Research paper
dddd
economic2.pdf
Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=cods20
Oxford Development Studies
ISSN: 1360-0818 (Print) 1469-9966 (Online) Journal homepage: https://www.tandfonline.com/loi/cods20
Does Economic Growth Raise Happiness in China?
John Knight & Ramani Gunatilaka
To cite this article: John Knight & Ramani Gunatilaka (2011) Does Economic Growth Raise Happiness in China?, Oxford Development Studies, 39:01, 1-24, DOI: 10.1080/13600818.2010.551006
To link to this article: https://doi.org/10.1080/13600818.2010.551006
Published online: 17 Mar 2011.
Submit your article to this journal
Article views: 1191
View related articles
Citing articles: 77 View citing articles
Does Economic Growth Raise Happiness in China?
JOHN KNIGHT & RAMANI GUNATILAKA
ABSTRACT Various measures of satisfaction with life or happiness in China appear not to have risen in recent years, despite China’s remarkable growth of income per capita. The paper brings together and integrates the results of four papers by the authors to provide a methodologically and substantively innovative explanation for this paradox. The four papers are based on a cross-section national household survey relating to 2002 and containing questions on subjective well-being. Their findings help to explain the time-series evidence: they highlight the importance of relative income, rising urban insecurity, rapid urbanization, and changing reference groups in preventing happiness from rising with income.
JEL Classification: D1, D6, O1, R2
1. Introduction
China’s remarkable rate of economic growth since the start of economic reform is
generally assumed to have raised the economic welfare of the Chinese people
dramatically. This is regarded as self-evident from the facts that, in little more than two
decades, average real income per capita has risen more than six times and that about 400
million people have been lifted out of poverty (Ravallion & Chen, 2007). Moreover,
within a quarter of a century China’s “human development index” has risen from 0.37 to
0.68 (United Nations Development Programme (UNDP), 2010, Table 2). The question in
the title appears either absurd or disingenuous.
Nevertheless, starting from the pioneering work of Easterlin (1974), economists have
increasingly asked this question of advanced economies. It has been shown that, in several
advanced economies—including the USA, Japan, the UK, France, Germany, Italy and the
Netherlands—income per capita rose consistently over one or more decades and yet the
mean subjective well-being score remained roughly constant (e.g. Blanchflower &
Oswald, 2004; Clark et al., 2008).
ISSN 1360-0818 print/ISSN 1469-9966 online/11/010001-24
q 2011 Oxford Department of International Development
DOI: 10.1080/13600818.2010.551006
The authors are grateful to the Leverhulme Trust for funding the research, to Richard Easterlin for good advice,
and to seminar participants in Beijing, Johannesburg, Nottingham, Oxford, Oslo, San Francisco, Shanghai and
Stockholm for helpful comments. The research was conducted while Ramani Gunatilaka was visiting the
Department of Economics in Oxford.
John Knight, Department of Economics, University of Oxford, Manor Road Building, Oxford OX1 3UQ, UK.
Email: [email protected]. Ramani Gunatilaka, Department of Econometrics and Business
Statistics, Monash University, Melbourne, Australia. Email: [email protected]
Oxford Development Studies, Vol. 39, No. 1, March 2011
Very few such studies appear to have been made for developing countries, probably
owing to a lack of relevant time-series data on subjective well-being. However, one would
expect that the happiness of people in poor countries is determined in a different way. For
instance, it is arguable that the greatest concern of poor people is to meet their basic
physical needs for food, shelter and clothing, whereas non-poor people are more
concerned with their position and achievements in relation to society. Thus, absolute
income might be important to happiness at low levels of income but relative income might
be more important at higher levels.
Could the findings for the advanced economies be true also for China? Although the
lack of appropriate data prevents us from answering the question directly, we approach it
indirectly on the basis of our four previous papers that reported research on subjective
well-being in China by means of a national household survey (Knight et al., 2009; Knight
& Gunatilaka, 2009, 2010a, b). We begin with some background evidence on subjective
well-being (the terms subjective well-being, happiness and satisfaction with life are used
interchangeably).
2. Background
Kahneman & Krueger (2006) presented a graph obtained from the Gallup Organization,
which had conducted surveys of respondents in China in 4 years within the period
1994–2005. The percentage of respondents who were somewhat satisfied or very satisfied
with life fell monotonically by 15% over that period, and the proportion of respondents
who were somewhat dissatisfied or very dissatisfied rose monotonically. Yet over that
period household real income per capita rose annually on average by 3.7% in rural China
and by 5.4% in urban China. Easterlin & Sawangfa (2010) provide evidence of the trend in
reported life satisfaction or happiness in China from three sources: the Gallup survey, the
Asiabarometer survey and the World Values survey. The results are shown in Table 1.
In each case the average life satisfaction score fell: from 2.82 in 1997 to 2.67 in 2004;
from 3.73 in 2003 to 3.68 in 2007; and from 6.83 in 1995 to 6.76 in 2007, respectively
Table 1. Mean life satisfaction or happiness in China over time
About 1995 About 2000 About 2005
Life satisfaction score Gallup survey (1–4 scale) 2.82 2.78 2.67 Year 1997 1999 2004
Asiabarometer survey (1–5 scale) 3.73 3.68 Year 2003 2006
World Values survey (1–10 scale) 6.83 6.53 6.76 Year 1995 2001 2007
Happiness score World Values survey (1–5 scale) 3.05 2.87 2.94 Year 1995 2001 2007
Notes: An earlier World Values survey is excluded because it was confined to the urban population. The 1995 World Values survey covered central China (two-thirds of the national population) and the 2001 and 2007 surveys were intended to be nationally representative. The 1994 Gallup survey is excluded because it had five rather than four response categories.
Sources: Easterlin & Sawangfa (2010); World Values survey, data for China.
2 J. Knight & R. Gunatilaka
(each survey used different units). The happiness score in the World Values survey also
fell, from 3.05 in 1995 to 2.94 in 2007. Unfortunately, these time-series data sets are not
rich enough to permit direct analysis of the reasons for their trends.
The question posed in the title thus cannot be dismissed out of hand and is worth
exploring further. To do so, it is necessary to review the reasons that have been put forward
for what has come to be known as the “Easterlin paradox” (Easterlin, 1974, 1995, 2001).
Easterlin’s own explanation, both in his original paper and subsequently, is that subjective
well-being is a positive function of income but a negative function of aspirations, and that
aspirations rise along with income, so cancelling out the positive effect of income.
Moreover, the reason that aspirations tend to rise with absolute income is that they are
influenced by relative income.
Any explanation would have to deal with the obvious fact that nearly everyone, in rich
as well as poor countries, if asked, would say they wanted more income, other things being
equal; and, if offered more income, would reveal their preference for it. Easterlin’s
explanation provides an answer: people want more income because they wish to raise their
relative income, or they recognize that the incomes of their comparator groups will rise, or
they fail to recognize that their aspirations will rise as well as their income. Thus, people
run on a “hedonic treadmill”.
Is the Easterlin paradox trivial? At least three arguments might be put forward. The most
basic criticism is that happiness scores are meaningless. However, this is not difficult to
refute owing to the widespread success in estimating happiness functions from sample
surveys. The individual happiness score is the dependent variable and various personal,
household and community characteristics are the explanatory variables. Many functions
produce significant coefficients with predictable signs and powerful regularities across
different countries and contexts.
A second criticism is that subjective reports of happiness are not comparable across
people. This would be important if the object was to compare two individuals, but in large
samples, comparing groups such as men and women or young and old, the problem is
much reduced. Issues of unobserved heterogeneity remain, but they merely warrant
caution in the interpretation of suspect coefficients.
A third criticism is that people redefine their happiness scores over time. For instance, if
people adjust their aspirations to the utility they normally experience, an improvement in
their normal utility would lead them to report no higher happiness than previously, even if
they were experiencing higher utility than previously. People are thus on an “aspirations
treadmill”, not a “hedonic treadmill”. A test of this argument requires separate measures of
“experienced utility” (“net affect” in psychology, or feelings) and of subjective well-being
(life satisfaction). Kahneman & Krueger (2006) presented evidence suggesting that
measures of net affect show as much adaptation as do measures of life satisfaction, and
accordingly reject this criticism. In any case, there is no consensus that there is such a thing
as utility independent of aspirations, i.e. that the utility which a person experiences can be
separated from their perception of happiness, however formed.
In contrast to these arguments, there is now a considerable literature providing
evidence—largely for advanced economies—that happiness is sensitive to relative income
(e.g. Frank, 1997; Clark & Oswald, 1998; Frey & Stutzer, 2002; Luttmer, 2004; Graham &
Felton, 2006; Clark et al., 2008). The effect of reference group income is normally
negative but a couple of studies have shown it to be positive (Senik, 2004; Kingdon &
Knight, 2007). There is also evidence-based research showing that aspirations are
Does Economic Growth Raise Happiness in China? 3
important to subjective well-being (Stutzer, 2004; Di Tella et al., 2003). This research
provides the justification for approaching the question posed in the title within the
framework of Easterlin’s explanation for his paradox.
3. Survey, Data and Method
The data used in this paper are from the national household survey, relating to 2002, of the
China Household Income Project (CHIP). This is the third CHIP cross-section national
household survey, containing rich socio-economic information. All three surveys (1988,
1995 and 2002) were designed by the CHIP research team, with hypotheses in mind, but
only the 2002 survey contained questions on subjective well-being.
There were just a couple of subjective well-being questions in the questionnaires for
the subsamples of (registered) urban resident households and rural–urban migrant
households, but the questionnaire for rural households contained a specially designed
module on subjective well-being. The analysis had to be based on a snapshot picture with
no panel element. The paper pioneers the analysis of the question posed in the title, but it
can only be a suggestive beginning.
The subjective well-being question that is available for all three subsamples can be
translated as “how happy are you nowadays?” Five answers were offered: very happy,
happy, so-so, unhappy and not at all happy. Answers to this question form the dependent
variable in much of the analysis. It was treated either as an ordinal or as a cardinal
measure, involving either ordered probit or ordinary least squares (OLS) estimation. In line
with the methodological study by Ferrer-i-Carbonnel & Frijters (2004), no substantive
differences were found between the results using the two measures, and accordingly only
the cardinal results are reported because they are easier to interpret. The household head,
or the main member present, was asked the question; the respondent is identified.
The explanatory variables in the happiness equations are a set of individual, household
and community socio-economic characteristics. We distinguish what we term basic
variables, conventional economic variables, comparison variables, insecurity variables
and attitudinal variables. We retain the specifications taken from the four papers but, to
simplify, generally report in the tables only those variables that are discussed in the text.
