Direct and indirect effects of body weight on adult wages

Economics and Human Biology 9 (2011) 381–392

Contents lists available at ScienceDirect

Economics and Human Biology journal homepage: http://www.elsevier.com/locate/ehb

Direct and indirect effects of body weight on adult wages Euna Han a,*, Edward C. Norton b,c, Lisa M. Powell d a

College of Pharmacy, Gachon University of Medicine and Science, Gachon Building, Room 1009, 534-2 Yeonsu3-dong, Yeonsu-gu, Incheon 406-799, South Korea Department of Health Management and Policy, University of Michigan, 1415 Washington Heights, M3108 SPH II, Ann Arbor, MI 48109, United States c Department of Economics, University of Michigan, 611 Tappan St., Lorch Hall, Ann Arbor, MI 48109, United States d Institute for Health Research and Policy and Department of Economics, University of Illinois at Chicago, 1747 West Roosevelt Road, Room 449, Chicago, IL 60608, United States b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 3 May 2010 Received in revised form 9 May 2011 Accepted 1 July 2011 Available online 4 August 2011

Previous estimates of the association between body weight and wages in the literature have been conditional on education and occupation. In addition to the effect of current body weight status (body mass index (BMI) or obesity) on wages, this paper examines the indirect effect of body weight status in the late-teenage years on wages operating through education and occupation choice. Using the National Longitudinal Survey of Youth 1979 data, for women, we find that a one-unit increase in BMI is directly associated with 1.83% lower hourly wages whereas the indirect BMI wage penalty is not statistically significant. Neither a direct nor an indirect BMI wage penalty is found for men. However, results based on clinical weight classification reveal that the indirect wage penalty occurs to a larger extent at the upper tail of the BMI distribution for both men and women via the pathways of education and occupation outcomes. Late-teen obesity is indirectly associated with 3.5% lower hourly wages for both women and men. These results are important because they imply that the total effect of obesity on wages is significantly larger than has been estimated in previous cross-sectional studies. ß 2011 Elsevier B.V. All rights reserved.

JEL classification: J31 I19 Keywords: BMI Obesity Wages Indirect effect Education Occupation choice

1. Introduction Over the past few decades, the prevalence of obesity has risen dramatically in the United States. One-third of American adults were reported to be obese in 2007– 2008 (Flegal et al., 2010). Given these trends, there is great interest in the economic consequences of obesity. Many economic studies have recently reported finding a negative effect of body mass index (BMI) or obesity on labor market outcomes, such as hourly wages, particularly for women (Averett and Korenman, 1996; Pagan and Davila, 1997; Cawley, 2004; Baum and Ford, 2004; Conley and Glauber,

* Corresponding author. Tel.: +82 32 820 4766; mobile: +82 10 9334 7870. E-mail addresses: [email protected], [email protected] (E. Han), [email protected] (E.C. Norton), [email protected] (L.M. Powell). 1570-677X/$ – see front matter ß 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.ehb.2011.07.002

2005; Norton and Han, 2008; Han et al., 2009) and the probability of employment (Sarlio-Lahteenkorva and Lahelma, 1999; Cawley, 2000; Paraponaris et al., 2005; Tunceli et al., 2006; Garcia and Quintana-Domeque, 2007; Lundborg et al., 2007; Morris, 2007; Norton and Han, 2008; Han et al., 2009). Economists are especially interested in understanding why BMI or obesity may affect labor market outcomes. Most explanations are conditional on having a job, and may explain differences either in initial wages or in wage growth. In addition to the relationship between current weight status and wages conditional on education and occupation, in this study, we provide empirical evidence on the wage penalty stemming from weight status in the lateteen years operating through two indirect pathways – education and occupation outcomes. Many important decisions that affect future employment and wages are made during the late teens and a large number of economic and vocational studies have explored whether behavioral

