Summary
shahdoh
1.pdf
R E S E A R C H P A P E R
Childhood Obesity and Unhappiness: The Influence of Soft Drinks and Fast Food Consumption
Hung-Hao Chang Æ Rodolfo M. Nayga Jr.
Published online: 21 March 2009 � Springer Science+Business Media B.V. 2009
Abstract A growing body of literature has examined the determinants of childhood obesity, but little is known about children’s subjective wellbeing. To fulfill this gap, this
paper examines the effects of fast food and soft drink consumption on children’s over-
weight and unhappiness. Using a nationwide survey data in Taiwan and estimating a
simultaneous mixed equation system, our results generally suggest a tradeoff in policy
implication. Fast food and soft drink consumption tend to be positively associated with
children’s increased risk of being overweight but they are also negatively associated with
their degree of unhappiness. Current and future policy/program interventions that aim to
decrease fast food and soft drinks consumption of children to reduce childhood obesity
may be more effective if these interventions also focus on ways that could compensate the
increase in degree of unhappiness among children.
Keywords Unhappiness � Childhood obesity � Fast food � Soft drink � Taiwan
1 Introduction
The World Health Organization (WHO) has reported that obesity has become a growing
threat to human health both in developing and developed countries (World Health Orga-
nization 2000). The prevalence of childhood obesity is also of increasing public concern
around the world. For example, in Taiwan, childhood obesity has increased dramatically
over the past decades. In 1970, only about 2% of Taiwanese school-children were con-
sidered obese. By 1988, this figure has risen to 17% (Wu 2001). To date, one in every four
H.-H. Chang (&) Department of Agricultural Economics, National Taiwan University, No 1, Roosevelt Rd Sec 4, Taipei 10617, Taiwan e-mail: [email protected]
R. M. Nayga Jr. Department of Agricultural Economics and Agribusiness, University of Arkansas, Fayetteville, AR 72701, USA e-mail: [email protected]
123
J Happiness Stud (2010) 11:261–275 DOI 10.1007/s10902-009-9139-4
children is now considered overweight in Taiwan (Taiwan Medical Association for the
study of obesity (TMASO) 2007).
Childhood obesity is a major public health problem that has both individual and
environmental causes. Among all of the factors that may be related to children’s body
weight, the promotion of healthy eating has become a target of health promotion and
research programs (Ludwig et al. 2001). Consequently, a number of nutrition and public
health studies suggest the importance of examining the influence of fast food consumption
on children’s weight (e.g., Bowman et al. 2004; Hsieh and FitzGerald 2005; Hui et al.
2003). In addition to fast food consumption, the association between children’s soft drinks
(i.e., sugar-sweetened beverages) consumption and obesity has also been discussed in the
literature (e.g., Andersen et al. 2005; Ariza et al. 2004; Berkey et al. 2004; Forshee et al.
2004; Troiano et al. 2000). Some studies suggest that consumption of sugar-sweetened
beverages leads to higher energy intakes, which may place children at risk for excess
weight gain and obesity (Andersen et al. 2005; Anderson et al. 2003; Troiano et al. 2000).
However, empirical evidence found in the previous studies is still inconclusive. For
instance, based on data consisting of 10,371 children drawn from the U.S. National Health
and Nutrition Survey, Troiano et al. (2000) found that soft drinks contributed to high
proportions of energy intakes among overweight children. In contrast, the results presented
by Andersen et al. (2005) using data from 3,139 children in Norway showed an insig-
nificant association between consumption of soft drinks and overweight.
Little is known as well about the relationship between fast food and soft drinks con-
sumption as well as the extent to which these consumption behaviors may affect children’s
subjective wellbeing or happiness. Addressing the association between children’s sub-
jective wellbeing and food consumption is of particular policy interest because the
prevalence of children’s mental health problems has been increasing dramatically in recent
years (e.g., Currie and Stabile 2006; Lindberg and Swanberg 2006). In Taiwan, approxi-
mately 38% of school children have psychological or behavioral problems (Lin 2002).
Understanding the effects of factors that are correlated with children’s body weight and
their psychological/mental health is crucial in that if the reduction in children’s body
weight improves their physical health, the decrease in the likelihood of subjective well-
being may result in an adverse effect on their psychological or mental health. Hence,
policies and programs that aim to improve children’s overall health should take these
effects on children objective (i.e., obesity) and subjective (i.e., unhappiness) wellbeing into
account.
Research focusing on individual’s subjective wellbeing has received great attention
recently due to the fact that the rational vision of economic decision-making has come
under increasing scrutiny (Graham 2005). Since the maximum utility framework cannot
explain many circumstances of individual’s decisions, an alternative indicator of human
wellbeing from the field of psychological science has been proposed to explain human
behavior (e.g., Anand and Clark 2006; Cummins 2000). 1
One of the common indicators is
‘‘happiness’’ or ‘‘unhappiness’’, which measures individual’s subjective wellbeing.
Recently, the use of happiness indicator has received increasing attention from economists
(e.g., Frey and Stutzer 2002; Graham 2005; Easterlin 2001; Kahneman and Krueger 2006).
Most of the studies related to subjective wellbeing are focused on the examination of the
factors associated with adult’s subjective wellbeing. However, not much attention has been
paid on children.
1 Several indicators from the psychological science can also be found in Frey and Stutzer (2002).
262 H.-H. Chang, R. M. Nayga Jr.
123
The focus of this paper is based on two important questions. First, what factors are
associated with children’s fast food and soft drink consumption? Second, how do these two
types of consumption influence children obesity and unhappiness? Using a nationwide
survey data in Taiwan, our empirical results suggest that children’s fast food and soft drink
consumption are influenced by children’s characteristics and household features. Addi-
tionally, both soft drinks consumption and fast food consumption are positively associated
with children’s overweight and negatively associated with degree of unhappiness. If the
likelihood of children’s overweight and happiness represent their objective and subjective
wellbeing, our finding suggests that consumption of fast food and soft drinks can result in a
tradeoff between children’s objective and subjective wellbeing.
