Content uploaded by Eiji Yamamura
Author content
All content in this area was uploaded by Eiji Yamamura on Sep 05, 2014
Content may be subject to copyright.
1
Charitable giving under inequality aversion and social
capital
Eiji Yamamura
Department of Economics, Seinan Gakuin University
6-2-92 Nishijin, Sawara-ku, Fukuoka 814-8511, Japan
Tel: +81-(0)92-823-4543, Fax: +81-(0)92-823-2506, e-mail: yamaei@seinan-gu.ac.jp
Abstract
A Japanese General Social Survey is used to re-examine how voluntary giving
is associated with inequality aversion, and how the relationship differs between high-
and low-income groups. This paper also investigates how social capital influences that
relationship. The key findings are that (1) the level of voluntary giving increases with
inequality aversion for high-income groups, but not for low-income groups, and (2)
social capital accumulated in the respondent’s residential area reinforces the positive
influence of inequality aversion on voluntary giving for high-income groups only.
JEL classification:
H41, C34, C35, D63, Z13.
Keywords:
Inequality aversion, charitable contribution, trust, private provision of
public goods.
2
1. Introduction
Derin-Güre and Uler (2010) used a General Social Survey (GSS) conducted in the
United States to examine how an individual’s inequality aversion influenced charitable
giving. They found a positive association between inequality aversion and charitable
giving for high-income earners and a negative association for low-income earners.
In addition to an individual’s characteristics, the economic environment of an
individual’s residential area is thought to influence a person’s charitable behavior
(Fehr and Schmidt 1999). For instance, labor union member density is positively
related to per capita donations (Zullo 2011). Putnam (2000) stated that the degree of
community involvement is considered a type of social capital, which facilitates
coordination and cooperation to increase social welfare. In contrast, “an envious or
malicious person presumably would feel better off if some other persons become worse
off in certain respects. He could “harm” himself (i.e., spend his own resources) to harm
others” (Becker 1996, 190). Social interaction among people plays an important role in
terms of an individual’s decision making because the utility function of a person
includes the reactions of others to his/her actions (Becker 1974). Frequent interactions
among community members are thought to increase negative externalities such as
envy toward richer members.
1
Accordingly, a hypothesis is proposed that
community
participation increases the positive effect of an individual’s inequality aversion on the
private voluntary contributions of high-income earners
. Using individual-level data
from Japan, this paper aims to test the above hypothesis.
1
Yamamura (2012) suggested that richer people prefer income redistribution in areas
with higher rates of community participation, which implies that negative
externalities lead richer people to be more altruistic.
3
2. Data and Model
This study uses data sourced from a Japanese General Social Survey (JGSS),
which was designed as a Japanese counterpart to the GSS in the United States.
2
Therefore, this paper presents findings that can be compared with those of Derin-Güre
and Uler (2010), who provided evidence based on GSS data. However, the JGSS only
provides data regarding private charitable contributions from 2005, although the
JGSS was conducted from 2000 to 2010. Hence, this paper uses a 2005 data set based
on JGSS data. Following Derin-Güre and Uler (2010), the study sample here was
divided into three groups (high-income, middle-income, low-income). Respondents who
considered themselves above or high above average income earners are classified as
high-income, average earners are classified as middle-income, and below average and
far below average are classified as low-income.
Following Derin-Güre and Uler (2010), the estimated function takes the
following form:
CONTRIBUT
i
=
0
+
1
INEQL
i
+
2
Social capital
m
+
3
Age
i
+
4
Marry
i
+
5
Number
of Child
i
+
6
Male
i
+
7
Schooling
i
+
8
Income
i
+
m
u
i
,
where
CONTRIBUT
i
is the private charitable contributions of individual
i
, which is
estimated by yen value. Regression parameters are denoted by
and
u
i
is the error
term. Individuals appear to be influenced by the various experiences and
circumstances that surround the area they grew up in. To capture this effect, this
study incorporates
m
, which are dummies for prefecture
m
where respondents
2
Data for this secondary analysis, “Japanese General Social Surveys (JGSS), Ichiro
Tanioka,” was provided by the Social Science Japan Data Archive, Information Center
for Social Science Research on Japan, Institute of Social Science, The University of
Tokyo.