382

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

choice and the environment of teenagers affect their education or occupational outcomes in adulthood (Blau et al., 1956; Heckman et al., 2006; Holland, 1985; Schoon, 2001; Schoon and Parsons, 2002). Our study builds on the previous literature and empirically tests whether in addition to the direct weight-related wage penalty, late-teen body weight status prior to the job search is associated with wages indirectly through two major determinants of wages, education and occupation (Mincer, 1974), the formation of which occurs in early adulthood in the life trajectory. Several previous studies that explore the association of choices in the lateteen years with occupation outcomes focus on young adulthood (see for example, Borghans et al., 2007; Heckman et al., 2006). Examining the relationship between body weight status in the early part of the life-cycle (i.e., the late teenage years) and future education, employment, and occupation outcomes in young adulthood, we are able to identify whether different levels in the stock of education or sorting into different occupations are potential mediating paths that contribute to the BMI wage penalty. Evidence on this indirect effect would substantially add to the debate about the relationship between weight and wages, highlighting the potential persistent negative externalities from obesity prevalence in the teenage years. Such evidence would also imply that the total effect of obesity on wages is significantly larger than has been estimated in previous cross-sectional studies. We examine the extent to which individuals with different body weight status have different labor market outcomes, particularly, hourly wages in their early thirties, and whether such differences may be attributable to education and occupation outcomes among service, sales, managerial or professional specialty, administrative support or clerical, or blue-collar jobs. That is, our study helps to understand the potential of an indirect effect of weight status in the late-teenage years on wages operating through education and occupation in addition to assessing the direct association of weight status on wages. We also build on a hypothesis from the psychology literature about the importance of social interactions (Frieze et al., 1990; Martel and Biller, 1987). Personal traits such as interactive skills have been explored as a source of individual heterogeneity that is related to economic behaviors in the economics literature as well (Borghans et al., 2008; Heckman et al., 2006; Mueller and Plug, 2006; Krueger and Schkade, 2005). In occupation outcomes, such traits are reported to affect both the productive capabilities and the individual preferences for occupations (Ham et al., 2009; Borghans et al., 2007). The lack of accumulation of some forms of human capital – such as interpersonal skills – may be due to stigma or lack of self-image during the teenage years. Obese children are reported to have social and psychological problems such as stigma and poor-self esteem (Daniels, 2005). The correlation of physical attractiveness (for which BMI or height may be an objective proxy measure) and less human capital accumulation during high school years also has been reported in economics literature. For example, Moan and Tekin (2009) show that less attractive high school students accumulate less human capital, particularly for female

students, which is correlated with their labor market outcomes in later years. Persico et al. (2004) show that taller people earn higher wages stemming from higher human capital such as good social skills built from more sport or club participation in high school compared to their shorter counterparts. Mobius and Rosenblat (2006) report physically attractive workers show more self-confidence. Han and colleagues (2009) also report larger negative relationships between adult contemporaneous BMI and wages in occupations requiring interpersonal skills based on the nature of relationship to people. Therefore, we categorize occupations by the amount of social interactions required using the Dictionary of Occupational Titles (DOT) in addition to Census occupation codes. The aggregate Census occupational codes generate occupation categories based on overall characteristics of each occupation (Pagan and Davila, 1997; Baum and Ford, 2004). Given that occupations in the same Census classification can have different requirements for social interaction with customers or colleagues, this additional control may partly explain the BMI wage penalty. The results from this study show that a one-unit change in BMI is associated with 1.8% direct wage penalty for women. No direct BMI or obesity wage penalty is found for men. It is notable that the indirect BMI wage penalty occurs at the upper tail of the late-teen BMI distribution given that we estimate a 3.5% wage penalty of obesity via the indirect pathways of education and occupation outcome for both women and men. 2. Previous literature Several studies have linked body weight status to labor market outcomes, mostly wages. Most of those studies find a negative contemporaneous effect of body weight status (either BMI or obesity) on hourly wages for women, but no significant effect for men (Cawley, 2004; Averett and Korenman, 1996; Conley and Glauber, 2005; Baum and Ford, 2004; Han et al., 2009). For women, the wage penalty for a one-unit increase in BMI is found not only for their own earnings and occupational prestige, but also their spouses’ earnings and occupational prestige (Conley and Glauber, 2005). The direction and the magnitude of the effects are different by race within each gender (Cawley, 2004). Also, the often-reported negative relationship between BMI and wages is larger in occupations requiring interpersonal skills with presumably more social interactions (Han et al., 2009). Body weight status is also estimated to affect labor market outcomes on the extensive margin, such as employment or occupation outcomes. Morris (2007) estimates a negative relationship between obesity and the probability of employment for British people for both genders. Increasing BMI also is estimated to raise the percentage of time spent unemployed during the working years and lowers the probability of employment after a period of unemployment for both genders for French people (Paraponaris et al., 2005). Both obese men and women are less likely to sort into managerial, professional and technical occupations (Pagan and Davila, 1997). However, the effect of obesity is not statistically significant