The remainder of this paper is constructed as follows. The data used in this study is
introduced in the following section. We then discuss the econometric strategy. After then
presenting and discussing the empirical results, we conclude with a brief summary and a
discussion of policy implications.
2 Data
We conduct the empirical analysis using data drawn from the National Health Interview
Survey at Taiwan (NHIS) in 2001. NHIS is an enumerative national survey conducted by
the National Health Research Institute of Taiwan (NHRI) to gather information on health
status and health behavior of the citizens. NHIS data were collected using a standardized
face-to-face interview between August 2001 and January 2002, and a multistage stratified
sampling scheme was used to select a probability sample. In the first stage, 359 townships
of Taiwan were divided into seven regions according to their geographic location. The
townships in each region were chosen based on their population size. Within each town-
ship, a number of lins (i.e., a smaller unit than township) were sampled with their
population size. 2
Therefore, NHIS is representative of the non-institutionalized population
in Taiwan (Lo et al. 2003). In all, 6,592 households and 26,658 adults were interviewed. In
addition, it captured survey data on a total of 3,977 children under 12 years old. These
adults and children data sets can be linked to each other based on the identification number
of each household. Since the cutoff points for body weight are only applicable for children
above 2-years old, we first limit our analysis to a sub-sample of children between 2 and 12-
years old. To capture the characteristics of the parents, we linked the adult data to the child
data set. 3
We then deleted the missing values of some key variables (such as body weight)
in the data, resulting in 2,366 children. 4
With respect to fast food and soft drinks consumption, respondents of NHIS are asked
the following questions: ‘‘How often does your child consume the following food item?’’
There are five categories that each respondent’s answer may fall into: never, seldom,
sometimes, often, and always. We assign a value between 0–4 (i.e., 0 for never, 1 for
seldom, 2 for sometimes, 3 for often, and 4 for always) to the answers for each food item
and sum up the scores for food items: French fries, pizza, and hamburger. This sum
represents our measurement for fast food consumption (the variable FASTFOOD). Also,
2 Detailed descriptions of the survey designs can be found in Lo et al. (2003) and Pan et al. (2003).
3 We exclude the children sample if their parents didn’t live with them.
4 To avoid the clustering effects as a result of multiple children from the same family, we follow the method
used in McIntosh et al. (2006) to randomly select only one child from each household.
Childhood Obesity and Unhappiness 263
123
the corresponding scores for soda and other sugar-sweetened beverages are our mea-
surement for soft drinks (the variable DRINK). In our sample, the mean values are 2.86 and
3.55 for children’s fast food and soft drink consumption, respectively.
The other two variables of particular interests are children’s weight status and the
degree of happiness. The weight status is defined by their body mass index (BMI). 5
Unlike
the case of adults, children’s BMI cannot be directly used to define their weight status, and
it has to be defined by age and gender based on the distribution of the population in the
same age. For instance, a child above 2-years old with a BMI above the 85th percentile and
less than the 95th percentile of the population is classified as at risk of being overweight.
The same child with a BMI above 95th percentile is considered at risk of being obese. In
this study, the cutoff points are made by age and gender and are determined by the Taiwan
Health Department. 6
In our sample, approximately 25% of children are considered over-
weight or obese. With respect to children’s degree of unhappiness, respondents of NHIST
are asked the following questions: ‘‘How often does your child feel unhappy, sad or
depressed?’’ Each respondent’s answer may fall into one of three categories: never,
sometimes, and often. Due to the low percentage of responses in the third category, we
combined the last two categories into a single category to represent if the child ever feels
unhappy, sad, or depressed. 7
In so doing, a binary indicator UNHAPPY is used to indicate
if the child feels unhappy. The sample statistics show that about 19% of children some-
times or often feel unhappy, sad or depressed. With respect to the influence of the fast food
and soft drinks consumption on children’s body weight and unhappiness, it appears that
children with fast food and soft drink consumption are likely to be overweight and less
unhappy (Table 1).
The other variables included in our analysis are built on the empirical specifications
from some of the previous studies (e.g., Anderson et al. 2003; Berkey et al. 2004; Bowman
et al. 2004; Crespo et al. 2001; Lin et al. 2004). Several variables of children’s charac-
teristics, household features, and geographical conditions are hypothesized to be associated
with children’s consumption, body weight, and happiness. In addition to these factors,
mother’s consumption of fast food and soft drink are included (FASTFOOD_M and
DRINK_M) to capture the importance of mothers’ consumption behavior on children.
Children’s characteristics are represented by age and gender (AGE and MALE). Because
soft drinks consumption has been shown to be associated with dental care (e.g., Marshall
et al. 2003), a dummy variable that specifies if the child has a routine visit with the dentist
(DENTIST) is included. Also, a dummy variable that specifies if the child has ever stayed
Table 1 Sample means of overweight and unhappiness
Children with fast food or soft drink consumption
If child is overweight 0.26
If child is unhappy 0.17
Children without fast food or soft drink consumption
If child is overweight 0.22
If child is unhappy 0.22
5 It is measured as the ratio of weight (in kilogram) to height squared (in meters).
6 The cutoff points for determining overweight or obese weight status in Taiwan can be found on the
website http://www.vghtpe.gov.tw/*nutr/forum/forum02/bmi2.htm. 7
The percentages of these categories in our sample are 80, 18, and 2, respectively.