4
resided at 15 years of age. A Japanese prefecture is the equivalent to a state in the
United States or a province in Canada. Dependent variable,
CONTRIBUT
m
is
interval-coded data and this study uses an interval regression model for the
estimations (Wooldridge 2002, 508–509).
3
The proxy for inequality aversion (
INEQL
) was generated as follows. Respondents
were questioned about income redistribution: “What is your opinion of the following
statement? It is the responsibility of the government to reduce the differences in
income between families with high incomes and those with low incomes.” There
were five response options, ranging from “1 (strongly disagree)” to “5 (strongly
agree)”. The proxy for inequality aversion,
INEQL,
are the values chosen by the
respondents. It is expected that that people who support income redistribution are
more inequality averse.
The degree of community participation within a respondent’s residential area is
considered as a type of social capital (Putnam 2000). The influence of neighbors
appears to be greater when people are more inclined to participate in community
activities. That is, people are thought to be influenced by neighbors to a greater
extent when they live in areas with higher levels of community participation. In
this paper, the proxy for social capital is generated using a survey conducted by the
Japan Broadcasting Corporation (Japan Broadcasting Corporation 1997). One of
the survey questions asked, “Do you actively participate in community activities?”
Respondents could choose one of three responses: “yes”, “unsure”, or “no”. Thus,
Social capital
m
is the number of respondents who answered “yes” within a
3
Detailed information about JGSS data such as interval-codes for private charitable
contributions is available from the author upon request. Basic statistics for the
variables used in this paper are also available upon request.
5
prefecture.
Derin-Güre and Uler (2010) showed that the coefficient of
INEQL
takes the
significant positive sign for high-income groups, whereas
INEQL
takes the significant
negative sign for low-income groups. In this paper, in addition to the dependent
variables used in Derin-Güre and Uler (2010), a
cross term between an individual’s
inequality aversion (
INEQL
i
) and degree of community participation in their
residential area (
Social capital
m
) was also incorporated. If the hypothesis presented in
the introduction is supported, the sign of
INEQL
i
*
Social capital
m
becomes positive for
high-income groups.
Various individual characteristics were used as control variables, including age,
marital status, number of children, gender, schooling years, and household income.
3. Results
Table 1 shows the re-examination of the results of the United States study by
Derin-Güre and Uler (2010). The signs for
INEQL
are negative and positive for
low-income and high-income groups, respectively. Furthermore,
INEQL
is not
statistically significant for low-income groups, whereas it is statistically significant for
high-income groups. The significant positive effect of
INEQL
is equivalent to that of
the United States study. However, the significant negative effect of
INEQL
was
observed in the earlier United States study but not for Japan.
It is interesting to note
that differences in socio-economic backgrounds between Japan and the United States
influence the charitable behavior of low-income individuals in the two countries, while
the difference does not change the behavior of high-income individuals.
Table 2 shows the cross term of
INEQL
i
*
Social capital
m
, and it produces the
6
positive sign in all columns. However,
INEQL
i
*
Social capital
m
becomes statistically
significant at the 5% level only for high-income groups. This implies that the level of
community participation within a residential area only increases the positive effect of
an individual’s inequality aversion on private charitable contributions for high-income
groups. The marginal effects of
INEQL
and
Social capital
both produce the positive
sign for high-income groups, whereas they yield the negative sign for low-income
groups. Furthermore, regarding high-income groups,
INEQL
is statistically significant
at the 1% level whereas
Social capital
is not statistically significant. Neither
INEQL
nor
Social capital
become statistically significant for low-income groups. Hence, the
results for the marginal effects of
INEQL
and
Social capital
are in line with results
shown in Table 1.
Thus, it can be argued that the results in Table 2 show that community
participation does not directly influence the private charitable contributions of
high-income groups, but indirectly increases the level of charitable contributions, and
therefore increases the positive effect of an individual’s inequality aversion.
4. Conclusions
The estimation results based on the JGSS show that voluntary giving increases
with inequality aversion for high-income groups, but not for low-income group. This is
consistent with evidence from the United States based on the GSS (Derin-Güre and
Uler 2010). Furthermore, the degree of community participation in a respondent’s
residential area only increases the positive influence of inequality aversion on
voluntary giving for high-income groups. Thus, social capital increases the
psychological cost for high-income earners not to give private voluntary contributions.
7
References
Becker, G. 1974. Theory of social interaction. Journal of Political Economics 82,
1063–1093.