The coefficients in the happiness functions represent associations and not necessarily
the hypothesized causal relationships. They might instead reflect the influence of
unobserved variables on both the dependent and the independent variables, or reverse
causation. In some cases we shall suggest reasons why the independent variable might
have a causal effect on happiness but without establishing causation, either because the
variable is not germane to the main argument or because a valid instrument is not
available. Where the interpretation is important to the story, as in the case of income, an
attempt is made to isolate the effect of exogenous variation in the independent variable by
means of instrumenting. Owing to the difficulty of finding persuasive instruments in a
cross-section analysis, the relevant statistical tests are presented in each table and the
theoretical and contextual plausibility of the instruments is discussed in footnotes.
4. Rural Happiness
We begin with rural happiness, drawing on the paper that analyses its determinants
(Knight et al., 2009). Despite the fact that rural-dwellers have relatively low incomes and
4 J. Knight & R. Gunatilaka
have been left behind in China’s economic development, it appears that dissatisfaction
with life is not widespread in rural China. No less than 62% of the sample reported
themselves to be happy or very happy, and only 9% not happy or not at all happy. With
very happy having a score of 4, happy 3, so-so 2, unhappy 1 and not at all happy 0, the
mean score was 2.67. Nevertheless, there is much variation in happiness scores, and this
variation can be explained well by the variables in the survey.
The happiness functions for rural households are reported in Table 2. The first and second
columns show the basic and the full OLS equations, respectively, and the third and fourth
columns the basic and the full instrumental variable (IV) equations, in which log (ln) income
level is instrumented. Many of the coefficients are statistically significant, have predictable
signs and display the regularities that are common to many happiness studies around the
world. For instance, the age–happiness profile has a U-shaped pattern, and being female,
being married and being in good health all raise happiness. The conventional economic
variables affect happiness, in line with basic economic theory, but the contributions
of ln income and net wealth (positive), and of working hours (negative) are weak.
We instrumented the income variable in case it was endogenous. 1 The effect was to raise the
coefficient on ln income. We had expected unobserved characteristics, such as a happy
disposition, to raise both income and happiness, so producing upward bias in the OLS
equation. The downward bias suggested either that aspirations raise income but lower
happiness or that there is attenuation bias resulting from measurement error. Even then,
the effect of a doubling of income was to raise the happiness score by only 0.4 points
(column 3).
Despite the apparent unimportance of income for happiness, 64% of the unhappy gave lack
of income as the reason for their unhappiness. A possible explanation for these discrepant
results is that happiness is not only a positive function of income but also a negative function
of aspirations, and that the latter can be governed by the income of the reference group.
The reference group is likely to be determined by the information that people possess and by
their social interactions. Most rural people report confining their reference groups to the
village: 68% make comparisons with their neighbours or fellow villagers.
Happiness is sensitive to respondents’ perceptions of their household’s position in the
village income distribution (as only 10 households were sampled in each village, it is not
possible to use actual instead of perceived position). Five categories are distinguished:
income perceived to be much above, above, at, below, and much below the village
average, with the middle category being the omitted variable in the dummy variable
analysis. The coefficients are large: that of the highest income category is greater than that
of the lowest by 1.05 (column 2). The notion of relative deprivation, as developed by
sociologists such as Runciman (1966), appears to be relevant. Thus, a rise or fall in income
tends to be offset if there is a simultaneous rise or fall in village income. Aspirations
appear to adjust to the income of the community, so producing a hedonic treadmill.
By contrast, income inequality in the county (as measured by the Gini coefficient of
income per capita of the sampled households) is found to raise happiness. Hirschman’s
(1973) “tunnel effect”—the analogy of two lines of cars jammed in a tunnel—might
provide the explanation: initially at least, the movement of one line raises expectations that
the other will also move. Thus, county income inequality might serve as a “demonstration
effect” of possible progress in the future.
Reference time is relevant as well as reference income: those whose current living
standards are considered to be higher than 5 years ago are happier than those whose living
Does Economic Growth Raise Happiness in China? 5
T a b le
2 . D e te rm
in a n ts o f h a p p in e ss
in ru ra l C h in a : O L S a n d IV
e st im
a te s
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
B a
si c
v a
ri a
b le
s A g e
4 5 .4 1
2 0 .0 1 1 7 7 1 * *
2 0 .0 1 6 6 3 5 * *
2 0 .0 2 1 5 4 3 * *
2 0 .0 2 6 9 2 7 * * *
A g e sq u a re d
2 1 7 4 .0 9
0 .0 0 0 1 7 9 * * *
0 .0 0 0 2 3 1 * * *
0 .0 0 0 2 3 3 * *
0 .0 0 0 3 1 1 * * *
M a le
0 .7 4
2 0 .0 6 6 8 9 7 * * *
2 0 .0 5 3 6 5 7 * *
0 .0 0 2 7 6 2
2 0 .0 9 0 8 2 6
M a rr ie d
0 .9 5
0 .1 3 3 2 0 5 * *
0 .1 1 4 7 8 2
D iv o rc e d
0 .0 0
2 0 .3 9 7 7 8 2 * *
2 0 .7 0 9 7 4 8 * * *
W id o w e d
0 .0 2
2 0 .2 4 4 5 9 5 * * *
2 0 .1 7 3 6 5
In g o o d h e a lt h
0 .7 4
0 .4 1 1 7 6 4 * * *
0 .2 8 9 4 3 3 * * *
0 .3 9 3 0 0 0 * * *
0 .3 0 4 5 4 9 * * *
C o
n v e n
ti o
n a
l e c o
n o
m ic
v a
ri a
b le
s L o g o f p e r c a p it a h o u se h o ld
in c o m e
7 .6 8
0 .1 6 0 2 3 7 * * *
0 .0 7 0 4 7 0 * * *
0 .9 6 8 7 0 1 * * *
0 .5 0 7 3 7 1
N e t w e a lt h (’ 0 0 0 y u a n )
3 7 .6 8
0 .0 0 1 7 8 5 * * *
0 .0 0 0 5 0 7 *
2 0 .0 0 3 9 9 0 * *
2 0 .0 0 1 8 9 3
W o rk in g h o u rs
(’ 0 0 p e r y e a r)
1 7 .0 9
2 0 .0 0 3 3 5 2 * * *
2 0 .0 0 1 5 0 4
2 0 .0 0 8 9 5 3 * * *
2 0 .0 0 3 7 0 3
C o
m p
a ri
so n
v a
ri a
b le
s H o u se h o ld
in c o m e m u c h a b o v e v il la g e a v e ra g e
0 .0 2
0 .2 1 6 2 5 1 * * *
0 .1 7 2 5 7 6 *
H o u se h o ld
in c o m e a b o v e v il la g e a v e ra g e
0 .1 9
0 .1 1 0 1 3 5 * * *
0 .1 0 3 9 9 1 * *
H o u se h o ld
in c o m e b e lo w
v il la g e a v e ra g e
0 .2 0
2 0 .2 7 0 0 8 5 * * *
2 0 .1 9 2 5 1 3 * * *
H o u se h o ld
in c o m e m u c h b e lo w
v il la g e a v e ra g e
0 .0 3
2 0 .8 4 3 0 1 6 * * *
2 0 .7 2 7 4 1 8 * * *
C u rr e n t li v in g st a n d a rd s b e tt e r th a n 5 y e a rs
a g o
0 .6 1
0 .1 8 1 1 3 9 * * *
0 .1 6 3 4 3 3 * * *
C u rr e n t li v in g st a n d a rd s w o rs e th a n 5 y e a rs
a g o
0 .0 5
2 0 .1 8 1 7 0 2 * * *
2 0 .1 8 2 9 2 6 * * *
E x p e c t b ig
in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .1 0
0 .1 8 9 2 4 5 * * *
0 .1 3 8 6 5 9 * *
E x p e c t sm
a ll in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .6 8
0 .0 8 8 2 5 5 * * *
0 .0 5 9 3 2 1
E x p e c t d e c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .0 4
2 0 .0 8 7 4 3 2
2 0 .0 7 1 1 4 4
G in i c o e ffi c ie n t o f h o u se h o ld
in c o m e p e r c a p it a
a t c o u n ty
le v e l
0 .2 8
0 .7 2 5 1 1 0 * * *
0 .7 7 2 5 5 1 * *
A tt
it u
d in
a l
v a
ri a
b le
s D e g re e o f h a rm
o n y a m o n g li n e a g e s
2 .8 2
0 .0 3 7 9 8 5 * *
0 .0 3 7 1 1 9
D e g re e o f h a rm
o n y in
v il la g e
2 .8 3
0 .0 7 3 6 0 0 * * *
0 .0 6 4 2 0 7 * *
A g re e th a t m o n e y is im
p o rt a n t
2 .3 3
2 0 .0 3 2 2 3 0 * *
2 0 .0 2 9 0 0 7
Im p o rt a n c e o f fa m il y
3 .9 0
0 .0 4 6 6 5 9
0 .0 7 4 3 2 5 *
Im p o rt a n c e o f fr ie n d s
3 .3 5
0 .0 4 9 5 3 4 * * *
0 .0 1 4 8 5 6
6 J. Knight & R. Gunatilaka
C o n st a n t
0 .7 8 6 0 9 0 * * *
0 .8 7 9 2 0 6 * * *
2 4 .5 1 4 8 3 4 * *
2 1 .8 1 8 5 6 3
R -s q u a re d /c e n tr e d
R -s q u a re d
0 .2 1 5
0 .3 4 0
2 0 .1 4 1
0 .2 2 7
N u m b e r o f o b se rv a ti o n s
8 8 7 2
7 0 0 0
5 1 9 8
4 2 2 8
S ig n ifi c a n c e o f e x c lu si o n re st ri c ti o n s in
fi rs t st a g e e q u a ti o n
F a th e r’ s y e a rs
o f e d u c a ti o n
* S p o u se ’s
e d u c a ti o n
* * *
* *
F -t e st o f e x c lu d e d in st ru m e n ts ( p -v a l)
0 .0 0 0 0
0 .0 0 5 4
S a rg a n te st /H a n se n
J -s ta ti st ic , fo r o v e r-
id e n ti fi c a ti o n o f a ll in st ru m e n ts ( p -v a l)
0 .9 7 8 8
0 .9 6 9 6
A n d e rs o n -R u b in -W
a ld ,
F -t e st ( p -v a l)
0 .0 0 0 4
0 .3 8 3 4
N o
te s:
1 . D e p e n d e n t v a ri a b le s:
sc o re
o f h a p p in e ss
b a se d o n c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts
a s fo ll o w s:
v e ry
h a p p y ¼
4 ; h a p p y ¼
3 ; so -s o ¼
2 ; n o t
h a p p y ¼
1 a n d n o t a t a ll h a p p y ¼
0 .