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

for limitations on the amount or types of paid work (Cawley, 2004). Two papers seek to disentangle the supply side from the consumer side effect in the association of obesity with labor market penalties. While no study carves out the consumer side labor market penalty related to obesity, Harper (2000) shows a positive effect of being physically attractive on the probability of employment for women, particularly on the probability of working in a managerial, professional, or clerical occupations. However, the study finds neither occupational sorting into customer-oriented occupations caused by physical attractiveness nor a wage penalty for non-attractive women working in customeroriented jobs. Carpenter (2006) compares the employment rate for obese people to normal-weight people before (1988) and after (1999) a 1993 court case, Cook vs. Rhode Island. In that case, obesity is ruled to be covered under the Rehabilitation Act of 1973 and the Americans with Disabilities Act for the first time by a federal appeals court. The data show that the employment rate increased 4% for obese women and 2% for obese men. There are a few papers that estimate the effect of obesity on educational achievements. Some studies report no effect of obesity on GPA (Crosnoe and Muller, 2004), grade progression and drop out status (Kaestner et al., 2009) among adolescents. Sabia (2007) reports that only white female adolescents have a GPA penalty for being obese. 3. Mechanisms for the BMI wage penalty The general finding that current weight status affects women’s wages could be due to differences in labor productivity, compensating wages for higher health care costs, using weight status as an observable measure of an unobservable trait (investment in human capital) that employers value, or pure discrimination. These explanations have all been proposed in the previous literature (Baum and Ford, 2004; Bhattacharya and Bundorf, 2009). Concerns about customers’ distaste for overweight or obese employees – particularly in sales – may also make employers shun overweight or obese employees. Everett (1990) and Puhl and Brownell (2001) demonstrate that employers perceived obese persons as unfit for public sales positions and as more appropriate for telephone sales involving little face-to-face contact. Carr and Friedman (2005) also report that class II obese (BMI  35) workers are more likely to report job-related discrimination in the workplace due to their weight and appearance, particularly, in ‘‘upper white-collar’’ occupations (including professional, executive, and managerial occupations). We argue that while these explanations are all important they may miss a long-run effect based on decisions in the past. If late-teen weight status is correlated with mid-life weight status, then late teens have a signal of their future weight status. This could be due to genetics, environment, behavior in terms of choices of diet and exercise, or a combination of all three. A heavy-set forward-looking teen may rationally decide to invest less in education than her trim classmate. A wage penalty lowers the returns to