264 H.-H. Chang, R. M. Nayga Jr.
123
at the hospital in the past 6 months (HOSPITAL) is included to reflect his/her health
condition. Several household characteristics are included as well. Family incomes are
defined in several categories (HHINC2, HHINC3, HHINC4, and HHINC5). To reflect the
effects of other household members on children’s behavior, the variable LONGCARE
indicates if any of the household members needs long term medical care. Finally, com-
pared to the northern area (the reference group), three regional dummies (SOUTH, EAST,
CENTER) are specified to indicate if the household is located in the south, east, and center
of Taiwan, respectively. Urbanization variables are also specified to indicate if the
household is located in the central city (CITY) or county (COUNTY). The sample statistics
of variables used in this study is listed in Table 2.
3 Econometrics Strategy
The econometric framework consists of two parts. First, we estimate a mixed structure
simultaneous-equation system for the effects of factors that are associated with children’s
fast food and soft drink consumption, and the likelihood of being overweight, and
Table 2 Sample statistics
Variable definition Mean SD
Dependent variables
FASTFOOD Child’s fast food consumption (0–12) 2.86 1.88
DRINK Child’s sugar drink (0–8) 3.55 1.95
OVERWEIGHT If the child is overweight or obese (=1) 0.25 0.43
UNHAPPY If the child is unhappy (=1) 0.19 0.40
Mother’s food consumption
FASTFOOD_M Mother’s fast food consumption (0–12) 1.52 1.73
DRINK_M Mother’s sugar drink consumption (0–8) 3.14 1.81
Children characteristics
AGE Age of the child in years 7.14 2.87
MALE If male (=1) 0.51 0.50
HOSPITAL If ever stay in hospital in the previous year (=1) 0.06 0.23
DENTIST If dentist visit in the past 6 months (=1) 0.56 0.50
Household factors
HHINC2 If monthly income NT$50,000 * 70,000 (=1) 0.24 0.43
HHINC3 If monthly income is NT$70,000 * 100,000 (=1) 0.17 0.38
HHINC4 If monthly income is NT$100,000 * 150,000 (=1) 0.11 0.32
HHINC5 If monthly income is more than NT$150,000 (=1) 0.07 0.25
LONGCARE If any family member needs long care (=1) 0.12 0.33
Region and local economy conditions
CITY If located in central city (=1) 0.56 0.50
COUNTY If located in county area (=1) 0.23 0.42
CENTER If located in central Taiwan (=1) 0.39 0.49
SOUTH If located in southern Taiwan (=1) 0.25 0.43
EAST If located in eastern Taiwan (=1) 0.04 0.20
2,366 children are included
Childhood Obesity and Unhappiness 265
123
unhappiness. To capture the potential correlations between these behaviors, this model is
estimated with a simulated maximum likelihood estimation method. To validate our
empirical specification, the second part of this section introduces the statistical tests for
testing the endogeneity assumption and the restricted exclusions.
3.1 Estimating the Mixed Structure Model
We estimate a four-equation simultaneous-equation system. The first two equations rep-
resent the fast food and soft drink consumption of children, and the other two equations
represent children’s likelihood of being overweight and unhappiness. This model is mixed
in that children’s consumption of fast food and soft drink are censored and the variables
representing children’s risk of being overweight and unhappiness are binary indicators.
Suppose i represents each child, the simultaneous equation system is specified as:
y�1i ¼ xib1i þ zik1 þ e1i y�2i ¼ xib2i þ zik2 þ e2i
y�3i ¼ xib3i þ y1ia1 þ y2ia2 þ e3i y�4i ¼ xib4i þ y1ic1 þ y2ic2 þ e4i
ð1Þ
yj ¼ y�j iff y � j [ 0 and yj ¼ 0 iff y
� j 50 ðj ¼ 1 and 2Þ
yk ¼ 1 iff y�k [ 0 and yk ¼ 0 iff y � k 50 ðk ¼ 3 and 4Þ
where y1 *
and y2 *
are the unobserved latent variables that represent children’s fast food
and soft drink consumption, respectively. y3 *
and y4 *
are the unobserved latent variables
that represent the likelihoods of being overweight and unhappy. Since not every child
consumes fast food or soft drink, these two variables may be censored at zero value. y3 and y4 are the binary indicators counterparts for the latent variables y3
* and y4
* . Xi are
vectors of common exogenous factors that are associated with child’s food consumption
and the risk of being overweight and unhappy. The vector Zi contains the excluded
variables that are associated with child’s food consumption but not directly affecting
his/her likelihood of being overweight and unhappy. The vectors (a; b; c; k) are the parameter of interests.
The vectors ej are random errors, which assume to follow a multivariate normal dis-
tribution u ð0; RÞ, with zero mean and the covariance R r21 ::: ::: ::: q21 r
2 2 ::: :::
q31 q32 1 ::: q41 q42 q43 1
2 664
3 775.
For identification purpose, the variances of e3; e4 are normalized to be 1 as the con- ventional strategy used in the binary probit model (Greene 2003). Overall, 16 regimes can
be realized by the data, and the probability of each regime can be derived. For instance, the
probabilities that (y1 [ 0, y2 [ 0, y3 = 1, y4 = 1) and (y1 = 0, y2 = 0, y3 = 0, y4 = 0) are
8 :
8 To save space, we didn’t present the derivatives of the probability of other regimes. However, they can be
derived in a similar way.
266 H.-H. Chang, R. M. Nayga Jr.
123
Pr ðy1 [ 0; y2 [ 0; y3 ¼ 1; y4 ¼ 1Þ
¼ fðe1; e2Þ Z1
�Xb3�y1 a1�y2 a2
Z1
�Xb4�y1c1�y2 c2
gðe1; e2je3; e4Þ de1de2 ð2Þ
Pr ðy1 ¼ 0; y2 ¼ 0; y3 ¼ 0; y4 ¼ 0Þ¼ Z�Xb1
�1
Z�Xb2
�1
Z�Xb3
�1
Z�Xb4
�1
gðe1; e2e3; e4Þ de1de2de3de4:
ð3Þ By combining the probability of the 16 possible regimes, the consistent estimates can be
obtained by implementing the maximum likelihood estimation method on the following
log-likelihood function.
log L ¼ XN i¼1
log Prðy1; y2; y3; y4Þ � di ð4Þ
where the dummy indicator di represents the specific regime that each individual may fall into.