2 . In d e p e n d e n t v a ri a b le s w it h c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts so
th a t g re a te r in te n si ty
is re p re se n te d b y a h ig h e r v a lu e a re : le v e l o f h a rm
o n y
a m o n g li n e a g e s, le v e l o f h a rm
o n y a m o n g v il la g e rs , a g re e m e n t th a t m o n e y is im
p o rt a n t; im
p o rt a n c e o f fa m il y , im
p o rt a n c e o f fr ie n d s; im
p o rt a n c e o f re li g io n .
3 . T h e o m it te d c a te g o ri e s in
th e d u m m y v a ri a b le
a n a ly se s a re : fe m a le
se x ; si n g le ; c u rr e n t li v in g st a n d a rd
th e sa m e a s 5 y e a rs
a g o .
4 . * * * , * * a n d * d e n o te
st a ti st ic a l si g n ifi c a n c e a t th e 1 % , 5 %
a n d 1 0 %
le v e ls , re sp e c ti v e ly .
5 . In st ru m e n te d v a ri a b le s re g re ss io n re su lt s a re
g e n e ra te d u si n g th e B a u m
e t
a l. (2 0 0 3 ) iv re g 2 .a d o p ro g ra m
fo r S ta ta .
6 . V a ri a b le s re la te d to
m a ri ta l st a te s h a v e b e e n e x c lu d e d fr o m
th e IV
sp e c ifi c a ti o n s b e c a u se
sp o u se ’s
y e a rs
o f e d u c a ti o n is u se d a s a n in st ru m e n t.
7 . M o d e ls (2 ) a n d (4 ) h a v e b e e n c lu st e re d a t v il la g e le v e l fo r ro b u st st a n d a rd
e rr o rs .
8 . N e t w e a lt h is d e fi n e d a s h o u se h o ld
fi n a n c ia l a ss e ts , p ro d u c ti v e a ss e ts a n d c o n su m e r d u ra b le s le ss
d e b ts .
S o
u rc
e : K n ig h t
e t
a l. (2 0 0 9 , T a b le s 6 a n d 1 1 ). N o t a ll e x p la n a to ry
v a ri a b le s c o n ta in e d in
th o se
ta b le s a re
re p o rt e d h e re .
Does Economic Growth Raise Happiness in China? 7
standards are now lower. By comparison with static expectations, those who expect an
increase in income over the next 5 years have a higher current happiness score whereas
those who expect a decrease have a lower score, other things being equal. This is
inconsistent with the standard assumption that current utility depends on current
consumption, not on expected future consumption; it suggests that people internalize their
future states into their current happiness. That being the case, it is consistent with the
psychological research findings (e.g. Rabin, 1998) that people tend to base their aspirations
on current incomes, and that they are better able to project their income into the future than
their aspirations.
There is some more evidence that aspirations are important for happiness. We can
distinguish those whose comparators are within and those whose comparators are beyond
the village. Relative income within the village appears to be less important, and the
coefficients showing the effect of future income on current happiness all have lower values
in the case of those with reference groups beyond the village (Knight et al., 2009, Table 9).
This suggests that the aspirations, relative to current income, of villagers with wider
horizons are raised by the higher incomes of their comparators.
We introduced a set of attitudinal variables into our happiness functions, in an attempt
to explore otherwise hidden influences. The significant coefficients suggest that rural
people who derive their satisfaction with life more from personal relationships and less
from material goods and services are happier, other things being equal, although reverse
causation is also possible.
5. Urban Happiness
Our discussion of urban happiness draws on another paper (Knight & Gunatilaka, 2010b).
Happiness functions for households with urban hukou (residence registration) are shown in
Table 3. Again, there are four columns: basic and full OLS and basic and full IV equations.
The full equations cover only respondents who were employed (64% of the total), the
reason being that we could then include a set of variables representing urban insecurity.
We obtained the conventional results for some of the standard variables. The coefficients
on the ln income variable are roughly twice the size of the coefficients in the
corresponding functions for rural residents (columns 1 and 2): it appears that urban people
may be more materialistic, in the sense that either aspirations for or need for income are
raised by urban living. When the income variable is instrumented, its coefficients lose their
significance—but instrumenting might not be necessary. 2
Our hypothesis is that urban people also experience relative deprivation. We found
two indicators that relative income is important for happiness. First, households in each
city were grouped into four income per capita quarters. Given the highest quarter as the
reference category, the coefficients on the quarters become monotonically more and
more negative, and the effect is both statistically significant and substantively important.
City mean income per capita across the cities varies sufficiently for this variable not
simply to reflect the variation in household incomes. Second, the log of average urban
income per capita in the province of residence has a negative coefficient (but significantly
so only in the OLS equation). In the urban case, unlike the rural case, the effect of
surrounding prosperity on aspirations may arouse feelings of relative deprivation.
Those who consider income distribution, both in the nation and in the city, to be fair
are happier, ceteris paribus, although it is unclear which way causation runs. As with
8 J. Knight & R. Gunatilaka
T a b le
3 . D e te rm
in a n ts o f h a p p in e ss
in u rb a n C h in a : O L S a n d IV
e st im
a te s
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
B a
si c
v a
ri a
b le
s A g e
4 6 .6 6
2 0 .0 4 7 2 8 9 * * *
2 0 .0 1 8 5 9 8
2 0 .0 4 5 1 2 7 * * *
2 0 .0 1 6 7 6 2
A g e sq u a re d
2 3 0 4 .1 5
0 .0 0 0 5 0 5 * * *
0 .0 0 0 2 3 3
0 .0 0 0 4 9 1 * * *
0 .0 0 0 2 9 7 * *
M a le
0 .4 5
2 0 .0 7 8 7 2 8 * * *
2 0 .0 7 5 0 4 2 * *
2 0 .0 9 0 3 4 2 *
2 0 .1 3 6 9 0 4 * * *
M a rr ie d
0 .9 4
0 .1 9 2 9 2 7 * *
0 .0 6 3 9 8 9
0 .1 7 3 1 2 1
2 0 .0 1 5 5 5 4
D iv o rc e d
0 .0 2
2 0 .2 1 2 9 2 4 *
2 0 .3 9 0 6 2 0 *
2 0 .2 3 6 7 1 2 *
2 0 .4 5 9 3 1 4 * * *
W id o w e d
0 .0 3
2 0 .0 1 9 1 2 6
2 0 .2 5 1 9 0 3 *
2 0 .0 4 6 5 1 3
2 0 .4 3 0 4 6 3 * * *
In g o o d h e a lt h
0 .6 0
0 .2 7 2 3 0 3 * * *
0 .1 6 2 3 7 8 * * *
0 .2 7 3 7 5 2 * * *
0 .1 6 7 7 3 1 * * *
C o
n v e n
ti o
n a
l e c o
n o
m ic
v a
ri a
b le
s L o g o f p e r c a p it a h o u se h o ld
in c o m e
8 .8 3
0 .3 2 2 3 8 6 * * *
0 .1 8 0 1 2 2 * * *
0 .2 5 0 9 6 8
2 0 .4 3 0 6 8 6
N e t w e a lt h (’ 0 0 0 y u a n )
4 5 .9 9
0 .0 0 0 2 0 9 * *
0 .0 0 0 1 8 9
0 .0 0 0 3 0 9
0 .0 0 0 6 0 9 * * *
W o rk in g h o u rs
(’ 0 0 p e r y e a r)
1 5 .2 9
2 0 .0 0 1 0 6 6
2 0 .0 0 1 2 1 4
2 0 .0 0 1 1 1
2 0 .0 0 4 4 2 8 *
C o
m p
a ri
so n
v a
ri a b
le s
E x te n t o f fa ir n e ss , in c o m e d is tr ib u ti o n in
C h in a
0 .7 7
0 .0 7 3 3 2 1 * *
0 .0 6 3 7 8 2 * *
E x te n t o f fa ir n e ss , in c o m e d is tr ib u ti o n in
c it y
0 .8 2
0 .1 0 0 7 1 7 * * *
0 .1 1 7 9 2 9 * * *
L iv in g st a n d a rd
in se c o n d h ig h e st q u a rt e r in
c it y
0 .3 2
2 0 .2 3 5 0 2 5 * *
2 0 .2 6 1 2 9 2 * *
L iv in g st a n d a rd
in th ir d h ig h e st q u a rt e r in
c it y
0 .5 6
2 0 .4 3 9 4 1 2 * * *
2 0 .6 1 0 8 6 5 * * *
L iv in g st a n d a rd
in lo w e st q u a rt e r in
c it y
0 .1 1
2 0 .9 2 5 2 5 6 * * *
2 1 .3 1 1 8 1 7 * * *
E x p e c t b ig
in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .0 2
0 .2 8 0 7 5 7 * *
0 .2 8 9 5 7 0 * * *
E x p e c t sm
a ll in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .4 6
0 .1 0 2 2 3 4 * *
0 .1 1 5 9 8 0 * * *
E x p e c t d e c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .1 9
2 0 .2 3 8 0 4 8 * * *
2 0 .2 1 4 8 4 6 * * *
L n a v e ra g e p e r c a p it a in c o m e in
p ro v in c e
8 .9 4
2 0 .1 6 6 2 8 4 * *
0 .3 4 8 8 1 8
In se
c u ri
ty v a ri
a b le
s U n e m p lo y e d
0 .0 5
2 0 .2 9 1 0 6 2 * * *
2 0 .3 2 0 8 5 9 * * *
C o rr u p ti o n is m o st im
p o rt a n t so c ia l p ro b le m
0 .2 1
2 0 .0 9 5 0 0 0 * * *
2 0 .1 0 5 7 9 2 * *
U n e m p lo y m e n t o r
x ia
g a
n g m o st im
p o rt a n t p ro b le m
0 .3 2
2 0 .0 8 8 7 7 6 *
2 0 .1 5 7 4 3 3 * * *
S o c ia l p o la ri z a ti o n is m o st im
p o rt a n t so c ia l p ro b le m
0 .0 6
2 0 .1 9 3 7 5 0 * *
2 0 .1 8 6 7 7 4 * * *
Im m o ra li ty
is m o st im
p o rt a n t so c ia l p ro b le m
0 .0 1
2 0 .3 8 4 3 7 9 *
2 0 .3 2 1 2 5 9 * *
E n te rp ri se
m a d e h ig h p ro fi t
0 .0 9
0 .0 2 9 2 7 9
0 .0 9 4 0 5 4 * *
E n te rp ri se
m a d e lo ss
0 .0 8
2 0 .0 8 0 0 6 7 *
2 0 .1 6 6 4 9 2 * * *
L a id
o ff w o rk
so m e ti m e in
2 0 0 2
0 .3 7
2 0 .1 3 4 1 2 2 * *
2 0 .2 3 5 9 8 6 * * *
(C o
n ti
n u
e s)
Does Economic Growth Raise Happiness in China? 9
T a b le
3 .