383

education for people with higher BMI. Therefore, people with high BMI may invest less in education. The perception about the wage penalty may or may not be real – either way if the teen acts on that perception then late-teen weight status could affect investment in education, with lighter teens investing more. This is all due to a perceived future wage penalty for higher weight status. This economic model that underlies the indirect BMI wage penalty is a form of statistical discrimination, and has been argued in similar contexts. If overweight or obese teenagers presciently observe smaller returns to human capital for overweight or obese people, then they may logically invest less in their human capital (Persico et al., 2004). Late-teen weight status may also affect choice of occupation, which could also result in the finding of a wage penalty. Suppose that the wage penalty is heterogeneous, being strong in high-paying jobs and nonexistent in low-paying jobs. Then a heavy person may decide to avoid an occupation with a high wage penalty so as to be in a job with no wage penalty. If the occupation with a high wage penalty also has high average salaries compared to the other occupation, then the observed wage difference may be explained by occupation sorting. In summary, we have described an economic model in which a future wage penalty for obesity affects current decisions about education and occupation choice because of the lower returns to these investments. This model implies that empirical work should try to separately identify direct and indirect effects of body weight on adult wages, and that the total effect could well exceed the direct effect already measured in previous studies. 4. Methods 4.1. Empirical models We present an empirical model that predicts wages as a function of the direct effect of weight status (measured by BMI or obesity), the indirect effect of weight status in the late teenage years operating through education and occupation outcome, and other factors. Following the labor literature, we take the logarithm of wages and estimate separate models for men and women. The main model is lnðWagei2 Þ ¼ g 0 þ g BMI BMIi2 þ g Educ Educi2 ðBMIi1 Þ þ g Occ þg

Census

Occ SI

OccCensi2 ðBMIi1 ; Educi2 ðBMIi1 ÞÞ

OccSIi2 ðBMIi1 ; Educi2 ðBMIi1 ÞÞ

X

þ g X i2 þ ei2 (1) where the subscripts i, 1, and 2 stand for individual, time 1 at the late-teenage years between age 16 and 20, and time 2 in the early thirties (hereafter also referred to as early career), respectively. The gs are parameters to be estimated, and e denotes the error term. BMI is a continuous measure of BMI. Educ is a continuous variable measuring the highest grade completed. OccCens is a vector of dummy variables measuring occupation outcome among service, sales, managerial or professional specialty,

384

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

administrative support or clerical jobs, with blue-collar jobs as the reference category based on aggregate Census occupational codes. OccSI is a dichotomous indicator for jobs requiring social interactions with colleagues or customers based on the Dictionary of Occupational Titles. Other covariates known to affect wages, such as demographics and the regional unemployment rate, are denoted by X. The direct association of a unit change in current BMI on the logarithm of wages in the early career stage is gBMI in Eq. (1), and this is what the prior literature conventionally estimates as the direct association of BMI with log wages controlling for both education and occupation outcomes. The indirect effect of BMI in late teenage years through education and occupation outcomes in the early career is shown by the full set of derivatives of the logarithm of wages with respect to BMI1 as in Eq. (2) below.   d lnðWageÞ2 @ lnðWageÞ2 @Educ2 ¼  dBMI1 @Educ2 @BMI1   @ lnðWageÞ2 @OccCens2  þ @OccCens2 @BMI1   @ lnðWageÞ2 @OccSI2  þ @OccSI2 @BMI1   @ lnðWageÞ2 @OccCens2 @Educ2   þ @OccCens2 @Educ2 @BMI1   @ lnðWageÞ2 @OccSI2 @Educ2   þ @OccSI2 @Educ2 @BMI1 (2) The first term in Eq. (2) estimates the indirect BMI wage penalty that operates through education. The second and third terms are the indirect BMI wage penalties through occupation outcomes, measured by the Census occupational categories (the second term) and the occupational characteristics of requiring social interactions with customers or colleagues (the third term). The last two terms account for the indirect effect of BMI on wages through education on occupation outcomes. To calculate the terms in Eq. (2), we estimate models to predict education and occupation as a function of late-teen BMI. We begin by assessing the effect of late-teen BMI on the stock of education as the years of schooling accumulated by the time an individual reaches their early thirties using OLS as in Eq. (3). Educi2 ¼ a0 þ aBMI BMIi1 þ aX X i2 þ hi2

(3)

where the as are the parameters to be estimated, h denotes the error term, and other variables are defined as before. Next, we estimate multinomial logit models of the effect of late-teen BMI on occupation outcomes in the early career based on the Census occupational codes (Eq. (4)) and the DOT codes for the requirement of social interaction (Eq. (5)). 0