Becker, G. 1996. Accounting for Tastes. London, Harvard University Press.
Derin-Güre, P., Uler, N. 2010.Charitable giving under inequality aversion. Economics
Letters 107, 208-210.
Fehr, E., Schmidt, K.M., 1999. A theory of fairness, competition, and cooperation.
Quarterly Journal of Economics 114(3), 817-868.
Hayami, Y. 2001. Development economics: From poverty to the wealth of nations.
(second edition). New York: Oxford University Press.
Japan Broadcasting Corporation. 1997. Data Book: Survey on Consciousness of
Prefecture Residents (Zenkoku Kenmin Ishiki Chosa 1996). Tokyo: Japan
Broadcasting Corporation Press.
Putnam, R. 2000. Bowling Alone: The Collapse and Revival of American Community, A
Touchstone Book, New York.
Wooldridge, J.M. 2002. Econometric Analysis of Cross Section and Panel Data. London:
MIT Press.
Yamamura, E., 2012. Effect of social capital on income redistribution preferences:
comparison of neighborhood externality between high- and low-income households,
mimeo.
Zullo, R., 2011.Labor unions and charity. Industrial and Labor Relation Review 64(4),
699-711.
8
Table 1. Interval regression estimates: dependent variable is the amount of the private
voluntary contribution
Note: Numbers in parentheses are
z
-statistics. *, **, and *** indicate significance at
the 10%, 5% and 1% levels, respectively.
(1)
Total
(2)
Low income
(3)
High income
INEQL
–1.12
(–0.20)
-
–5.76
(–0.70)
-
30.1*
(1.95)
-
Age
2.78***
(5.92)
2.14***
(3.32)
3.88***
(2.20)
Marry
–3.94
(–0.22)
22.6
(1.08)
–175.0***
(–2.90)
Number of children
1.54
(0.26)
1.78
(0.23)
–63.3***
(–3.34)
Male
2.29
(0.20)
–14.5
(–0.91)
68.3**
(2.20)
Schooling
5.95**
(2.03)
3.30
(0.84)
8.37
(0.77)
Income
0.04***
(3.25)
–0.03
(–1.12)
0.11***
(3.61)
Constant
–247.5***
(–3.84)
–103.8
(–1.22)
–295.0
(–1.24)
Region dummies (region
respondents lived in at 15 years of
age)
Yes
Yes
Yes
Size of local government dummies
Yes
Yes
Yes
Log-likelihood
–2119
–901
–303
Left-censored observations
183
89
17
Right-censored observations
8
3
4
Observations
764
336
116
9
Table 2. Interval regression estimates: dependent variable is the amount of the private
voluntary contribution
Note: Numbers in parentheses are
z
-statistics. *, **, and *** indicate significance at
the 10%, 5% and 1% levels, respectively.
(1)
Total
(2)
Low income
(3)
High income
INEQL*Social capital
0.31
(0.35)
1.63
(1.30)
7.11**
(2.40)
INEQL
–16.1
(–0.37)
-
–83.6
(–1.38)
-
–0.28**
(–2.16)
-
Social capital
–1.07
(–0.30)
–6.73
(–1.30)
–19.8*
(–1.84)
Age
2.78***
(5.92)
2.12***
(3.28)
4.36**
(2.42)
Marry
–4.11
(–0.23)
24.6
(1.18)
–163.3***
(–2.67)
Number of children
1.47
(0.25)
2.36
(0.31)
–63.0***
(–3.36)
Male
2.02
(0.18)
–16.0
(–0.99)
48.6
(1.41)
Schooling
5.99**
(2.04)
3.36
(0.85)
7.79
(0.70)
Income
0.04***
(3.23)
–0.03
(–1.26)
0.10***
(3.50)
Constant
–196.1
(–1.08)
215.9
(0.84)
589.6
(1.04)
Marginal effect of
INEQL
–0.93
(–0.16)
–3.63
(–0.33)
41.6***
(2.60)
Marginal effect of
Social capital
0.12
(0.10)
–0.13
(–0.08)
4.43
(1.01)
Region dummies (region respondents
lived in at 15 years of age)
Yes
Yes
Yes
Size of local government dummies
Yes
Yes
Yes
Log-likelihood
–2119
–901
–300
Left-censored observations
183
89
17
Right-censored observations
8
3
4
Observations
764
336
116