C o
n ti
n u
e d
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
C o n st a n t
0 .3 7 2 5 2 1 *
2 .9 9 0 4 3 4 * * *
0 .9 0 6 4 9 6
3 .5 5 4 0 6 3 * * *
R -s q u a re d /c e n tr e d
R -s q u a re d
0 .1 1 7
0 .2 4 2
0 .1 1 5
0 .1 3 8
N u m b e r o f o b se rv a ti o n s
6 4 9 5
4 1 5 2
6 4 9 5
4 1 5 1
S ig n ifi c a n c e o f in st ru m e n ts in
fi rs t st a g e e q u a ti o n
N u m b e r o f la b o u r fo rc e p a rt ic ip a n ts
* P a re n ts ’ m e m b e rs h ip
o f C o m m u n is t P a rt y
* * *
F a th e r’ s y e a rs
o f e d u c a ti o n
* * *
M o th e r’ s y e a rs
o f e d u c a ti o n
* * *
F -t e st o f e x c lu d e d in st ru m e n ts ( p -v a l)
0 .0 0 0 0
0 .0 0 0 0
S a rg a n te st , fo r o v e r- id e n ti fi c a ti o n o f a ll in st ru m e n ts ( p -v a l)
0 .1 2 7 4
0 .6 9 0 5
A n d e rs o n -R u b in
te st o f jo in t si g n ifi c a n c e o f e n d o g e n o u s
re g re ss o rs in
m a in
e q u a ti o n , F -t e st ( p -v a l)
0 .2 1 7 2
0 .2 2 6 9
N o
te s:
1 . D e p e n d e n t v a ri a b le : sc o re
o f h a p p in e ss
is b a se d o n c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts a s fo ll o w s: v e ry
h a p p y ¼
4 ; h a p p y ¼
3 ; so -s o ¼
2 ; n o t
h a p p y ¼
1 a n d n o t a t a ll h a p p y ¼
0 .
2 . M o d e ls (2 ) a n d (4 ) a re
li m it e d to
su b sa m p le
o f e m p lo y e d re sp o n d e n ts .
3 . O m it te d c a te g o ri e s in
th e d u m m y v a ri a b le
a n a ly se s a re : fe m a le
se x ; si n g le ; e m p lo y e d o r la b o u r fo rc e n o n -p a rt ic ip a n t; n o t h e a lt h y ; li v in g st a n d a rd
in h ig h e st
q u a rt e r in
c it y ; n o c h a n g e in
in c o m e e x p e c te d in
n e x t 5 y e a rs ; e n v ir o n m e n ta l d e g ra d a ti o n m o st
im p o rt a n t so c ia l p ro b le m ; e n te rp ri se
m a d e m a rg in a l p ro fi t;
e m p lo y e d a ll o f 2 0 0 2 .
4 . In d e p e n d e n t v a ri a b le s w it h c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts so
th a t a h ig h e r v a lu e d e n o te s g re a te r in te n si ty
a re : e x te n t o f fa ir n e ss , in c o m e
d is tr ib u ti o n in
C h in a , e x te n t o f fa ir n e ss , in c o m e d is tr ib u ti o n in
th e c it y .
5 . M o d e ls (2 ) a n d (4 ) a re
c lu st e re d a t p ro v in c e le v e l fo r ro b u st st a n d a rd
e rr o rs .
6 . In st ru m e n te d v a ri a b le s re g re ss io n re su lt s a re
g e n e ra te d u si n g th e B a u m
e t
a l. (2 0 0 3 ) iv re g 2 .a d o p ro g ra m
fo r S ta ta .
S o
u rc
e : K n ig h t &
G u n a ti la k a (2 0 1 0 b , T a b le s 7 a n d A 7 ). N o t a ll e x p la n a to ry
v a ri a b le s c o n ta in e d in
th o se
ta b le s a re
re p o rt e d h e re .
10 J. Knight & R. Gunatilaka
rural-dwellers, expected future income is important to current happiness, possibly because
people internalize their future states and they also assume that their aspirations in the
future will be the same as their current aspirations.
There was a high rate of retrenchment by state-owned enterprises in the years prior to
the survey, and retrenched workers faced great difficulties in finding re-employment.
The social security system was in transition from being employer-based to insurance-
based, and unemployment benefits were not reliable, so that many unemployed workers
received very little. We expected the new uncertainties of urban living to depress
happiness. We therefore explored the effect of insecurity on urban-dwellers’ happiness.
The experience of current unemployment, and of having been laid off in the past, had
a significant negative coefficient, as also did the dummy variable denoting that a
worker’s employing enterprise made a loss, which would increase the employee’s
chances of being made redundant. Durkheim’s (1897) notion of anomie might be
relevant. He defined anomie as normlessness, when social rules break down and people
do not know what to expect of each other. The remarkable economic progress, the rapid
creation of markets, the withdrawal of institutional support and the demise of ideology
might have created a state of anomie. The survey does not include good attitudinal
questions for identifying anomie. However, respondents were asked what they
considered to be the most important social problem. Three suggestive pointers are the
negative coefficients in the happiness function on corruption, on social polarization, and
on immorality.
6. A Rural–Urban Comparison
We went on to make a comparison of rural and urban China (Knight & Gunatilaka, 2010b).
It is well known that China has a considerable rural–urban divide (Knight & Song, 1999).
The ratio of urban-to-rural household income per capita has exceeded 2.0-to-1 throughout
the period of economic reform, and has actually risen in recent years, despite the economic
reforms and marketization that partly integrated the rural and urban sectors. In 2002, the
year of the survey, the ratio from the survey stood at 3.1-to-1. We would therefore expect a
correspondingly large divide in subjective well-being. Yet the survey also shows that,
when happiness is converted into a cardinal value, the urban score is no higher than the
rural score. Indeed, the reported mean urban happiness (2.5) is actually lower than the
mean rural happiness (2.7). How can this result be explained?
We first calculated a standard Oaxaca decomposition of these mean differences in
happiness using those variables in the equations that are identical in the two subsamples.
The difference in income, of course, simply added to the puzzle. What raised the happiness
of rural people was their superior happiness generation function. Unfortunately, much of
the work was being done by the difference in the intercept terms, which remained
unexplained. It was necessary to produce an explanation from the separate and
non-identical rural and urban happiness functions.
It is possible that in some societies there is a cultural unwillingness to report happiness,
or alternatively unhappiness, and that comparisons made across culturally distinct groups
might be misleading as a result. Thus, a greater willingness of urban than of rural people to
report being less than happy might explain our results. We cannot reject the hypothesis,
but one piece of evidence points against it. As we shall see, rural migrant households living
in the cities reported having lower average happiness than did urban households. A culture
Does Economic Growth Raise Happiness in China? 11
of not wishing to admit being unhappy is not observable among households that were
recently part of rural society.
Our preferred explanation, based on Tables 2 and 3, runs as follows. On the one hand,
dissatisfaction with life is not widespread in rural China, despite the relative poverty and
low socio-economic status of its people in Chinese society. The basic reasons are that they
have limited information sets and narrow reference groups, they expect their income to rise
in the future, and they place a high value on personal and community relationships. On the
other hand, the relatively low happiness of urban people, despite their relatively high
income and their expectations of higher income in the future, has to do with the nature of
the urban society that has emerged in recent years. High aspirations, governed by reference
groups, appear to give rise to the relative deprivation that makes for unhappiness.
In addition, the greater insecurity associated with redundancy, unemployment and various
other urban social ills also makes city-dwellers unhappy.
7. Migrant Happiness
Rural–urban migration in China has grown remarkably in recent years: the number of
rural–urban migrants was about 120 million in 2002. Many of the migrants are in the cities
temporarily but settlement is increasingly permitted. The higher income to be obtained in
the city than in the village appears to provide a strong incentive to migrate. The 2002
national household survey contained a unique feature—a nationally representative
subsample of rural–urban migrants, i.e. rural hukou households living in urban areas.
Their subjective well-being is analysed in Knight & Gunatilaka (2010a). The average
happiness score of these, fairly settled, migrants is lower than that of rural residents. This
appears to be inconsistent with the economic theories of rural–urban migration based on
utility maximization. We looked at three main possible explanations: in terms of the
hardships of urban life that they experience; in terms of self-selection; and in terms of
revised aspirations.
We proceeded by estimating migrant happiness functions (Table 4), both OLS (columns
1 and 2) and IV (columns 3 and 4). Again we found the usual results for several of the basic
variables. The coefficient on ln income per capita is significantly positive but its values
indicate that a doubling of income raises the happiness score by only 0.13 points (column
2). Although the coefficient is raised, the inference that income level is unimportant is not
altered by instrumenting the income variable. 3 The coefficient tends to rise with length of
stay, suggesting either that there is a process of self-selection or that migrants may become
more materialistic as they lay down deeper urban roots. Although current income is not
important to happiness, expectations of income over the next 5 years enter powerfully and
significantly into the current happiness score. Again, this suggests that anticipated future
happiness is absorbed into current happiness, but also that people are bad at forecasting
how their aspirations will change if income changes and that they judge their future
happiness on the basis of their current aspirations. An alternative interpretation is that
expected future income determines current consumption, in line with the “permanent
income” theory of consumption, and that the relationship would therefore not survive if
current income were replaced by current consumption in the happiness function. However,
this substitution made no notable difference to the coefficients on expected income
(Knight & Gunatilaka, 2010a, p. 117).
