Beruflich Dokumente
Kultur Dokumente
Incentives and
Appearance Maintenance
Uri Gneezy* and Jason Shafrin**
Abstract:
-1-
1. Introduction
individuals (Sobal, Rauschenbach and Frongillo, 1992; Hahn, 1993). We propose and test
one reason for this increase in body weight after entering into marriage: the incentives to
includes having a healthy body weight. In order to have a healthy body weight, many
People exert effort to maintain their appearance in order to attract a mate. The
maintaining appearances hypothesis says that upon entering into a monogamist domestic
relationship, one’s incentive to maintain attractive body weight decreases since one
already has a mate and is less active in the dating market. This is an incentive explanation
stating that getting into long term relationship decreases the incentives to maintain
weight.
13 year panel data set from the Netherlands. We exploit variation in the type of domestic
probability of termination will gain less weight than those who enter into domestic
We find that individuals who enter into cohabitation relationships gain less weight
than those who enter into traditional marriage relationships. Further, we observe that
-2-
having a child reduces the probability that a relationship terminates. The marginal impact
of having a child decreases the probability a couple separates more for cohabitators and
traditional marriages already had a lower separation probability so the marginal effect of
having a child is less than is the case for cohabitators or marriages with prenuptial
agreements. As our theory predicts, married individuals with a prenuptial agreement gain
more weight after having a child than would be the case when an individual in a
traditional marriage has a child. For cohabitators the results are imprecise because fewer
cohabitators have children than married couples. Overall, our results show a correlation
between the probability a domestic relationship will terminate and subsequent weight
gain.
2. Background
counterparts (Sobal, Rauschenbach and Frongillo, 1992; Hahn, 1993). This result persists
after controlling for age and other covariates. Interpreting this empirical finding,
however, is more difficult. Cross-sectional analyses do not take into account the fact that
marriage and obesity interact through two different mechanisms: marital selection and
marital causation (Sobal, Rauschenbach and Frongillo, 1992). Marital selection describes
the phenomenon that overweight or obese individuals are less likely to date or marry than
their healthy-weighted peers (Averett and Korenman, 1999; Gortmaker et al., 1993;
Cawley, Joyner and Sobal, 2006). As we have noted, however, it has been widely found
that married individuals have higher levels of BMI than non-married individuals. This is
-3-
due to the marital causation pathway. Marital causation claims that something about
being married directly affects an individual’s weight. It has been widely reported in the
medical literature that individuals who enter marriage gain weight and those who exit
marriage lose weight (Sobal, Rauschenbach and Frongillo, 1992; Rissanen et al., 1991;
Kahn and Williamson, 1990; Kahn and Williamson, 1991; Kahn, Williamson and
Stevens, 1991). These findings hold despite the fact that married individuals engage in
healthier behaviors and have lower mortality rates than their non-married peers
(Umberson, 1992).
What is it about being married that leads to weight gain? Academics have not
been lacking in explanations. Craig and Truswell (1988) claim that marriage may lead to
higher food intake which will cause more weight gain. Marriage may alter activity levels
as well. Verhoef, Love and Rose (1992) and Myers, Weigel, Holliday (1989) observe that
married individuals are less likely to be active or exercise, but this finding is not universal
(King et al. 1998). On average, marriage decreases smoking rates and smoking cessation
can induce weight gain (Waldron and Lye, 1989; Wee et al., 2001). The closest
explanation to the maintaining appearances hypothesis that we test can be found in Sobal
(1984), who proposes the possibility that marriage may reduce the incentive to maintain
Empirically testing this plethora of theories is not trivial. When individuals get
married, a variety of changes happen all at once. Statistically testing a single explanation
is often infeasible with available data. To solve this problem, we utilize variation in the
different domestic relationships. Individuals can either be married under the traditional
-4-
common property law, married with a prenuptial agreement, or they can live together but
not be officially married (i.e., cohabitation). It is most difficult for individuals to exit a
traditional marriage, and easiest for individuals to exit if the couple is cohabitating but
Our hypothesis is that individuals who enter a traditional marriage will gain the
most weight after entering into a domestic relationship because there is a lower
probability that their marriage will dissolve. The likelihood that those in a traditional
marriage will re-enter the dating market is low and thus individuals in this type of
domestic relationship will have less motivation to maintain their external appearance or
keep a healthy body weight. On the other hand, those who enter into a cohabitation
arrangement, but are not married will gain the least amount of weight upon entering into
a domestic relationship because the probability that the relationship will dissolve is the
highest among the three institutions. Thus, a higher probability that the relationship will
terminate will compel individuals to maintain their appearance and body weight.