To estimate Eq. 4, it is necessary to evaluate the multivariate normal density function.
The simulated maximum likelihood (SML) method based on the Geweke-Hajivassiliour-
Keane (GHK) smooth recursive conditioning simulator is used. It has been shown that the
GHK simulator is unbiased for any number of replications and generates smaller variances
than other simulators (Train 2003). Additionally, it has been shown to be the most
unambiguously reliable method for simulating normal probabilities (Hajivassiliou et al.
1996). Simulating the multivariate normal probability in the likelihood function is a
practical alternative to numerical evaluation of the probability integrals.
Given the consistent estimates of the Eq. 4, the effects of the fast food and the soft drink
consumption on children’s overweight and unhappiness can be evaluated according to the
average treatment effects on the sample means (Greene 2003). For instance, the effects of
fast food consumption on children’s overweight and unhappiness can be shown as:
ATEðy3Þ¼ Uðxb̂3 þ â1 �y1 þ â2 �y2Þ� Uðxb̂3 þ â2 �y2Þ ð5Þ
ATEðy4Þ¼ Uðxb̂4 þ ĉ1 �y1 þ ĉ2 �y2Þ� Uðxb̂4 þ ĉ2 �y2Þ ð6Þ
where x represents the vectors of the means values of the explanatory variables. 9
3.2 Statistical Tests
To validate the empirical specification, two statistical tests are conducted: weak instru-
ments test and test for endogeneity. The endogeneity issue arises if the error terms of the
simultaneous-equation mixed model are correlated. The endogeneity test also justifies the
need to estimate the four equation system simultaneously. Since this equation system is
estimated by the maximum likelihood estimation method, a likelihood ratio test can be
9 The interpretation of the treatment effect is similar to the continuous outcome variable case discussed on
page 787–788 in Greene (2003). However, it is of note that the average treatment effect in our case is the differences in the predicted probability, instead of the expected values as the continuous variable case.
Childhood Obesity and Unhappiness 267
123
used to test if all of the error correlations are exogenous. Specifically, the null hypothesis is
that all of the estimated correlation coefficients are equal to zero. The test statistic follows
the chi-square distribution with degree of freedom 6 in our case.
Unlike the two-stage estimation procedure of the instrumental variable method for
which exclusion conditions are necessary for model identification, the identification cri-
teria are met due to the nonlinear functional form and distributional assumption for the
current mixed system (i.e., Eq. 4). However, some exclusion restrictions are useful to
avoid full reliance on the functional form nonlinearity for identification. 10
In this study,
mother’s fast food and soft drink consumption are selected as the exclusion variables in
children’s body weight and unhappiness equations. To validate our exclusion variables,
the weak instruments test is conducted to investigate if the excluded variables have
enough explanatory power for model identification. Although most of the applications of
the weak instrument tests proposed by Staiger and Stock (1997) are for the linear model,
Kan (2007) provided a modification to a non-linear model. Using a similar strategy in Kan
(2007), the weak instruments test involves estimating the binary tobit model for children’s
fast food or soft drink consumption and testing if the excluded instruments (i.e., mother’s
consumption of fast food and soft drink beverage) have enough explanatory power. A
likelihood ratio test is implemented and the test statistic follows a chi-square distribution
(say x2) with the degree of freedom M (2 in our case). To be consistent with Staiger and Stock (1997), the likelihood ratio test statistic can be covered with a F statistic as: x2=M �FðM;1Þ.11
4 Empirical Results
The estimations of the mixed simultaneous equation system and the results of the statistical
tests are exhibited in Table 3. We begin our discussion of the results by looking at the
findings of the statistical tests (the bottom in Table 3). Since mother’s fast food and soft
drink consumption are used as exclusion variables that directly affect children’s con-
sumption of these two food items, but not directly affect the risk of being overweight and
unhappiness of children, a statistical test is conducted to see if these exclusions are sta-
tistically weak. The weak instrument test is conducted separately for children’s fast food
consumption and soft drink equation. The results of the tests are encouraging. The values
of the likelihood tests are 42.17 and 33.22 for children’s fast food and soft drink equations,
respectively. These are asymptotically equal to 21 and 17 of the F statistics,12 and both of which are higher than 10 (the critical value) proposed by Staiger and Stock (1997).
Therefore, we may conclude that mother’s fast food and soft drink consumption are sta-
tistically strong to identify her children’s fast food and soft drink consumption.
As in the conventional simultaneous equation system, error correlations accommodate
endogeneity between equations (Amemiya 1973). The investigation of the correlations of
children’s fast food, soft drink consumption, overweight, and unhappiness are also of
interest since it could justify the need to estimate a mixed system. A likelihood ratio
test is conducted to investigate if these correlations are statistically equal to zero
10 Discussions of the model identification through the nonlinear functional form are on page 121 in Maddala
(1983). 11
A detailed discussion of the proposed test procedure can be found on page 71–72 in Kan (2007). 12
The asymptotic statistics of the likelihood ratio test to the F test is discussed in Kan (2007).