12 J. Knight & R. Gunatilaka
T a b le
4 . H a p p in e ss
fu n c ti o n s o f ru ra l– u rb a n m ig ra n ts : O L S a n d IV
e st im
a te s
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
B a
si c
v a
ri a
b le
s M a le
0 .6 1
2 0 .2 6 8 3 7 4 * *
2 0 .1 9 8 8 9 3 *
0 .8 9 7 3 0 8
0 .8 7 1 1 6 8
M a rr ie d
0 .9 0
2 0 .0 5 9 8 1
0 .0 4 6 9 3 3
1 .2 7 0 8 8 1
1 .1 6 4 3 4 9
M a le
a n d m a rr ie d
0 .5 5
0 .3 4 9 1 2 8 * * *
0 .2 4 3 2 1 9 * *
2 0 .6 9 6 7 0 1
2 0 .6 9 0 3 3 9
In g o o d h e a lt h
0 .9 0
0 .1 2 3 0 8 6 * *
0 .1 2 9 4 2 7
0 .0 7 6 2 1 1
0 .0 9 8 6 0 6
D u ra ti o n o f u rb a n re si d e n c e (y e a rs )
7 .5 1
0 .0 1 3 5 8 0 *
0 .0 0 8 7 3 1
0 .0 1 9 4 8 6 *
0 .0 1 6 4 1 6
D u ra ti o n o f u rb a n re si d e n c e , sq u a re d
8 4 .8 3
2 0 .0 0 0 5 4 7 *
2 0 .0 0 0 3 9 1
2 0 .0 0 0 8 4 8 * *
2 0 .0 0 0 7 6 8 *
C o
n v e n
ti o
n a
l e c o
n o
m ic
v a
ri a
b le
s L o g o f p e r c a p it a h o u se h o ld
in c o m e
8 .5 5
0 .2 0 8 1 0 2 * * *
0 .1 8 6 2 8 6 * * *
0 .6 3 4 2 0 8 * * *
0 .6 3 5 4 8 7 * * *
N e t fi n a n c ia l a ss e ts (’ 0 0 0 y u a n )
1 6 .5 1
2 0 .0 0 0 2 4 7
0 .0 0 0 3 4 9
2 0 .0 0 1 6 2 2 *
2 0 .0 0 1 7 1 9 *
W o rk in g h o u rs
(’ 0 0 p e r y e a r)
3 1 .9 4
0 .0 0 0 0 9 3
0 .0 0 0 5 8 1
0 .0 0 1 8 4 2
0 .0 0 3 4 2 4 *
C o
m p
a ri
so n
v a
ri a
b le
s E x p e c t b ig
in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .0 7
0 .2 9 8 3 9 8 * * *
0 .2 4 5 2 0 7 * *
0 .2 7 2 6 2 9 * *
0 .2 1 2 3 4 5 *
E x p e c t sm
a ll in c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .5 5
0 .0 2 6 1 7 6
0 .0 0 5 9 7 7
0 .0 3 1 9 4 8
0 .0 1 3 4 3 0
E x p e c t d e c re a se
in in c o m e o v e r n e x t 5 y e a rs
0 .1 0
2 0 .4 0 3 2 9 9 * * *
2 0 .3 8 3 0 0 4 * * *
2 0 .3 4 1 6 2 3 * * *
2 0 .3 2 4 7 8 5 * * *
L o g o f a v e ra g e p e r c a p it a u rb a n in c o m e in
c it y o f
c u rr e n t re si d e n c e
8 .9 7
2 0 .1 2 0 4 3 2
2 0 .1 3 4 5 6 4
2 0 .3 2 6 7 6 7 * *
2 0 .3 2 5 6 4 4 * *
H a
rs h
n e ss
o f
c it
y li
fe v a
ri a
b le
s L iv in g w it h fa m il y m e m b e rs
0 .8 8
0 .1 3 4 7 2 6
0 .1 4 7 5 4 2 *
N u m b e r o f re la ti v e s a n d fr ie n d s in
c it y
7 .1 9
0 .0 0 3 8 6 9 *
0 .0 0 2 6 5 8
0 .0 0 1 8 1 0
0 .0 0 1 5 8 1
C h il d st il l in
v il la g e
0 .3 2
2 0 .1 2 4 9 7 7 * *
2 0 .1 2 7 7 2 3 * *
2 0 .2 1 0 3 4 6 * * *
2 0 .2 1 3 7 0 7 * * *
N o h e a ti n g
0 .6 5
2 0 .1 4 9 8 6 5 * *
2 0 .1 3 8 5 2 1 * *
2 0 .1 8 2 6 3 1 * * *
2 0 .1 9 6 8 9 0 * * *
S a ti sf a c ti o n w it h jo b
1 .9 8
0 .0 7 3 5 2 7 *
0 .0 6 6 5 8 9 * *
In d e x o f d is c ri m in a ti o n
5 .3 5
2 0 .0 3 2 1 9 6 * * *
2 0 .0 2 9 6 9 6 * * *
C a n fi n d a n o th e r jo b in
2 w e e k s
0 .1 1
2 0 .0 9 9 6 7 6
2 0 .1 8 1 5 7 8 * *
C a n fi n d a n o th e r jo b in
1 m o n th
0 .2 3
2 0 .1 2 1 3 3 9 * *
2 0 .2 2 1 8 3 4 * * *
C a n fi n d a n o th e r jo b in
2 m o n th s
0 .1 0
2 0 .1 4 7 8 2 0 *
2 0 .1 7 0 0 8 0 *
C a n fi n d a n o th e r jo b in
6 m o n th s
0 .1 3
2 0 .1 9 1 7 0 4 * *
2 0 .2 0 0 8 1 3 * *
(C o
n ti
n u
e s)
Does Economic Growth Raise Happiness in China? 13
T a b le
4 .
C o
n ti
n u
e d
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
N e e d m o re
th a n 6 m o n th s to
fi n d a n o th e r jo b
0 .1 7
2 0 .2 1 4 0 1 2 * * *
2 0 .2 0 8 3 9 5 * *
C o n st a n t
1 .0 2 4 8 0 8
1 .5 3 6 9 1 6
2 1 .3 4 9 4 1 5
2 0 .7 2 0 1 1 5
R -s q u a re d /c e n tr e d
R -s q u a re d
0 .1 0 0
0 .1 2 9
0 .0 4 6
0 .0 7 0
N u m b e r o f o b se rv a ti o n s
1 8 5 0
1 7 1 5
1 1 1 5
1 1 0 0
S ig n ifi c a n c e o f in st ru m e n ts in
fi rs t st a g e e q u a ti o n
M o th e r’ s y e a rs
o f e d u c a ti o n
* *
* *
S p o u se ’s
y e a rs
o f e d u c a ti o n
* * *
* * *
E a rn in g s p e r m o n th
b e fo re
m ig ra ti n g
* * *
* * *
F -t e st o f e x c lu d e d in st ru m e n ts ( p -v a l)
0 .0 0 0 0
0 .0 0 0 0
S a rg a n te st , fo r o v e r- id e n ti fi c a ti o n o f a ll
in st ru m e n ts ( p -v a l)
0 .5 2 0 7
0 .6 3 0 0
A n d e rs o n -R u b in
te st o f jo in t si g n ifi c a n c e o f
e n d o g e n o u s re g re ss o rs
in m a in
e q u a ti o n ,
F -t e st ( p -v a l)
0 .0 1 3 0
0 .0 2 8 3
N o
te s:
1 . D e p e n d e n t v a ri a b le : sc o re
o f h a p p in e ss
b a se d o n c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts
a s fo ll o w s:
v e ry
h a p p y ¼
4 ; h a p p y ¼
3 ; so -s o ¼
2 ; n o t
h a p p y ¼
1 a n d n o t a t a ll h a p p y ¼
0 .
2 . M o d e ls (1 ) a n d (3 ) a re
fo r th e fu ll sa m p le . M o d e ls (2 ) a n d (4 ) a re
fo r th e su b sa m p le
o f e m p lo y e d p e rs o n s.
3 . T h e o m it te d c a te g o ri e s in
th e d u m m y v a ri a b le
a n a ly se s a re : si n g le
fe m a le ; n o t h e a lt h y ; n o c h a n g e in
in c o m e e x p e c te d in
th e n e x t 5 y e a rs .
4 . * * * , * * a n d * d e n o te
st a ti st ic a l si g n ifi c a n c e a t th e 1 % , 5 %
a n d 1 0 %
le v e ls , re sp e c ti v e ly .
5 . M o d e ls h a v e b e e n c lu st e re d a t c it y le v e l fo r ro b u st st a n d a rd
e rr o rs .
6 . In st ru m e n te d v a ri a b le s re g re ss io n re su lt s a re
g e n e ra te d u si n g th e B a u m
e t
a l. (2 0 0 3 ) iv re g 2 .a d o p ro g ra m
fo r S ta ta .
S o
u rc
e : K n ig h t &
G u n a ti la k a (2 0 1 0 a , T a b le s 2 a n d 3 ).
14 J. Knight & R. Gunatilaka
When migrants who reported that they were unhappy or not at all happy were asked the
reason for their unhappiness, over two-thirds said that their income was too low. This
pointer to the possible importance of perceptions of relative deprivation is confirmed by
the negative, large and significant coefficient on per capita income of urban residents in the
destination province. This effect is stronger for the migrants who had been in the city for
more than the median length of time, 7.5 years (Knight & Gunatilaka, 2010a, Table 2).
The migrants appear to compare their own situations with those of others living in their
new surroundings, and to do so increasingly as they become more settled.
China’s political economy accords urban hukou people a set of rights and privileges
that are denied to rural hukou people residing in the cities (Knight & Song, 1999, 2005).
The migrants are generally “second-class citizens”. When we included various proxies for
these disadvantages in the happiness function, job dissatisfaction, perceptions of
discrimination against migrants and measures of job insecurity had significant negative
coefficients. The unsatisfactory conditions in which migrants live and the unpleasant and
insecure nature of their employment depress happiness.
We explored the reasons why migrants were on average less happy than peasants by
conducting a decomposition analysis using happiness functions with identical
determinants for the migrant and rural subsamples (Table 5). The objective was to
explain the migrant shortfall in mean happiness score, equal to 0.31, distinguishing
between the contributions of the different mean values of the explanatory variables and
those of their coefficients. The effect of characteristics was actually to increase the
Table 5. Decomposition of the difference in mean happiness score between rural–urban migrants and rural residents: percentage contribution to the difference
Using the rural happiness function
Using the migrants’ happiness function
Due to characteristics
Due to coefficients
Due to characteristics
Due to coefficients
Ln income per capita 255.51 1.13 255.39 1.01 Health 226.39 114.99 25.81 94.41 Income expectations 14.71 32.98 11.34 36.36 Age 13.97 2138.82 6.69 2131.54 Education 22.55 22.61 20.13 20.18 Male 24.70 223.87 0.74 229.30 Marital status 2.49 21.82 0.89 20.22 Ethnicity 1.10 2.12 0.13 3.10 Communist Party 5.01 1.38 0.40 5.99 Unemployment 0.09 0.02 0.10 0.02 Working hours 16.65 223.94 5.53 212.81 Net financial assets 213.43 21.28 0.29 7.56 Constant term 0.00 140.48 0.00 140.48 Sum (percentage) 248.56 148.56 235.23 135.23 Sum (score) 20.1485 0.4544 20.1078 0.4137
Notes: The mean happiness scores are 2.6764 in the case of rural residents and 2.3703 in the case of migrants, creating a migrant shortfall of 0.3061 (set equal to þ100%) to be explained by the decomposition. Thus, the combined contributions of characteristics and coefficients sum to 100%. The composite variables are age and age squared for age, married, single, divorced and widowed for marital status, and big increase, small increase and decrease for income expectations.