3. Data
We use the DNB Household Survey (DHS) from the CentERdata of the
Netherlands. The data set is panel in nature and has 66,810 observations between 1995
and 2007. The dependent variable of interest is an individual’s body mass index (BMI).
Marriage status in the DHS is defined into one of six categories: 1) married under
-5-
The DNB Household Survey consists of 66,810 observations, but not all of the
observations are used in our empirical work. We eliminate 33,746 observations where the
height or weight of the individual is missing. Without these two variables, we cannot
calculate the person's BMI or measure whether or not they are obese. The sample is
limited to individuals who are between 15 and 65 years old in order to eliminate children
and the elderly. The reduced age range leads to 7,430 observations being deleted.
Observations with missing data for marital status (2,111), income or education (2,714)
variables did not materially affect the results. The deletion of extreme value of height,
weight, BMI or income reduced the sample by 73 observations, but this did not affect the
results of the coefficients of interest significantly. Thus, the sample that will be used in
the empirical portion of this paper has 20,736 observations for 8,067 unique individuals.
Table 1 shows the sample means for the variables used in subsequent regressions.
agreements make up 9.1% and 9.5% of individuals are cohabitators. Seventy five percent
of the individuals are in one of these three types of domestic relationships. The
remaining individuals can be classified into three other categories. Never married
individuals comprise 19.5% of the sample, 4.5% of the observations are divorced, and
In order to test our hypothesis we need to have variation in the type of marriage.
Among Dutch adults in relationships, 75.3% are in a traditional marriage, 12.1% are
married with a prenuptial agreement and 12.6% are cohabitators with no official marriage
-6-
marriage than in the U.S. Cohabitation acquired virtually equal status with marriage
under Dutch law in the 1980s. For example unlike in the United States, Dutch
Figure 1 shows the average BMI levels across marital status. The data on BMI
are qualitatively similar to the data from previous studies. Individuals in traditional
marriages weigh more than individuals in marriages with a prenuptial agreement, but
individuals in either type of marriage weigh more than cohabitators. Single individuals
have the lowest levels of BMI. Figure 1 also examines BMI by age as well. A clear
trend emerges that BMI increases with age. Controlling for age attenuates BMI
differences across marital status categories. Because individuals move in and out of
We propose that individuals who enter into traditional marriages will gain more
weight because the probability that they will re-enter the dating market is low. In order
to test this hypothesis, the type of domestic relationship must influence the probability a
couple will dissolve their relationship. Table 2 shows the results of this test. Using a
logit regression, we investigate if the probability of separation is correlated with the type
of relationship. We find that individuals who cohabitate are five times more likely to
terminate their relationship after three years than married individuals. The probability of
separation for married individuals with a prenuptial agreement is not statistically different
from people with a traditional marriage. Thus, we predict that cohabitators will gain less
-7-
Having children should also affect the probability of separation. Couples who
otherwise would separate may stay together for the sake of the children. The degree to
which children alter the probability a couple terminates their relationship depends on the
parent’s relationship. Couples who in more secure relationships have a lower probability
that their relationship will dissolve. For traditional marriages, the marginal effect of
having children on separation probabilities is low because the marriage was already
secure. On the other hand, having children may significantly decrease the separation
probabilities; couples with stable relationships are more likely to have children. Yet if
hypothesis predicts that having children should affect weight gain through the probability
a relationship terminates.
Columns 2 and 3 of Table 2 show the results of two logit regressions describing
how having children affects the probability a marriage dissolves. Column two shows that
having children reduces the probability that after three years a couple has separated.