268 H.-H. Chang, R. M. Nayga Jr.
123
T a b
le 3
E st
im a ti
o n s
o f
th e
e c o n o m
e tr
ic m
o d e l
F a st
fo o d
S o ft
d ri
n k
O v e rw
e ig
h t
U n h a p p in
e ss
C o e ffi
c ie
n t
S E
C o e ffi
c ie
n t
S E
C o e ffi
c ie
n t
S E
C o e ffi
c ie
n t
S E
F A
S T
F O
O D
– –
– –
0 .0
1 7
0 .0
0 5
2 0
.0 7 5
0 .0
3 3
D R
IN K
– –
– –
0 .1
0 8
0 .0
4 1
- 0
.0 4 1
0 .1
1 7
F A
S T
F O
O D
_ M
0 .4
9 3
0 .0
2 4
0 .1
3 3
0 .0
2 5
– –
– –
D R
IN K
_ M
0 .0
1 1
0 .0
2 3
0 .1
1 9
0 .0
2 3
– –
– –
A G
E 0
.1 0 9
0 .0
1 5
0 .0
6 6
0 .0
1 5
0 .0
0 7
0 .0
1 1
0 .0
9 3
0 .0
1 2
M A
L E
0 .0
5 4
0 .0
8 0
0 .0
9 9
0 .0
8 3
- 0
.0 2
7 0
.0 5
7 0
.0 4 0
0 .0
6 1
H O
S P
IT A
L -
0 .1
4 8
0 .1
7 5
2 0
.4 6 9
0 .1
8 3
0 .1
8 1
0 .1
2 7
- 0
.2 2 2
0 .1
5 6
D E
N T
IS T
0 .1
6 4
0 .0
8 4
0 .1
5 7
0 .0
8 8
- 0
.0 6
9 0
.0 6
1 0
.1 5 2
0 .0
6 5
H H
IN C
2 -
0 .1
3 4
0 .1
0 3
0 .2
6 0
0 .1
0 7
0 .0
5 5
0 .0
7 7
0 .0
2 1
0 .0
8 5
H H
IN C
3 -
0 .0
5 8
0 .1
1 6
0 .0
8 7
0 .1
2 2
2 0
.1 8
0 0
.0 8
7 0
.1 1 3
0 .0
8 9
H H
IN C
4 -
0 .0
8 4
0 .1
3 6
0 .1
4 7
0 .1
4 2
- 0
.1 0
9 0
.1 0
0 0
.0 3 9
0 .1
0 5
H H
IN C
5 -
0 .2
6 1
0 .1
7 0
- 0
.2 9 1
0 .1
7 8
2 0
.2 8
3 0
.1 2
8 -
0 .0
8 6
0 .1
3 5
L O
N G
C A
R E
- 0
.1 4 5
0 .1
2 3
0 .0
9 3
0 .1
2 8
2 0
.2 0
2 0
.0 9
3 0
.2 7 0
0 .0
9 1
C IT
Y 0
.2 4 3
0 .1
0 9
2 0
.2 2 4
0 .1
1 3
- 0
.1 2
4 0
.0 8
1 0
.2 5 1
0 .0
9 3
C O
U N
T Y
0 .0
4 9
0 .1
2 9
0 .2
3 6
0 .1
3 4
- 0
.0 6
1 0
.0 9
5 0
.1 9 7
0 .1
0 4
C E
N T
E R
- 0
.0 8 3
0 .1
0 0
0 .0
5 2
0 .1
0 5
- 0
.0 4
7 0
.0 7
3 -
0 .0
5 3
0 .0
7 6
S O
U T
H -
0 .1
5 1
0 .1
1 1
0 .1
9 3
0 .1
1 6
0 .1
8 0
0 .0
8 0
- 0
.0 0 3
0 .0
8 8
E A
S T
- 0
.2 5 2
0 .2
1 7
0 .6
2 4
0 .2
2 3
- 0
.1 8
8 0
.1 7
7 -
0 .2
2 7
0 .1
9 1
C o n
st a n
t 1
.0 7 7
0 .1
9 4
2 .2
2 6
0 .2
0 2
- 0
.1 2
2 0
.2 8
6 2
1 .7
5 2
0 .3
4 8
Childhood Obesity and Unhappiness 269
123
T a b
le 3
c o
n ti
n u e d
F a st
fo o
d S
o ft
d ri
n k
O v
e rw
e ig
h t
U n
h a p
p in
e ss
C o e ffi
c ie
n t
S D
C o e ffi
c ie
n t
S D
C o e ffi
c ie
n t
S D
C o e ffi
c ie
n t
S D
C o
rr e la
ti o
n S
D
(F A
S T
F O
O D
,D R
IN K
) 0
.2 7
7 0
.0 5 9
F A
S T
F O
O D
1 .9
0 4
0 .3
0 8
(F A
S T
F O
O D
,O V
E R
W E
IG H
T )
0 .0
8 5
0 .0
1 3
S U
G A
R D
R IN
K 2
.0 0 6
0 .3
0 7
(F A
S T
F O
O D
,U N
H A
P P
Y )
2 0
.0 6
8 0
.0 3 1
O V
E R
W E
IG H
T 1
.0 0 0
–
(D R
IN K
,O V
E R
W E
IG H
T )
0 .2
0 7
0 .0
3 9
U N
H A
P P
Y 1
.0 0 0
–
(D R
IN K
,U N
H A
P P
Y )
0 .0
9 5
0 .2
1 0
(O V
E R
W E
IG H
T ,U
N H
A P
P Y
) 2
0 .0
7 5
0 .0
4 3
W e a k
IV T
e st
a T
e st
v a lu
e C
ri ti
c a l
v a lu
e E
n d
o g
e n
e it
y T
e st
b T
e st
v a lu
e C
ri ti
c a l
v a lu
e
F a st
fo o
d E
q u
a ti
o n
2 1
1 0
2 8
3 1
3
D ri
n k
e q
u a ti
o n
1 7
1 0
A v e ra
g e
T re
a tm
e n t
e ff
e c ts
c C
o e ffi
c ie
n t
S E
C o e ffi
c ie
n t
S E
F a st
fo o
d 0
.0 2 0
0 .0
0 5
2 0
.0 5 0
0 .0
2 3
S o
ft d
ri n
k 0
.1 5 1
0 .0
8 0
2 0
.0 4 6
0 .0
2 7
a T
h e
n u
ll h
y p
o th
e si
s H
0 :
th e
IV s
a re
st a ti
st ic
a ll
y w
e a k
b T
h e
n u
ll h
y p
o th
e si
s H
0 :
a ll
th e
c o rr
e la
ti o n s
a re
z e ro
. L
R te
st a n d
th e
c ri
ti c a l
v a lu
e is
x 2 (0
.9 5
,6 )