Source: Knight & Gunatilaka (2010a, Table 5).
Does Economic Growth Raise Happiness in China? 15
difference in mean happiness scores: in particular, the migrants had higher mean income.
The explanation was therefore to be found in the superior happiness function of rural
people. Here the expectations of future income were crucial. With static income as the
reference category, the coefficients of the migrants were uniformly lower, suggesting that
migrants had higher income aspirations relative to their current income. This can be
expected if aspirations depend on the income of the relevant comparator group. The rural
respondents are representative of rural society and so their mean income is close to the
mean income of their likely comparator group; but the migrant subsample is
unrepresentative of urban society: migrants tend to occupy the lower ranges of the urban
income distribution. If migrants make comparisons with urban-born residents as well as
with other migrants, their aspirations will be high in relation to their current income.
An equivalent exercise was conducted to decompose the difference in mean happiness
between migrants and urban hukou residents (Table 6), the migrant shortfall in happiness
score being 0.11. In this case the difference in coefficients makes no net contribution to the
explanation. Two differences in mean characteristics can explain all of the difference: the
higher mean income of urban residents and their superior position in the urban income
Table 6. Decomposition of the difference in mean happiness score between rural–urban migrants and urban residents: percentage contribution to the difference
Using the urban happiness function
Using the migrants’ happiness function
Due to characteristics
Due to coefficients
Due to characteristics
Due to coefficients
Ln income per capita 43.20 457.57 28.15 472.62 Income expectations 247.03 66.43 239.92 59.32 Living standard in second highest quarter in city
216.81 9.40 233.68 26.28
Living standard in third highest quarter in city
28.19 74.32 211.71 77.84
Living standard in lowest quarter in city
194.35 226.79 175.93 28.37
Age 1.52 2562.72 32.85 2594.05 Male 11.53 262.39 24.08 246.78 Education 28.65 8.22 211.54 11.11 Marital status 0.18 2.63 21.96 4.77 Ethnicity 22.12 3.19 20.34 1.40 Communist Party 15.69 1.00 7.63 9.06 Unemployment 26.68 22.01 20.68 28.01 Health 254.21 78.08 228.01 51.89 Working hours 21.08 22.20 10.50 10.62 Net financial assets 1.69 3.85 22.46 8.01 Constant term 0.00 296.38 0.00 296.38 Sum (percentage) 123.41 223.41 120.67 220.67 Sum (score) 0.1372 20.0260 0.1342 20.0230
Notes: The mean happiness scores are 2.4845 in the case of urban residents and 2.3703 in the case of migrants, creating a migrant shortfall of 0.1143 (set equal to þ 100%) to be explained by the decomposition. Thus, the combined contributions of characteristics and coefficients sum to 100%. The composite variables are age and age squared for age, married, single, divorced and widowed for marital status, and big increase, small increase and decrease for income expectations.
Source: Knight & Gunatilaka (2010a, Table 4).
16 J. Knight & R. Gunatilaka
distribution. Position in the city income distribution has a powerful effect on happiness,
and this is true for both samples. A far higher proportion of migrants than of urban
residents fall into the lowest quarter of city households in terms of living standards. If the
income of the relevant comparator group influences aspirations, the inferior position of
migrants in the city income distribution can thus explain why they appear to have higher
aspirations in relation to their current income.
There might be selection on the basis of unobserved characteristics. For instance,
migrants might be inherently unhappy people who have unsuccessfully sought happiness
through migration. Our test was to use the residual (actual minus predicted) happiness
score as a proxy for inherent disposition and to introduce this variable into a probit
equation predicting that the migrant reported urban living to yield more happiness than
rural living. The coefficient was significantly positive and large, implying that the
unobserved characteristic was acquired after migration. Thus, this explanation lacked
empirical support (Knight & Gunatilaka, 2010a, pp. 121–122).
8. Aspiration Income and Happiness
The argument of this paper has centred on peoples’ aspirations in relation to their income,
and yet the evidence has been only indirect. We should ideally measure aspirations, or at
least aspirations for income. There is a proxy for “aspiration income” in the rural data set,
which is analysed in Knight & Gunatilaka (2009). Respondents were asked: “What is the
minimum income needed to sustain the household for a year?” It was possible to justify
this as a proxy for income aspirations. The strategy was first to analyse its determinants
and then to include income aspirations as an additional argument in the happiness
functions for the rural sample that had previously been estimated.
Table 7 reports the determinants of income need, with ln household income need as the
dependent variable. Column 1 shows the OLS and column 2 the IV estimation with ln
household income instrumented. Among the demographic and physiological determinants
of income need, we see that good health (reducing income need), satisfaction with the
village clinic (reducing income need) and size and composition of the household are
important, and that the age, sex and marital status of the respondent may be. The equations
contain several variables that might influence aspirations for income. In particular, the
coefficient on ln household income is both positive and significant. The coefficient is 0.19
(OLS) and 0.57 (IV), i.e. a doubling of actual income increases the perceived minimum
income by 19 or 57%, respectively. 4 Years of education also has a significantly positive
coefficient: the more education the respondent had received, the higher the income needed.
With static living standard as the base category, those whose current living standard is
worse than 5 years ago have a significantly higher aspiration for income. By contrast,
financial assets may have a negative effect, i.e. more wealth appears to provide security
rather than to raise aspirations. Those whose main reference group is outside the village
have higher aspiration income, as do households whose income is below their village’s
average household income.
At the second stage, we added the aspiration income variable to the function estimating
happiness, again converted into a cardinal score. Table 8 shows OLS and IV estimates,
columns 1 and 3 having ln per capita income need as the only aspiration variable and
columns 2 and 4 having a full list of aspiration variables, respectively. Both ln income per
capita and ln income needed per capita are instrumented. When other variables that are likely
Does Economic Growth Raise Happiness in China? 17
Table 7. Determinants of income need: OLS and IV estimates
OLS IV
Mean or proportion (1) (2)
Log of total household income 8.97 0.189531*** 0.568303***
Aspiration variables Net financial assets (’000 yuan) 5.52 20.001471 20.005687*** Current living standards better than 5 years ago 0.60 20.002653 20.024994 Current living standards worse than 5 years ago 0.05 0.200728*** 0.209451*** Education (years) 7.14 0.025571*** 0.023523*** Main reference group beyond village 0.11 0.100161** 0.085029* Household income much above village average 0.02 0.070678 0.009413 Household income above village average 0.18 0.052439* 0.004858 Household income below village average 0.20 0.053649* 0.125806*** Household income much below village average 0.03 0.188881** 0.365798***
Conditioning variables Age 45.26 0.023823*** 0.009860 Age squared 2159.52 20.000281*** 20.000121 Male 0.75 20.052313** 20.013750 Married 0.95 0.176272** 0.183060* Divorced 0.00 0.225053 0.257623 Widowed 0.02 0.13155 0.068548 In good health 0.74 20.080823*** 20.085976*** Satisfaction with clinic 2.34 20.049787*** 20.058235*** Senior citizens, male, aged 65 þ 0.10 0.174814*** 0.165848*** Senior citizens, female, aged 65 þ 0.11 0.116292*** 0.099134** Adult males of 18–64 years 1.48 0.128360*** 0.083446*** Adult females of 18–64 years 1.39 0.142299*** 0.101945*** Teenage males of 11–17 years 0.36 0.107355*** 0.067175*** Teenage females of 11–17 years 0.31 0.105044*** 0.079474*** Children, males less than 11 years 0.27 0.024714 0.005915 Children, females less than 11 years 0.22 0.034355 0.022870 Constant 5.410067*** 2.491840* R-squared 0.14 Number of observations 6231 5356 Significance of exclusion restrictions in first stage equation Father’s years of education * Productive assets ***
F-test of excluded instruments ( p-val) 0.0000 Hansen J-test for over-identification of all instruments ( p-val)
0.7236
Anderson-Rubin test of joint significance of endogenous regressors in main equation, F-test ( p-val)
0.0296
Notes: 1. Dependent variables: logarithm of minimum income needed (mean 8.455, standard deviation 0.731). 2. Independent variables with cardinal values assigned to qualitative assessments so that greater intensity is represented by a higher value are: satisfaction with clinic.
3. The omitted categories in the dummy variable analyses are: female sex;married; not healthy; current living standard the same as 5 years ago; main reference group within village; household at average village income.
4. ***, ** and * denote statistical significance at the 1%, 5% and 10% levels, respectively. 5. Instrumented variables regression results are generated using the Baum et al. (2003) ivreg2.ado program for Stata.
6. The models have been clustered at village level for robust standard errors. Source: Knight & Gunatilaka (2009, Tables 3 and 4). Not all explanatory variables contained in those
tables are reported here.