Couples with children are half as likely to separate as those without children. Although
not shown in the table, having more than one child has no marginal effect on separation
probabilities after taking into account having had the first child.
similarly across all three types of domestic relationships. Cohabitators are still more
likely to terminate their relationship than other couples. However, the marginal effect of
having children reduces the probability a couple separates significantly more for
-8-
cohabitating couples and married couples with prenuptial agreements compared to
couples in a traditional marriage. Having a child reduces the odds of separation about
four times more for cohabitators than for traditional marriages. For marriages with
prenuptial agreements, the effect is even larger. Having a child reduces the odds of
separation almost ten times more for marriages with prenuptial agreements than those
without.
1. Individuals who enter into a cohabitation arrangement should gain less weight
2. Having children should lead to a larger weight gain for cohabitators and married
marriages.
In order to test whether the type of domestic relationship affects body weight, we
utilize a fixed effects regression. Using ordinary least squares would bias the results due
to the fact that healthier people generally select into marriage. The fixed effects
-9-
The fixed effects regression is valid if the change in marital status is uncorrelated
with unobserved changes that also affect weight. This strong condition is unlikely to
hold. Since our major research question is to test how weight changes across domestic
similar for cohabitators, and both groups of married individuals. In other words, we
assume that sharing a residence with your significant other has a similar affect on weight
for all three groups studied. We attribute any additional differences in weight gain across
Column 1 of Table 3 shows the results of the fixed effects regression. People gain
weight when entering any of the three domestic relationships. BMI increases by 0.51
after individuals become married and by 0.53 after individuals are married with a
prenuptial agreement. When people begin to cohabitate, however, BMI increases by only
0.27. For individuals of average height, these coefficients imply that the average person
gains 1.7 kg (3.7 lb) after they get married, but would only have gained 0.9 kg (1.9 lb.) if
they had decided to cohabitate without being officially married.1 Weight increases are
significantly less than for cohabitators than for married couples (p<.045). Weight gain
after marriage was similar between individuals who had a prenuptial agreement and those
who did not (p<.783). Cohabitators have a significantly higher probability that their
relationship will dissolve than either of the married groups. Thus, cohabitators have a
strong incentive to maintain their weight because the probability they we re-enter the
1
BMI = (Weight in kg)/(Height in m)2. Thus, multiplying the regression coefficients by the average height
squared, (1.753)2, recalibrates the coefficients in terms of the change in weight for the average person.
- 10 -
It is possible that married individuals are more likely to have children and thus
the increased weight gain upon getting married may be caused entirely by pregnancy. In
order to control for this possibility, we examine males and females separately. Columns
2 of Table 3 show the results for males and column 3 shows the results for females.
BMI. On the other hand, females gain significant weight after settling down. BMI
increases by 1.3 after entering marriage, but only increases by 0.7 when woman enter a
cohabitating relationship. Weight gain for cohabitators is significantly less than for either
Are increased pregnancy rates during marriage driving the results here? Column
4 looks at the same fixed effects regression for females who did not become pregnant
during the sample. Females who did not become pregnant still gained weight after
entering a domestic relationship, but those who entered into a cohabitation relationship
gained less weight. Cohabitators increased BMI by 0.5 compared to 0.8 for traditional
marriages and 0.9 for marriages with a prenuptial agreement. Non-pregnant women who
entered into cohabitation arrangements gained significantly less weight that those in
Marriage dissolution probabilities affect body weight more for females than for males.
Having children reduces the probability a couple separates more for cohabitators
and married individuals with a prenuptial agreement than for individuals in a traditional
marriage. According to our second hypothesis, greater increases in BMI should follow
- 11 -
from larger decreases in the probability of separation. Having a child significantly
reduces the separation probabilities of both cohabirators and marriage individuals with a
the dating market was already low so the marginal impact of having a child is smaller.
Our empirical strategy is to use a fixed effect regression to see if people who
cohabitate or have prenuptial agreements gain more weight after have a child. Let Δz=(zit
greater than the coefficient on (Δtraditional marriage)(Δkids). The marginal weight gain
from having kids should be larger for couples in previously less committed relationships.