OccCensi2 ¼ d þ d

BMI

Educ

BMIi1 þ d

Educi2 ðBMIi1 Þ

X

þ d X i2 þ ni2 0

OccSIi2 ¼ b þ b þ & i2

BMI

BMIi1 þ b

Educ

where the ds and bs are the parameters to be estimated, n and denote the error terms, and other variables are as defined before. After estimating Eqs. (1), (3), (4) and (5), we fill in values for Eq. (2) to estimate the full derivative of wages in the early career with respect to late-teen BMI. We bootstrap the standard errors of all calculations based on Eq. (2). We also run all equations using indicators for obesity (BMI  30 kg/m2) instead of the linear measure of BMI, and calculate the indirect and direct portion of obesity wage penalty as discrete differences between obesity and nonobesity. We control for state-level labor market characteristics for all the estimations, including per capita average income, the Consumer Price Index, and the unemployment rate in the resident’s area at the time of survey. State-level macroeconomic conditions control for time-varying macroeconomic shocks at the state level that could affect both individual health conditions and participation in the labor market (Ruhm, 2000). We estimate all models of log wages using sibling fixed effects to help control for unobserved heterogeneity. In estimating indirect pathways of BMI on wages via education and occupation choice, the measurement for our variable of interest (lateteen BMI) precedes the measurement of the dependent variables. This lagged structure is often used in the previous literature when the relationship of body weight and labor market outcomes is explored (Averett and Korenman, 1996; Gortmaker et al., 1993; Norton and Han, 2008). We also control for family environment through parents’ level of education that may affect teenagers’ time use, health behaviors, investment in human capital, and occupational aspirations. Previous studies imply that parents’ achievement, which is measured with parents’ level of education in this study, can influence individuals’ labor market outcomes through intergenerational transfers of achievements (Bradley, 1991; Bourdieu and Passeron, 1977; Borghans et al., 2008; Bynner, 1998). We control for self-esteem and the level of intelligence that may be correlated with both late-teen BMI and wages at the early career. In addition, we control for experience in the job market as the years of employment and on-the-jobtraining as a dichotomous indicator for participating in any vocational trainings as covariates. However, despite controlling for many factors, we cannot control for all genetic and environmental factors that affect career decisions, education, diet and exercise. We also acknowledge that such lagged structure does not fully address the potential endogeneity bias stemming from any omitted variables, which is likely to cause a downward bias in the estimated coefficients in the wage regression. Our results cannot be generalized for all people given that we restrict our sample to individuals with valid wage information in all estimations and do not estimate the unconditional effect of independent variables on wages.

(4)

4.2. Data

(5)

We draw individual-level data from the NLSY79, a nationally representative sample of 12,686 young men and women who were 14–22 years of age at the initial survey

X

Educi2 ðBMIi1 Þ þ b X i2

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

in 1979. Four years of data (1981, 1982, 1985, and 1986) are pooled to create averaged late-teen BMI (the first two waves of NLSY79, 1979 and 1980, are not used because of lack of height information). For current BMI, we average BMI between age 30 and 34 spanning seven survey waves of the NLSY79 including the 1989, 1990, 1992, 1993, 1994, 1996, and 1998 waves. Our four primary dependent variables are wages, the stock of education, occupation outcome based on the Census occupational classification and occupation outcome regarding occupations requiring social interactions. These four outcomes are defined at age 30 when possible (drawn from the 1989, 1990, 1992, 1993, 1994, 1996, and 1998 waves of the data); we filled any missing values with information until age 34.

385

Wages are measured by the hourly rate of pay at the current job. We deflate hourly wages to year 2000 dollars. For the stock of education, we use a continuous measure of the years of schooling completed. The summary statistics in Table 1 show that by one’s early career, men and women earn on average $13.6 and $12.1 per hour, respectively (in 2000 dollars). The average years of schooling completed is approximately 14 years for both genders. We categorize individuals by their occupational characteristics. Employment and occupational outcomes have the following five categories: (1) employed in service occupations; (2) employed in managerial or professional specialty occupations;