18 J. Knight & R. Gunatilaka
T a b le
8 . D e te rm
in a n ts o f h a p p in e ss : O L S a n d IV
e st im
a te s
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
L o g o f p e r c a p it a h o u se h o ld
in c o m e (’ 0 0 0 y u a n )
7 .5 8
0 .2 3 4 7 5 9 * * *
0 .1 5 3 4 2 8 * * *
1 .1 0 4 8 7 3
0 .3 0 7 6 4 6
A sp
ir a ti
o n
v a ri
a b le
s L o g o f p e r c a p it a m in im
u m
in c o m e n e e d e d
7 .0 7
2 0 .0 8 1 3 8 1 * * *
2 0 .0 6 3 2 6 5 * * *
2 0 .3 4 6 7 4 3
2 0 .0 9 4 0 6 8
A g re e m e n t w it h st a te m e n t th a t m o n e y is im
p o rt a n t
2 .3 5
2 0 .0 3 6 9 8 1 * * *
2 0 .0 5 1 3 6 6 * *
H o u se h o ld
in c o m e m u c h a b o v e v il la g e a v e ra g e
0 .0 2
0 .2 5 9 1 8 9 * * *
0 .1 9 4 9 2 6 * *
H o u se h o ld
in c o m e a b o v e v il la g e a v e ra g e
0 .1 8
0 .0 8 2 0 6 9 * * *
0 .0 5 4 8 7 2 *
H o u se h o ld
in c o m e b e lo w
v il la g e a v e ra g e
0 .2 0
2 0 .3 2 1 8 2 3 * * *
2 0 .2 9 2 9 4 2 * * *
H o u se h o ld
in c o m e m u c h b e lo w
v il la g e a v e ra g e
0 .0 3
2 0 .8 0 2 9 5 2 * * *
2 0 .7 3 7 0 8 1 * * *
C u rr e n t li v in g st a n d a rd s b e tt e r th a n 5 y e a rs
a g o
0 .6 0
0 .2 0 1 3 0 2 * * *
0 .1 7 9 9 1 9 * * *
C u rr e n t li v in g st a n d a rd s w o rs e th a n 5 y e a rs
a g o
0 .0 5
2 0 .0 8 9 7 0 2 * *
2 0 .0 7 8 6 3
B a si
c v a ri
a b le
s A g e (y e a rs )
4 5 .2 6
2 0 .0 0 9 8 8 9
2 0 .0 1 5 1 4 5 * *
2 0 .0 2 2 5 3 0 *
2 0 .0 1 9 2 0 3 * * *
A g e sq u a re d
2 1 5 9 .5 2
0 .0 0 0 1 7 0 * *
0 .0 0 0 2 3 2 * * *
0 .0 0 0 2 6 3 * *
0 .0 0 0 2 7 4 * * *
M a le
0 .7 5
2 0 .0 8 3 3 5 1 * * *
2 0 .0 8 2 5 3 0 * * *
2 0 .0 2 8 2 2 2
2 0 .0 6 6 4 6 7 * *
M a rr ie d
0 .9 5
0 .1 4 8 8 1 8 * *
0 .1 3 9 8 2 1 * *
0 .1 3 3 1 1 2
D iv o rc e d
0 .0 0
2 0 .4 2 3 3 3 6 * *
2 0 .4 7 7 5 4 4 * * *
2 0 .4 8 3 5 4 7 * *
W id o w e d
0 .0 2
2 0 .2 9 9 2 0 7 * * *
2 0 .2 2 0 0 5 3 * *
2 0 .2 2 8 8 5 6 * *
W o rk in g h o u rs
(’ 0 0 p e r y e a r)
1 7 .0 7
2 0 .0 0 3 0 7 2 * * *
2 0 .0 0 2 3 9 1 * *
2 0 .0 0 7 5 7 9
2 0 .0 0 2 8 7 7 * *
N e t fi n a n c ia l a ss e ts (’ 0 0 0 y u a n )
5 .5 2
0 .0 0 1 5 7 6 * *
0 .0 0 1 3 8 4 *
2 0 .0 1 0 8 8
2 0 .0 0 0 8 1 1
In g o o d h e a lt h
0 .7 4
0 .4 2 3 0 1 8 * * *
0 .3 4 4 8 0 0 * * *
0 .3 7 8 4 1 5 *
0 .3 4 4 2 4 5 * * *
C o n st a n t
0 .7 0 1 4 4 9 * * *
1 .4 6 7 2 6 8 * * *
2 3 .1 4 3 9 1 1
0 .6 8 9 2 9 5
R -s q u a re d
0 .2 3 1
0 .3 0 8
N u m b e r o f o b se rv a ti o n s
6 6 1 7
6 5 3 8
3 8 9 6
5 6 2 0
S ig n ifi c a n c e o f e x c lu si o n re st ri c ti o n s fo r ln
o f p e r
c a p it a h o u se h o ld
in c o m e in
fi rs t st a g e e q u a ti o n
F a th e r’ s y e a rs
o f e d u c a ti o n
* * *
S p o u se ’s
y e a rs
o f e d u c a ti o n
* * *
S e n io r c it iz e n s, m a le , a g e d 6 5 þ
* * *
S e n io r c it iz e n s, fe m a le , a g e d 6 5 þ
* * *
A d u lt m a le s o f 1 8 – 6 4 y e a rs
* * *
(C o n ti
n u e s)
Does Economic Growth Raise Happiness in China? 19
T a b le
8 .
C o n ti
n u e d
(1 )
(2 )
(3 )
(4 )
M e a n o r p ro p o rt io n
O L S
O L S
IV IV
A d u lt fe m a le s o f 1 8 – 6 4 y e a rs
* * *
T e e n a g e m a le s o f 1 1 – 1 7 y e a rs
* * *
T e e n a g e fe m a le s o f 1 1 – 1 7 y e a rs
* * *
C h il d re n , m a le s le ss
th a n 1 1 y e a rs
* * *
C h il d re n , fe m a le s le ss
th a n 1 1 y e a rs
* * *
F -t e st o f e x c lu d e d in st ru m e n ts ( p -v a l)
0 .0 0 0 0
0 .0 0 0 0
S ig n ifi c a n c e o f e x c lu si o n re st ri c ti o n s fo r ln
o f n e e d
m in im
u m
in c o m e in
fi rs t st a g e e q u a ti o n
F a th e r’ s y e a rs
o f e d u c a ti o n
n s
S p o u se ’s
y e a rs
o f e d u c a ti o n
* * *
S e n io r c it iz e n s, m a le , a g e d 6 5 þ
n s
S e n io r c it iz e n s, fe m a le , a g e d 6 5 þ
* * *
A d u lt m a le s o f 1 8 – 6 4 y e a rs
* * *
A d u lt fe m a le s o f 1 8 – 6 4 y e a rs
* * *
T e e n a g e m a le s o f 1 1 – 1 7 y e a rs
* * *
T e e n a g e fe m a le s o f 1 1 – 1 7 y e a rs
* * *
C h il d re n , m a le s le ss
th a n 1 1 y e a rs
* * *
C h il d re n , fe m a le s le ss
th a n 1 1 y e a rs
* * *
F -t e st o f e x c lu d e d in st ru m e n ts ( p -v a l)
0 .0 0 0 0
0 .0 0 0 0
S a rg a n te st fo r o v e r- id e n ti fi c a ti o n o f a ll in st ru m e n ts ( p -v a l)
0 .2 2 1 3
A n d e rs o n -R u b in
te st o f jo in t si g n ifi c a n c e o f e n d o g e n o u s
re g re ss o rs in
m a in
e q u a ti o n , F -t e st ( p -v a l)
0 .0 0 0 9
0 .0 0 1 1
N o
te s:
1 . D e p e n d e n t v a ri a b le s:
sc o re
o f h a p p in e ss
b a se d o n c a rd in a l v a lu e s a ss ig n e d to
q u a li ta ti v e a ss e ss m e n ts
a s fo ll o w s:
v e ry
h a p p y ¼
4 ; h a p p y ¼
3 ; so -s o ¼
2 ; n o t
h a p p y ¼
1 a n d n o t a t a ll h a p p y ¼
0 . T h e m e a n v a lu e o f h a p p in e ss
is 2 .6 3 , st a n d a rd
d e v ia ti o n 0 .8 8 , a n d c o e ffi c ie n t o f v a ri a ti o n 0 .3 3 .
2 . In d e p e n d e n t v a ri a b le s w it h c a rd in a l v a lu e s a ss ig n e d to q u a li ta ti v e a ss e ss m e n ts so
th a t a h ig h e r v a lu e d e n o te s g re a te r in te n si ty : a g re e m e n t th a t m o n e y is im
p o rt a n t.
3 . T h e o m it te d c a te g o ri e s in
th e d u m m y v a ri a b le
a n a ly se s a re : fe m a le
se x ; m a rr ie d ; n o t h e a lt h y ; h o u se h o ld
a t a v e ra g e v il la g e in c o m e ; c u rr e n t li v in g st a n d a rd
th e
sa m e a s 5 y e a rs
a g o .
4 . * * * , * * a n d * d e n o te
st a ti st ic a l si g n ifi c a n c e a t th e 1 % , 5 %
a n d 1 0 %
le v e ls , re sp e c ti v e ly .
5 . V a ri a b le s re la te d to
m a ri ta l st a te s h a v e b e e n e x c lu d e d fr o m
th e IV
sp e c ifi c a ti o n in
M o d e l (3 ) b e c a u se
sp o u se ’s
y e a rs
o f e d u c a ti o n is u se d a s a n in st ru m e n t.
6 . In st ru m e n te d v a ri a b le s re g re ss io n re su lt s a re
g e n e ra te d u si n g th e B a u m
e t
a l. (2 0 0 3 ) iv re g 2 .a d o p ro g ra m
fo r S ta ta .
S o
u rc
e : K n ig h t &
G u n a ti la k a (2 0 0 9 , T a b le s 5 a n d 6 ). N o t a ll e x p la n a to ry
v a ri a b le s c o n ta in e d in
th o se
ta b le s a re
re p o rt e d h e re .
20 J. Knight & R. Gunatilaka
to represent aspirations are introduced as well as minimum income need, they generally have
significant coefficients. However, our particular interest here lies in aspiration income. As
expected, the coefficient on ln household income per capita is significantly positive in the
OLS specifications, with an average value of about 0.20; the coefficients are higher but not
significantly positive in the less precise IV estimates. Ln minimum income needed has a
significantly negative coefficient, averaging 20.07 in the OLS estimates; it is more negative
but not significant in the IV estimates. 5 Although the conventional statistical tests of good
instruments are passed, it is not possible to find a set of instruments that reliably identify the
separate effects of the two income variables. Nevertheless, this set of results provides direct,
albeit only suggestive, evidence that, other things being equal, having higher aspirations for
income can reduce happiness. Moreover, a comparison of the positive and negative
coefficients suggests that people run on a partial “hedonic treadmill”.
9. Conclusion
We are now in a position to consider whether these sets of results, all based on cross-
section evidence, can provide an answer to and explanation for the time-series question
posed in the title: Does economic growth raise happiness in China?
First, in all three data sets—the rural, the urban and the migrant—current income has a
positive and significant effect on happiness. However, in none of these subsamples is the
coefficient on current income substantively large. Clearly, there are other, more important,
determinants of individual subjective well-being.
Second, this pure effect of individual income level is further weakened by the fact that
economic growth will tend to raise incomes generally. In so far as the income of the
reference group rises as well as own income, the decline in relative income reduces
individual happiness. As the economy grows, it is important to “keep up with the Zhous”.
Third, the higher the incomes to which people aspire, the lower their subjective
well-being.
Fourth, aspirations are influenced by peoples’ reference groups and reference times. For
rural people, the reference group is generally their fellow villagers, for urban people it is
their fellow citizens within the city, and for rural–urban migrants, it is also other people
living in the city, urban as well as rural hukou holders. It is not the income of “any old
Zhou” that produces feelings of relative deprivation but the income of the “Zhous you
know”—those who fall into a person’s reference group.