Column 1 of Table 4 shows the results of this regression. The marginal weight
gain from having kids is indeed larger for individuals whose marriage has a prenuptial
agreement compared to than individuals whose marriage does not contain a prenuptial
agreement (p<.009). Separation probabilities change more after having children for
married couples with a prenuptial agreement than couples in traditional marriages and
thus weight gain after the birth of a child should be larger for married couples who have a
Although we find a large change in weight after the birth of a child for couples
- 12 -
cohabitators compared to those with traditional marriage (p<.434). We expected that
after the birth of a child, cohabitator’s weight would increase more than if they were in a
traditional marriage. The data does not bear this out. Since cohabitators have fewer
children, however, there is less variation in the right hand side variable. Thus it is not
surprising that the empirical estimates of the marginal impact of having a child for
The second and third columns of table 4 display the same fixed effects regressions
separately for males and females. For both sexes, having a child increases BMI more for
Once again the coefficients on the interaction of the cohabitation and kids variables are
imprecisely measured.
7. Conclusion
Many studies have found that married individuals weigh more than their non-
married peers. Our paper attempts to answer why this may be the case. Maintaining a
healthy body weight involves costly activities such as exercise and food preparation and
the payoff to “looking good” is likely higher for individuals on the dating market than
those who have exited through marriage. Our paper looks at variation in the type of
you will re-enter the dating market. A high probability that you will return to the dating
market increases one’s incentive to maintain his body weight; a low probability of
returning to the dating market decreases one’s incentive to maintain his body weight.
- 13 -
Our data provide evidence that the probability of re-entering the dating market
directly effects body weight. Cohabitatiors are the most likely to re-enter the dating
market. We find that individuals who being to cohabitate gain less weight than
individuals who get married. Children also affect the probability a relationship dissolves.
The marginal effect of having kids on relationship dissolution probabilities is larger for
cohabitators and married couples with prenuptial agreements. After a couple with a
prenuptial agreement has a child, they tend to gain more weight than would be the case if
the same couple had a child under a traditional marriage scenario. Since the traditional
marriage was more secure to begin with, having a child does not affect relationship
dissolution probabilities nearly as much as in the case for marriage with prenuptial
agreement. As our theory predicts, weight gain after having a child is higher for married
Understanding the mechanisms through which weight gain occurs is important for
might fund programs which destabilize marriage in hopes of reducing of obesity. While
obesity rates may drop as a result of these programs, this is not a wise policy to pursue.
Numerous studies have shown that the health benefits from marriage are great and likely
outweigh the cost marriage imposes on society from increased obesity (Hahn 1993,
Umberson 1992). This paper contributes to the understanding of how weight gain occurs.
Extending this paper to find practical policies which could maintain marriage stability
- 14 -
REFERENCES
Burkhauser, Richard V. and John Cawley. 2008. "Beyond BMI: The Value of More
Accurate Measures of Fatness and Obesity in Social Science Research" Journal of Health
Economics, 27(2): 519-529.
Cawley, John, Kara Joyner, and Jeffery Sobal. 2006. "Size Matters: The Influence of
Adolescents' Weight and Height on Dating and Sex" Rationality and Society, 18(1):
67-94.
Craig, P. L. and A. S. Truswell. 1988. "Changes in Food Habits when People Get
Married: Analysis of Food Frequencies." In Food Habits in Australia, ed. A. S. Truswell
and M. L. Wahlqvist, 151-159. Melbourne.
Elder, Glen H.,Jr. 1969. "Appearance and Education in Marriage Mobility" American
Sociological Review, 34(4): 519-533.
Finkelstein, Eric A., Ian C. Fiebelkorn, and Guijing Wang. 2003. "National Medical
Spending Attributable to Overweight and Obesity: How Much, and Who's Paying?"
Health Affairs.
Gortmaker, Steven L., Aviva Must, James M. Perrin, Arthur M. Sobol, and William
H. Dietz. 1993. "Social and Economic Consequences of Overweight in Adolescence and
Young Adulthood" The New England Journal of Medicine, 329(14): 1008-1012.
Hahn, Beth A. 1993. "Marital Status and Women's Health: The Effect of Economic
Marital Acquisitions" Journal of Marriage and the Family, 55(2): 495-504.
