25 2700 0.01 0.10 2700 0.00 0.06
Religion: Other 2700 0.02 0.12 2700 0.01 0.11 2700 0.06 0.24
Settlement type: Village 2700 0.52 0.50 2700 0.60 0.49 2700 0.31 0.46
Settlement type: Small town 2700 0.15 0.36 2700 0.03 0.17 2700 0.23 0.42
Settlement type: Regional center 2700 0.28 0.45 2700 0.19 0.39 2700 0.24 0.43
Settlement type: Capital 2700 0.05 0.21 2700 0.18 0.39 2700 0.23 0.42
ln population in settlement 2700 10.08 2.54 2700 9.52 2.20 2700 9.70 1.32
53
Table 9: Summary characteristics of the households, by country (Conti.)
Serbia Tajikistan Ukraine
Characteristics N Mean STD N Mean STD N Mean STD
Self-reported well-being 2393 2.75 1.06 2684 3.47 0.95 2684 2.67 1.03
Socially Excluded 994 0.11 0.32 849 0.58 0.49 1183 0.19 0.39
ln Household consumption (actual) 2047 6.77 0.73 2148 5.96 0.87 2491 6.36 0.90
ln Mean consumption (country) 2401 7.01 0.00 2700 6.51 0.00 2700 6.78 0.00
ln Mean consumption (region) 2401 7.00 0.15 2700 6.50 0.18 2700 6.76 0.19
ln Mean consumption (district) 2401 7.00 0.21 2700 6.40 0.54 2700 6.75 0.25
ln Mean consumption (village) 1137 6.87 0.24 1873 6.37 0.63 840 6.52 0.47
ln Mean consumption (small town) 534 6.98 0.20 379 6.31 0.67 1020 6.74 0.27
ln Mean consumption (reg. center) 358 7.06 0.20 194 6.29 0.22 678 6.85 0.35
ln Mean consumption (capital) 372 7.28 0.00 249 6.51 0.27 162 6.97 0.00
House owner 2401 0.86 0.34 2700 0.94 0.25 2700 0.94 0.23
Land owner 2401 0.51 0.50 2700 0.57 0.50 2700 0.55 0.50
Unemployed 2297 0.16 0.36 2510 0.12 0.32 2637 0.06 0.23
Not in the labor force 2401 0.29 0.46 2700 0.48 0.50 2700 0.45 0.50
Gender 2401 0.50 0.50 2700 0.57 0.50 2700 0.56 0.50
Age 2401 43.52 16.40 2700 36.33 15.73 2700 45.07 18.64
Age squared 2401 21.62 15.58 2700 15.67 13.57 2700 23.79 17.79
Ethnicity in Settlement (<10%) 2401 0.04 0.21 2700 0.09 0.29 2700 0.03 0.17
Ethnicity in Settlement (10-20%) 2401 0.01 0.12 2700 0.04 0.19 2700 0.02 0.15
Ethnicity in Settlement (20-40%) 2401 0.04 0.20 2700 0.06 0.23 2700 0.03 0.17
Ethnicity in Settlement (40-60%) 2401 0.06 0.24 2700 0.11 0.32 2700 0.26 0.44
Ethnicity in Settlement (60-80%) 2401 0.10 0.30 2700 0.23 0.42 2700 0.21 0.41
Ethnicity in Settlement (>80%) 2401 0.74 0.44 2700 0.47 0.50 2700 0.44 0.50
Marital Status 2401 0.57 0.50 2700 0.66 0.47 2700 0.55 0.50
Education 2401 11.60 3.12 2700 11.01 2.72 2700 12.35 2.92
ln Household size 2401 1.03 0.52 2700 1.63 0.47 2700 0.95 0.49
Religion: Orthodox 2401 0.87 0.33 2700 0.01 0.12 2700 0.69 0.46
Religion: Catholic 2401 0.06 0.25 2700 0.00 0.02 2700 0.09 0.29
Religion: Islam 2401 0.04 0.19 2700 0.98 0.15 2700 0.00 0.00
Religion: Atheist 2401 0.01 0.08 2700 0.00 0.06 2700 0.15 0.35
Religion: Other 2401 0.02 0.14 2700 0.01 0.08 2700 0.07 0.26
Settlement type: Village 2401 0.47 0.50 2700 0.69 0.46 2700 0.31 0.46
Settlement type: Small town 2401 0.22 0.42 2700 0.14 0.35 2700 0.38 0.48