= 1
2 .9
5 c
E v
a lu
a te
d a t
th e
sa m
p le
m e a n s
B o ld
a re
si g n
ifi c a n
t a t
1 0
% le
v e l
T h
e e st
im a te
d lo
g -l
ik e li
h o
o d
v a lu
e is
- 1
1 ,7
7 4
270 H.-H. Chang, R. M. Nayga Jr.
123
(i.e., uncorrelated). The result of the likelihood ratio test is 283, which is higher than the
conventional significance level of the chi-squared distribution (x2(0.95,6) = 12.95). Therefore, this result is supportive of the joint estimation of a mixed system to improve the
efficiency of the estimators. Additionally, the estimated correlation coefficient between
child’s fast food and soft drink consumption is 0.277, and it is statistically significant. This
shows that fast food and soft drink consumption are positively correlated due to common
unobserved characteristics. Additionally, 8.5 and -6.8% correlations are found between
child’s fast food consumption and the risk of being overweight and unhappiness, respec-
tively. These results indicate that fast food consumption of children is associated with risk
of being overweight and unhappiness due to unobserved factors. Similar evidence is
revealed between child’s soft drink consumption and being overweight or obese (20.7%).
4.1 Factors Associated with Children’s Fast Food and Soft Drink
The variables that significantly influence children’s fast food and soft drink consumption,
overweight, and unhappiness include mothers’ food consumption, children and family
characteristics, and geographical or regional factors. To begin, we first examine the factors
that are associated with child’s consumption of fast food and soft drinks. Results point out
that mother’s fast food and soft drink consumption positively contributes to children’s food
consumption. This result may not be inconsistent with the belief that mothers have some
control on children’s food choice, or may reflect the fact that mothers usually spend more
time than other family members on children’s food consumption (e.g., Lindsay et al. 2006).
Children’s characteristics also affect their food consumption. For instance, children’s
age is positively correlated with their consumption of fast food and soft drink. Also,
children who have regular dental examinations consume more fast food and soft drink than
those who do not have regular dental examinations. Household characteristics also play an
important role on children’s fast food consumption. Compared to those in lower income
households, children living in households with income between NT$50,000 and NT$
70,000 are more likely to consume soft drinks.
Finally, geographical and locational variables also significantly affect children’s food
consumption. Compared to children living in small towns (the reference group), children
who live in city and county areas consume more fast food and soft drinks. This may reflect
the fact that children living in city areas have easy access to convenience stores or res-
taurants that provide fast food and soft drink, which increases the frequency of
consumption. It is also no surprise that fast food consumption is lower on children living in
households located in eastern Taiwan. Compared to the north part of Taiwan (reference
group), eastern Taiwan has less business and commercial activities. Therefore, the fast
food consumption of children may be lower due to the limited availability of fast food
restaurants.
4.2 Factors Associated with Children’s Overweight and Unhappiness
Perhaps, the most interesting finding and contribution of this paper is the association
between children’s fast food and soft drink consumption, and their objective and subjective
wellbeing. Results show that children who consume more fast food and soft drink are more
likely to be overweight but are less likely to be unhappy. The estimated average treatment
effects of children’s fast food and soft drink on overweight are 0.02 and 0.151. That is,
after controlling for the socioeconomic factors and other exogenous determinants,
Childhood Obesity and Unhappiness 271
123
compared to children who don’t consume these two food products, those children have
more likely to be overweigh by 2 and 15%, respectively. However, those children have 5
and 4.6% lower probabilities of being unhappy. Several points are noticeable from this
finding. First, this result is continuing to confirm the findings of previous studies regarding
the positive association between children fast food consumption and body weight (e.g.,
Chen et al. 2006). Second, this finding contributes to the debate of the influences of
children’s soft drink consumption on body weight. Consistent with the findings in Troiano
et al. (2000), a positive association between soft drinks consumption and the risk of being
overweight or obese is evident. Third, our results demonstrate that children’s consumption
of fast food and soft drink are positively correlated with their happiness.
Other factors that are significantly associated with children’s weight status include the
characteristics of children, household, and geographical and regional conditions. For
example, compared to children living in lower income households, children who live in
households with family income between NT$70,000 and NT$100,000 are less likely to be
overweight or obese. This result is consistent with the finding in Anderson et al. (2003)
using the NHANES 1988–1994 data in the United States. Household characteristics also
influence children’s likelihood of being overweight and obese. Results indicate that chil-
dren living in households with at least one member needing long care medical service are
less likely to be overweight or obese. Regions and geographic variables also significantly
determine children’s body weight. Compared to children living in the north part of Taiwan,
children who live in households located in the southern part have higher risk of being
overweight or obese.
Factors that determine children’s unhappiness are similar to the findings of the weight
status. Consistent with the findings in previous studies indicating the importance of the
association between children’s age and the psychological/mental health (e.g., Frey and
Stutzer 2002), results show that child’s age is negatively associated with their happiness.
Consistent with the finding of factors determining children body weight, children living in
households with long care family members are more likely to feel unhappy, sad or
depressed. Urbanization and geographic locations are also significant determinants of
children’s happiness. Compared to children who live in the rural area, those who live in the
city or county areas have higher propensity to be unhappy. This may reflect the envi-
ronmental effects of living in a city than in a rural area (e.g., pollution, noise, security,
etc.). The effects of environmental factors on children’s subjective wellbeing can be further
confirmed by the finding that, compared to children in the northern area of Taiwan,
children living in the eastern part of Taiwan are found to be less unhappy.