- 15 -
Kahn, H. S. and D. F. Williamson. 1990. "The Contributions of Income, Education and
Changing Marital Status to Weight Change among US Men" International journal of
obesity, 14(12): 1057-1068.
------. 1991. "Is Race Associated with Weight Change in US Adults After Adjustment for
Income, Education, and Marital Factors?" The American Journal of Clinical Nutrition,
53(6 Suppl): 1566S-1570S.
Kahn, H. S., D. F. Williamson, and J. A. Stevens. 1991. "Race and Weight Change in
US Women: The Roles of Socioeconomic and Marital Status" American Journal of
Public Health, 81(3): 319-323.
Mokdad, Ali H., Mary K. Serdula, William H. Dietz, Barbara A. Bowman, James S.
Marks, and Jeffrey P. Koplan. 1999. "The Spread of the Obesity Epidemic in the
United States, 1991-1998" JAMA: The Journal of the American Medical Association,
282(16): 1519-1522.
Rand, C. S., J. M. Kuldau, and L. Robbins. 1982. "Surgery for Obesity and Marriage
Quality" JAMA: The Journal of the American Medical Association, 247(10): 1419-1422.
Sobal, J., B. S. Rauschenbach, and E. A. Frongillo Jr. 1992. "Marital Status, Fatness
and Obesity" Social science and medicine, 35(7): 915-923.
- 16 -
Sobal, Jeffery, Barbara Rauschenbach, and Edward A. Frongillo. 2003. "Marital
Status Changes and Body Weight Changes: A US Longitudinal Analysis" Social Science
and Medicine, 56(7): 1543-1555.
Sturm, Roland. 2003. "Increases in Clinically Severe Obesity in the United States,
1986-2000" Archives of Internal Medicine, 163(18): 2146-2148.
Umberson, D. 1992. "Gender, Marital Status and the Social Control of Health Behavior"
Social Science and Medicine (1982), 34(8): 907-917.
Verhoef, M. J., E. J. Love, and M. S. Rose. 1992. "Women's Social Roles and their
Exercise Participation" Women and Health, 19(4): 15-29.
Waldron, I. and D. Lye. 1989. "Family Roles and Smoking" American Journal of
Preventive Medicine, 5(3): 136-141.
Wee, Christina C., Nancy A. Rigotti, Roger B. Davis, and Russell S. Phillips. 2001.
"Relationship between Smoking and Weight Control Efforts among Adults in the United
States" Archives of Internal Medicine, 161(4): 546-550.
- 17 -
Wolfers, Justin and Betsey Stevenson. 2007. "Marriage and Divorce: Changes and their
Driving Forces" Journal of Economic Perspectives, 21(2): 27-52.
Wooldridge, Jeffrey M. 2001. Econometric Analysis of Cross Section and Panel Data,
First ed: The MIT Press.
Yang, Zhou and Allyson G. Hall. 2008. "The Financial Burden of Overweight and
Obesity among Elderly Americans: The Dynamics of Weight, Longevity, and Health
Care Cost" Health services research, 10.1111/j.1475-6773.2007.00801.x.
- 18 -
TABLES
Marriage Status
Traditional Marriage 0.568 0 1
Marriage with a Prenup 0.091 0 1
Cohabitation 0.095 0 1
Divorced 0.045 0 1
Widowed 0.007 0 1
Kids 1.21 1.23 0 7
Smoker 0.316 0.465 0 1
Male 0.515 0 1
Age 41.1 11.1 18 65
Primary 0.102 0 1
Pre-vocational 0.155 0 1
Education
Pre-university 0.120 0 1
Vocational 0.461 0 1
University 0.148 0 1
Other 0.014 0 1
Income (loon) 35.0 39.8 0 850
Year 2000.0 4.1 1995 2007
n 20518
Table 2
- 19 -
Table 3: Your significant other and Weight Gain
Female -
Not
Total Male Female Pregnant
- 20 -
Table 4: The marginal effect of having kids on BMI
Total Male Female
- 21 -
Figure 1
27.0
26.0
24.0
Cohabitation
23.0 Single
22.0
21.0
Total 15-30 31-40 41-50 51-65
Age
- 22 -