Settlement type: Regional center 2401 0.15 0.36 2700 0.07 0.26 2700 0.25 0.43
Settlement type: Capital 2401 0.15 0.36 2700 0.09 0.29 2700 0.06 0.24
ln population in settlement 2401 9.86 2.58 2700 8.97 2.09 2700 10.69 2.52
54
Table 10: Relative consumption and well-being across countries
Dependent Variable Baseline IV for Household
Self-reported Wel l-being consumption
(1) (2) (3) (4) (5) (6)
ln Household consumption 0.165*** 0.157*** 0.177*** 0.169*** 0.458*** 0.521***
ln Mean HHC in locality10.127***0.102*** 0.088***
House owner 0.322*** 0.310*** 0.319*** 0.307*** 0.259*** 0.254***
Land owner 0.136*** 0.141*** 0.136*** 0.140*** 0.146*** 0.145***
Unemployed 0.411***0.429*** 0.412*** 0.430*** 0.343*** 0.331***
Not in the labor force 0.061*** 0.077*** 0.062***0.078***0.020 0.011
Female 0.014 0.017 0.014 0.017 0.019 0.019
Age 0.045***0.044*** 0.045*** 0.044*** 0.045*** 0.044***
Age-squared 0.043*** 0.042*** 0.043*** 0.042*** 0.043*** 0.043***
Ethnicity in Settlement (<10%)
Ethnicity in Settlement (10-20%) 0.025 0.034 0.019 0.028 0.042 0.037
Ethnicity in Settlement (20-40%) 0.106*0.167*** 0.117** 0.177*** 0.139*** 0.145***
Ethnicity in Settlement (40-60%) 0.021 0.072 0.027 0.076 0.062 0.068
Ethnicity in Settlement (60-80%) 0.061 0.086*0.067 0.089*0.076 0.077*
Ethnicity in Settlement (>80%) 0.040 0.074*0.048 0.080*0.057 0.062
Marital status 0.152*** 0.157*** 0.152*** 0.157*** 0.127*** 0.123***
Education 0.043*** 0.042*** 0.043*** 0.042*** 0.028*** 0.026***
ln Household size 0.051** 0.085*** 0.055** 0.088*** 0.180*** 0.201***
Religion: Orthodox
Religion: Catholic 0.088*0.178*** 0.087*0.179*** 0.230*** 0.240***
Religion: Islam 0.111** 0.057 0.105** 0.053 0.068 0.069
Religion: Atheist 0.111** 0.142*** 0.109** 0.145*** 0.128*** 0.129***
Religion: Other 0.105 0.114*0.104 0.115*0.110*0.108*
Settlement type: Village
Settlement type: Small town 0.040 0.047 0.034 0.041 0.046 0.042
Settlement type: Regional center 0.056 0.080** 0.056 0.075*0.059 0.053
Settlement type: Capital 0.014 0.007 0.019 0.010 0.008 0.008
ln Population in settlement 0.007 0.005 0.011 0.001 0.015*0.015*
Country dummy: Kazakhstan
Country dummy: Moldova 0.201*0.197*** 0.184*0.180***0.201*** 0.188***
Country dummy: Macedonia 0.717*** 0.312*** 0.633*** 0.233*** 0.569*** 0.545***
Country dummy: Serbia 0.954***0.568*** 0.874*** 0.486*** 0.841*** 0.820***
Country dummy: Tajikistan 0.359** 0.309*** 0.318** 0.346*** 0.287*** 0.319***
Country dummy: Ukraine 0.518***0.720***0.433***0.659***0.843*** 0.816***
cut1 0.864***0.980***1.548***1.503*** 0.398 0.192
cut2 0.193 0.059 0.490*0.465** 1.426*** 1.220***
cut3 1.095*** 0.947*** 0.412 0.424** 2.302*** 2.097***
cut4 2.456*** 2.287*** 1.774*** 1.765*** 3.629*** 3.424***
Regional Dummies Y es No Y es N o N o No
N 13466 13466 13466 13466 14730 14729
1Locality is defined as district.