5 Concluding Remarks
Extensive literature has examined the factors that are associated with children’s objective
wellbeing (i.e., body weight), but relatively little is known about children’s subjective
wellbeing (i.e., unhappiness). Focusing on children’s consumption of fast food and soft
drink, this paper investigates the association between children’s fast food and soft drink
consumption, and the risk of being overweight and unhappy. No other known study has
evaluated the effect of fast food and soft drinks consumption on children’s obesity and
unhappiness. Moreover, in contrast to previous studies, we estimate a mixed system
allowing for the endogeneity between children’s food consumption and wellbeing.
Using a nationwide survey data in Taiwan, our results show that children’s character-
istics and household factors, as well as the geographical conditions, are associated with
272 H.-H. Chang, R. M. Nayga Jr.
123
children’s consumption of fast food and soft drinks. Specifically, mother’s fast food and
soft drinks consumption behavior is found to be positively associated with their children’s
fast food and soft drinks consumption. In our analysis, the error correlations between fast
food and soft drink consumption, and children’s weight status and happiness are significant
and interrelated. Therefore, it is necessary to estimate these equations simultaneously. With
respect to the effects of children’s fast food and soft drink consumption on children’s
wellbeing, our results suggest a trade-off in the influence of food consumption on child-
hood obesity and unhappiness since while consumption of fast food and soft drinks is
positively associated with childhood obesity, they also tend to make them less unhappy.
This finding is of particular policy interest since current policy strategies for decreasing the
prevalence of childhood obesity have focused on child’s objective wellbeing (e.g., body
weight), and far less attention has been paid to the examination of the subjective wellbeing.
Interestingly, our results also imply that consumption of fast food and soft drinks is
negatively correlated with degree of unhappiness. While the definitive assessment of the
mechanisms underlying these associations is complicated, these findings may imply that
there could be a tradeoff between children’s obesity and happiness with reductions in fast
food and soft drink consumption. It is then possible that policy interventions would be
insufficiently effective in reducing childhood obesity if they do not take into account the
effect of policy prescriptions such as reducing fast food and soft drinks consumption on
children’s happiness. Policy or program prescriptions should take this tradeoff into account
to facilitate reduction in childhood obesity without sacrificing children’s degree of hap-
piness. Hence, current and future policy/program interventions that aim to decrease fast
food and soft drinks consumption of children to reduce childhood obesity may be more
effective if these interventions also focus on ways that could compensate the potential
reduction in degree of happiness of children.
Future research should accommodate the other types of food items (e.g., fruit and
vegetable etc.) that may increase children’s happiness but at the same time may not
increase body weight. However, including more items of food consumption is chal-
lenging in model estimation. One caveat from our analysis is that due to data limitations,
our measure for food consumption is based on frequency scores and not actual amount
consumed by children and our measure for degree of children’s unhappiness is based on
subjective observations from parents. In addition, we could not include an exercise
variable in our model due to data limitations. Admittedly, the addition of an exercise
variable may moderate the effects of fast food and soft drink consumption in the
analysis. Hence, future research should replicate our analysis with more measures related
to exercise and other lifestyle behaviors to test the robustness of our findings. With data
availability, the addition of psychological factors should also be included in the analysis
to further assess the robustness of our findings. For example, consumption of fast food
and soft drink may be addictive or may lead to depression and feeling of being
exhausted. If this is the case, then it is possible that consumption of fast food and soft
drink may just be a temporary relief from an addiction and may then not be associated
with a true happy feeling.
Acknowledgments Hung-Hao Chang acknowledges partial funding support from the National Science Counsel of Taiwan under Grant No: 95-2415-H-002-041. The data used in the analysis is provided by the Bureau of Health Promotion, Department of Health and National Health Research Institute in Taiwan. The interpretation and conclusions do not represent those of Department of Health and National Health Research Institute. The authors accept responsibility for any remaining errors or omissions.
Childhood Obesity and Unhappiness 273
123
References
Amemiya, T. (1973). Regression analysis when the dependent variable is truncated normal. Econometrica, 41, 997–1016. doi:10.2307/1914031.
Anand, P., & Clark, A. (2006). Symposium introduction: life satisfaction and welfare economics. Journal of Socio-Economics, 35, 177–179. doi:10.1016/j.socec.2005.11.013.
Andersen, L., Lillegaard, I., Overby, N., Lytle, L., Klepp, K., & Johansson, L. (2005). Overweight and obesity among Norwegian school children: changes from 1993 to 2000. Scandinavian Journal of Public Health, 33, 99–106. doi:10.1080/140349404100410019172.
Anderson, P., Butcher, K., & Levine, P. (2003). Economic perspectives on children obesity. Economic Perspectives, QIII, 30–48.
Ariza, A., Chen, E., Binns, H., & Christoffel, K. (2004). Risk factors for overweight in five- to six-year-old Hispanic-American children: a pilot study. Journal of Urban Health, 81, 150–161. doi:10.1093/jurban/ jth091.
Berkey, C., Rockett, H., Field, A., Gillman, M., & Colditz, G. (2004). Sugar-added beverages and adolescent weight changes. Obesity Research, 12, 778–788. doi:10.1038/oby.2004.94.
Bowman, S., Gortmaker, S., Ebbeling, E., Pereira, M., & Ludwig, D. (2004). Effects of fast-food con- sumption on energy intake and diet quality among children in a national household survey. Pediatrics, 113, 112–118. doi:10.1542/peds.113.1.112.
Chen, L., Fox, K., Haase, A., & Wang, J. (2006). Obesity, fitness, and health in Taiwanese children and adolescents. European Journal of Clinical Nutrition, 60, 1367–1375. doi:10.1038/sj.ejcn.1602466.
Crespo, C., Smit, E., Trojano, R., Bartlett, S., Macera, C., & Anderse, R. (2001). Television watching, energy intake, and obesity in U.S. children: results from the third National Health and Nutrition Examination Survey, 1988–1994. Archives of Pediatrics & Adolescent Medicine, 155, 360–365.