Fifth, China’s national Gini coefficient of household income per capita rose from 0.39 in
1988 to 0.47 in 2002 (Gustaffson et al., 2008, p. 19). It is likely that this rising income
inequality reduced happiness, but the relationship is complicated by the importance of
local reference groups and the possibility of demonstration effects as well as relative
deprivation effects.
Sixth, aspirations for income are much influenced by reference time income, and this is
governed primarily by the present. It is current income—both absolute and relative—that
mainly determines aspirations for income. However, there appears also to be a ratchet
effect: previous income can also influence aspirations, so that experience of a past fall in
income reduces happiness, other things being equal. In general terms, the analysis
highlights the important role that aspirations play in peoples’ perceptions of their own
well-being.
Does Economic Growth Raise Happiness in China? 21
Seventh, expectations of future income are important for current happiness. This
suggests that a gloomier view of the economy’s prospects could be serious for well-being,
and maybe even for political stability.
Using this framework of empirical findings, we can see that the changes in the economy
and in the society that stem from, or go along with, economic growth are likely to have
implications for overall happiness in China. The effects of income growth itself are limited
because of the resultant growth in aspirations, this being a function of both own and
relative income. The importance of relative income for subjective well-being in all three
subsamples, together with rising income inequality over time, helps to explain the failure
of happiness scores to rise with income levels. The new urban insecurities and
uncertainties generated by economic reform and marketization have a negative impact on
the subjective well-being of the growing number of urban residents. In particular,
rural–urban migrants—rapidly expanding in number—suffer both from their second-class
status in the cities and from the widening of their reference groups to include the more
affluent urban hukou population. By extending the reference groups of rural-dwellers
beyond the village, migration can also have the effect of reducing rural happiness. These
findings help to explain why mean happiness in China appears not to have risen, and may
even have fallen, in recent years.
Does China’s experience apply generally in developing countries? Easterlin &
Sawangfa (2010) made a careful descriptive study of 12 developing countries for which
sufficiently long and comparable time-series data were available. All 12 experienced
rising real income per capita. Three (China, India and Chile) appeared to experience a fall
in their average satisfaction with life, whereas nine showed a rise in their score, although
in only two (Mexico and Venezuela) was the rise statistically significant. However, the
change in the score was not positively related to the growth of income per capita, and in
every case the change in the score fell short of that predicted by the cross-section
relationship within a country.
Is China’s experience common among transition economies? Easterlin (2008) examined
happiness scores in 13 ex-communist countries of Eastern Europe. Happiness collapsed
when their economies collapsed, but it failed to recover commensurately with income.
This was attributed to the non-income changes that accompanied the transition to
capitalism, such as rising unemployment, inequality and insecurity. China’s gradualist
reform avoided economic collapse but, as in other transition economies, the socio-
economic changes accompanying economic transition appeared to reduce happiness.
Our analysis raises, and also illuminates, some basic normative and policy issues.
To what extent should subjective well-being enter into the social welfare function and be
accepted as one of the criteria for policy-making? Ultimately, a value judgement is
required. Powerful and plausible regularities were observed in the analysis. Thus, in
making that value judgement, it is difficult simply to dismiss as irrelevant peoples’
reported perceptions of their own welfare. There are some difficult policy trade-offs
between the gains from economic growth and the losses from the socio-economic changes
accompanying growth, and these have not been sufficiently recognized.
For over a quarter of a century China’s reformist policy-makers gave the highest priority
to the achievement of rapid economic growth. In the last 5 years, however, the balance of
policy objectives has moved somewhat in the direction of creating a “harmonious society”,
for instance, showing greater concern for reducing income inequality and for improving
social security. That move can be seen as a response to the issues that underlie this paper.
22 J. Knight & R. Gunatilaka
Notes
1 The task was to find variables that were well correlated with the income variable but for which it was
plausible that they made no direct contribution to happiness. The instruments chosen (father’s education
and spouse’s education, in years) were unlikely to affect own happiness (even own education had a
significant positive effect only in the most basic specification). The F-test of excluded instruments
shows that the instruments are not weak ( p-values 0.000 and 0.0054), the Sargan/Hansen over-
identification test (if one instrument is exogenous then at least one other is exogenous) is passed
( p-value . 0.96), and the Anderson-Rubin-Wald test (of joint significance of the endogenous
regressors) suggests that instrumenting is necessary ( p-value ¼ 0.0004) in column 3 but may not be
necessary ( p-value ¼ 0.3834) when more regressors are added in column 4. 2 The instruments listed in the notes to the table (combinations of parental or household characteristics)
seem unlikely to affect happiness directly; they are not weak; they pass the Sargan test; however, the
Anderson-Rubin test fails to provide evidence of endogeneity. 3 The instruments for ln household income per capita (mother’s education, spouse’s education and
earnings per month before migrating) were unlikely to affect current happiness; they passed the
statistical tests of relevance and validity and, according to the Anderson-Rubin test, they were needed. 4 The instruments for ln total household income (father’s education and productive assets, i.e. rural
household machinery and equipment) were in themselves unlikely to influence the household’s
perceived need for income. According to the test results shown in the table, they were relevant and
valid, and instrumenting was necessary. 5 In our view, the instruments (combinations of father’s education, spouse’s education and household
composition) were unlikely to affect happiness directly. They were highly correlated with both ln
household per capita actual income and needed income: in each case the F-test of excluded instruments
was highly significant ( p-value ¼ 0.000). The Sargan test for over-identification of all instruments was
passed, and the Anderson-Rubin test indicated that instrumenting was needed.
References
Baum, C., Schaffer, M. & Stillman, S. (2003) Instrumental variables and GMM: estimation and testing,
Stata Journal, 3(1), pp. 1–31.
Blanchflower, D. & Oswald, A. (2004) Well-being over time in Britain and the USA, Journal of Public
Economics, 88(7–8), pp. 1359–1386.
Clark, A. & Oswald, A. (1998) Comparison-concave utility and following behaviour in social and economic
settings, Journal of Public Economics, 71(1), pp. 133–155.
Clark, A., Frijters, P. & Shields, M. (2008) Relative income, happiness and utility: an explanation for the Easterlin
paradox and other puzzles, Journal of Economic Literature, 46(1), pp. 95–144.
Di Tella, R., Haisken-DeNew, J. & MacCulloch, R. (2007) Happiness, Adaptation to Income and to Status in an
Individual Panel, NBER Working Paper 13159.
Di Tella, R., MacCulloch, R. & Oswald, A. (2003) The macroeconomics of happiness, Review of Economics and
Statistics, 85(4), pp. 809–827.
Durkheim, E. (1897 [1952]) Suicide. A Study in Sociology, translated by J.A. Spaulding & G. Simpson
(London: Routledge and Kegan Paul).
Easterlin, R. (1974) Does economic growth improve the human lot? Some empirical evidence, in: P. David &
M. Reder (Eds) Nations and Households in Economic Growth: Essays in Honor of Moses Abramovitz,
pp. 98–125 (New York and London: Academic Press).
Easterlin, R. (1995) Will raising the incomes of all increase the happiness of all? Journal of Economic Behaviour
and Organization, 27(1), pp. 35–48.
Easterlin, R. (2001) Income and happiness: towards a unified theory, Economic Journal, 111(473), pp. 465–484.
Easterlin, R. (2008) Lost in transition: life satisfaction on the road to capitalism, Journal of Economic Behavior
and Organization, 71(2), pp. 130–145.
Easterlin, R. & Sawangfa, O. (2010) Happiness and economic growth: does the cross section evidence predict
time trends? Evidence from developing countries, in: E. Diener, J. Helliwell & D. Kahneman (Eds)
International Differences in Well-being (New York: Oxford University Press).
Ferrer-i-Carbonnel, A. & Frijters, P. (2004) How important is methodology for estimates of the determinants of
happiness? Economic Journal, 114(497), pp. 640–659.
Does Economic Growth Raise Happiness in China? 23
Frank, R. F. (1997) The frame of reference as a public good, Economic Journal, 107, pp. 1832–1847.
Frey, B. & Stutzer, A. (2002) What can economists learn from happiness research? Journal of Economic
Literature, 40(2), pp. 402–435.
Graham, C. & Felton, A. (2006) Inequality and happiness: insights from Latin America, Journal of Economic
Inequality, 4(1), pp. 107–122.
Gustaffson, B., Shi, L. & Sicular, T. (Eds) (2008) Inequality and Public Policy in China (Cambridge:
Cambridge University Press).
Hirschman, A. (1973) The changing tolerance for income inequality in the course of economic development,
Quarterly Journal of Economics, 87(4), pp. 544–566.
Kahneman, D. & Krueger, A. (2006) Developments in the measurement of subjective well-being, Journal of
Economic Perspectives, 20(1), pp. 3–24.
Kingdon, G. & Knight, J. (2007) Community, comparisons and subjective well-being in a divided society,
Journal of Economic Behavior and Organization, 64, pp. 69–90.
Knight, J. & Gunatilaka, R. (2010a) Great expectations? The subjective well-being of rural–urban migrants,
World Development, 38(1), pp. 113–124.
Knight, J. & Gunatilaka, R. (2010b) The rural–urban divide: Income but not happiness? Journal of Development
Studies, 46(3), pp. 506–534.
Knight, J. & Gunatilaka, R. (2009) Income, Aspirations and the Hedonic Treadmill in a Poor Society,
University of Oxford, Department of Economics Discussion Paper No. 468, December.
Knight, J. & Song, L. (1999) The Rural – Urban Divide. Economic Disparities and Interactions in China (Oxford:
Oxford University Press).
Knight, J. & Song, L. (2005) Towards a Labour Market in China (Oxford: Oxford University Press).
Knight, J., Song, L. & Gunatilaka, R. (2009) The determinants of subjective well-being in rural China,
China Economic Review, 20(4), pp. 635–649.
Luttmer, E. (2004) Neighbors as negatives, Relative earnings and well-being, Quarterly Journal of Economics,
120(3), pp. 963–1002.
Rabin, M. (1998) Psychology and economics, Journal of Economic Literature, 36, pp. 11–46.
Ravallion, M. & Chen, S. (2007) China’s (uneven) progress against poverty, Journal of Development Economics,
82(1), pp. 1–42.
Runciman, W. G. (1966) Relative Deprivation and Social Justice (Berkeley: University of California Press).
Senik, C. (2004) When information dominates comparison: learning from Russian subjective panel data,
Journal of Public Economics, 88, pp. 99–123.
Stutzer, A. (2004) The role of income aspirations in individual happiness, Journal of Economic Behavior and
Organization, 54(1), pp. 84–109.
United Nations Development Programme (2010) Human Development Report 2010 (New York: UNDP).
24 J. Knight & R. Gunatilaka