2*, **, and *** indicate significance at the 10%, 5%, and 1% level respectively.
55
Table 11: Relative consumption and well-being in Kazakhstan
Dependent Variable Baseline IV for Household
Self-reported Wel l-being consumption
(1) (2) (3) (4) (5) (6)
ln Household consumption 0.230*** 0.218*** 0.233*** 0.231*** 0.511*** 0.495***
ln Mean HHC in locality10.019 0.072 0.017
House owner 0.445*** 0.450*** 0.445*** 0.449*** 0.375*** 0.376***
Land owner 0.248*** 0.244*** 0.248*** 0.241*** 0.255*** 0.255***
Unemployed 0.417***0.417*** 0.416*** 0.416*** 0.343*** 0.346***
Not in the labor force 0.117** 0.164*** 0.118** 0.165*** 0.133** 0.135**
Female 0.075 0.076*0.075 0.076*0.072 0.072
Age 0.044***0.044*** 0.044*** 0.044*** 0.043*** 0.043***
Age-squared 0.046*** 0.045*** 0.046*** 0.045*** 0.045*** 0.045***
Ethnicity in Settlement (<10%)
Ethnicity in Settlement (10-20%) 0.168*0.186*0.167*0.184*0.190** 0.191**
Ethnicity in Settlement (20-40%) 0.154 0.053 0.153 0.052 0.063 0.062
Ethnicity in Settlement (40-60%) 0.348*** 0.227*** 0.347*** 0.233*** 0.208** 0.208**
Ethnicity in Settlement (60-80%) 0.243** 0.065 0.241** 0.063 0.038 0.039
Ethnicity in Settlement (>80%) 0.073 0.175** 0.073 0.175** 0.176** 0.177**
Marital status 0.148*** 0.154*** 0.148*** 0.154*** 0.127** 0.129**
Education 0.040*** 0.037*** 0.040*** 0.037*** 0.021** 0.021**
ln Household size 0.141*** 0.221*** 0.141*** 0.221*** 0.319*** 0.315***
Religion: Orthodox
Religion: Catholic 0.084 0.107 0.084 0.107 0.192 0.190
Religion: Islam 0.255*** 0.210*** 0.254*** 0.210*** 0.239*** 0.238***
Religion: Atheist 0.017 0.012 0.017 0.011 0.030 0.029
Religion: Other 0.185 0.072 0.185 0.076 0.056 0.056
Settlement type: Village
Settlement type: Small town 0.010 0.106 0.010 0.103 0.106 0.106
Settlement type: Regional center 0.144 0.287*** 0.146 0.291*** 0.246** 0.246**
Settlement type: Capital 0.103 0.278*0.108 0.290** 0.268*0.264*
ln Population in settlement 0.020 0.021 0.021 0.017 0.033 0.033
cut1 0.386 0.815** 0.482 1.140** 0.495 0.514
cut2 0.771** 0.313 0.674 0.011 1.602** 1.621**
cut3 1.544*** 1.062*** 1.448** 0.739 2.341*** 2.360***
cut4 2.961*** 2.426*** 2.865*** 2.102*** 3.700*** 3.719***
Regional Dummies Y es N o Y es N o N o No
N 2484 2484 2484 2484 2653 2653
1Locality is defined as district.
2*, **, and *** indicate significance at the 10%, 5%, and 1% level respectively.
56
Table 12: Relative consumption and well-being in Moldova
Dependent Variable Baseline IV for Household
Self-reported Wel l-being consumption
(1) (2) (3) (4) (5) (6)
ln Household consumption 0.230