Cummins, R. (2000). Personal income and subjective-wellbeing: a review. Journal of Happiness Studies, 1, 133–158. doi:10.1023/A:1010079728426.
Currie, J., & Stabile, M. (2006). Child mental health and human capital accumulation: the case of ADHD. Journal of Health Economics, 25, 1094–1118. doi:10.1016/j.jhealeco.2006.03.001.
Easterlin, R. (2001). Income and happiness: toward a unified theory. The Economic Journal, 111, 465–484. doi:10.1111/1468-0297.00646.
Forshee, R., Anderson, P., & Storey, M. (2004). The role of beverage consumption, physical activity, sedentary behavior, and demographics on body mass index of adolescents. International Journal of Food Sciences and Nutrition, 55, 463–478. doi:10.1080/09637480400015729.
Frey, B., & Stutzer, A. (2002). Happiness and Economics. Princeton: Princeton University Press. Graham, C. (2005). Insights on development from the economics of happiness. The World Bank Research
Observer, 20, 201–231. doi:10.1093/wbro/lki010. Greene, W. (2003). Econometric analysis. Englewood Cliffs, NJ: Prentice Hall. Hager, R. (2006). Television viewing and physical activities in children. The Journal of Adolescent Health,
39, 656–661. doi:10.1016/j.jadohealth.2006.04.020. Hajivassiliou, V., McFadden, D., & Ruud, P. (1996). Simulation of multivariate normal rectangle proba-
bilities and their derivatives: theoretical and computational results. Journal of Econometrics, 72, 85– 134. doi:10.1016/0304-4076(94)01716-6.
Hsieh, P., & FitzGerald, M. (2005). Childhood obesity in Taiwan: review of the Taiwanese literature. Nursing & Health Sciences, 7, 134–142. doi:10.1111/j.1442-2018.2005.00218.x.
Hui, L., Nelson, E., Yu, L., Li, A., & Fok, T. (2003). Risk factors for childhood overweight in 6-to 7-y-old Hong Kong children. International Journal of Obesity, 27, 1411–1418. doi:10.1038/sj.ijo.0802423.
Kahneman, D., & Krueger, A. (2006). Development in the measurement of subjective well-being. The Journal of Economic Perspectives, 20, 3–24. doi:10.1257/089533006776526030.
Kan, K. (2007). Cigarette smoking and self-control. Journal of Health Economics, 26, 61–81. doi: 10.1016/j.jhealeco.2006.07.002.
Lin, C. (2002). The prevention of children’s mental health problem. Student Affairs, 83, 119–123. in Chinese.
Lin, B., Huang, C., & French, S. (2004). Factors associated with women’s and children’s body mass indices by income status. International Journal of Obesity, 28, 536–542. doi:10.1038/sj.ijo.0802604.
Lindberg, L., & Swanberg, I. (2006). Well-being of 12-year-old children related to interpersonal relations, health habits and mental distress. Scandinavian Journal of Caring Sciences, 20, 274–281. doi: 10.1111/j.1471-6712.2006.00405.x.
Lindsay, A., Sussner, K., Kim, J., & Gortmaker, S. (2006). The role of parents in preventing childhood obesity. The Future of Children, 16, 169–186. doi:10.1353/foc.2006.0006.
274 H.-H. Chang, R. M. Nayga Jr.
123
Lo, K. H., Pan, L. Y., Lin, M. C., Chang, H. Y., & Shih, Y. T. (2003). The analysis of special physical inactivity, higher energy intakes and body mass index of children in Taiwan: results of 2001 National Health Interview Survey. Taiwan Journal of Public Health, 22, 474–482.
Ludwig, D., Peterson, K., & Gortmaker, S. (2001). Relation between consumption of sugar-sweetened drinks and childhood obesity: a prospective, observational analysis. Lancet, 357, 505–508. doi: 10.1016/S0140-6736(00)04041-1.
Maddala, G. (1983). Limited dependent and quantitative variables in Econometrics. Cambridge: Cambridge University Press.
Marshall, T., Levy, S., Broffitt, B., Warren, J., Eichenberger-Gilmore, J., Burns, T., et al. (2003). Dental caries and beverage consumption in young children. Pediatrics, 112, e184–e191. doi:10.1542/peds. 112.3.e184.
McIntosh, A., Davis, G., Nayga, R., Jr., Anding, J., Torres, C., Kubena, K., Perusquia, E., Yeley, G., & You, W. (2006). Parental time, role strain, and children’s fat intake and obesity-related outcomes. Contractor and Cooperator Report No 263, Economic Research Service, US Department of Agriculture, Wash- ington DC.
Pan, L. Y., Chang, H. Y., & Shih, Y. T. (2003). The 2001 National Health Interview Survey: geographic variation and family aggregation of helmet use among children aged 6 to 12 in Taiwan. Taiwan Journal of Public Health, 22, 483–491.
Staiger, D., & Stock, J. (1997). Instrumental variables regression with weak instruments. Econometrica, 65, 557–586. doi:10.2307/2171753.
Taiwan Medical Association for the study of obesity (TMASO). (2007). Children’s obesity. http://www. obesity.org.tw.
Train, K. (2003). Discrete choice methods with simulation. New York: Cambridge University Press. Troiano, R., Briefel, R., Carroll, M., & Bialostosky, K. (2000). Energy and fat intakes of children and
adolescents in the United States: data from the National Health and Nutrition Examination Surveys. The American Journal of Clinical Nutrition, 72, 1343S–1353S.
World Health Organization. (2000). Obesity: preventing and managing the global epidemic-report of a WHO consultation. WHO Technical Report Series, 894.
Wu, T. (2001). Childhood obesity in Taiwan. Endocrinology and Dialectology, 14, 38–39.
Childhood Obesity and Unhappiness 275
123
Copyright of Journal of Happiness Studies is the property of Springer Science & Business Media B.V. and its
content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.