Table 1 Weighted summary statistics. Variables

Women Mean

Dependent variables Hourly wages Stock of education: highest grade completed Occupation outcome based on Census codes Service Managerial or professional specialties Sales Administrative support/clerical Blue-collar Occupation outcome based on requirement of social interaction Independent variable of interest BMI in the late-teenage years Obese in the late-teenage years BMI Obese Race White Black Hispanic Highest grade completed by mother Highest grade completed by father Married Number of children AFQT scores Years of employment Participated any vocational training Height (m) Pregnancy Being pregnant within 2 years from the time of interview Being pregnant between 2 and 4 years from the time of interview Being pregnant between 4 and 6 years from the time of interview Not being pregnant within 6 years from the time of interview Mean age for the late teen BMI observation Time between the teen BMI observation and the adult wage observation Regional variables Unemployment rate: <6% Unemployment rate: 6–9% Unemployment rate: 9–2% Unemployment rate: >12% Number of private businesses per 10,000 at the state-level State per capita average yearly income in $1000 The Consumer Price Index West North-central South Urban N

906

Men SD

Range

Mean

SD

12.06 14.03

9.12 2.35

13.60 13.72

12.78 2.51

[1, 660.69] [3, 20]

0.19 0.35 0.09 0.23 0.14 0.70

0.39 0.47 0.29 0.43 0.36 0.50

0.11 0.26 0.08 0.07 0.48 0.58

0.31 0.44 0.27 0.23 0.50 0.50

[0, [0, [0, [0, [0, [0,

23.83 0.08 26.59 0.24

4.18 0.29 6.37 0.43

26.24 0.18 27.22 0.22

4.34 0.39 4.62 0.41

[15.62, 57.66] [0, 1] [15.78, 62.95] [0, 1]

0.58 0.26 0.16 11.92 12.03 0.60 1.32 51.77 9.32 0.48 1.64

0.49 0.44 0.38 3.15 3.88 0.49 1.10 27.23 1.80 0.50 0.07

0.60 0.23 0.17 12.07 12.32 0.61 1.06 49.68 9.70 0.42 1.79

0.49 0.43 0.38 3.15 3.84 0.50 1.11 29.39 1.38 0.49 0.08

[0, 1] [0, 1] [0, 1] [0, 20] [0, 20] [0, 1] [0, 3] [1, 99] [0, 11] [0, 1] [1.22, 1.98]

0.23 0.29 0.35 0.13 18.66 11.80

0.42 0.45 0.48 0.38 0.60 0.99

N/A N/A N/A N/A 18.64 11.71

N/A N/A N/A N/A 0.60 0.85

[0,1] [0, 1] [0, 1] [0, 1] [17, 19.50] [10.50, 16.50]

0.40 0.41 0.10 0.09 250.59 21.90 1.51 0.19 0.25 0.40 0.80

0.49 0.49 0.30 0.28 22.46 2.57 0.06 0.39 0.43 0.49 0.40

0.40 0.41 0.11 0.08 251.17 21.99 1.51 0.20 0.26 0.36 0.81

0.49 0.49 0.32 0.27 23.29 2.51 0.06 0.40 0.44 0.48 0.40

[0, 1] [0, 1]

1068

1] 1] 1] 1] 1] 1]

[0, 1] [205.35, 362.10] [15.42, 30.12] [1.39, 1.58] [0, 1] [0, 1] [0, 1] [0, 1]

386

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

(3) employed in sales occupations; (4) employed in administrative support or clerical occupations; and (5) employed in blue-collar occupations. We adopt the aggregate occupational categories in the 1980 Census occupational classification system (U.S. Census Bureau, 1996). A blue-collar occupation is an aggregate category of the following census occupational categories: technicians and related support; farming, forestry, and fishing; precision production, craft, and repair; and operators, fabricators, and laborers. The majority of employed women (with positive wages) in their early thirties have either managerial or professional specialty occupations (35%) or administrative support or clerical occupations (23%). For men, the most frequent aggregate occupational category in their early thirties is blue-collar occupations (48%), followed by managerial or professional specialty occupations (26%) (Table 1). Not shown in the tables are the distributional differences among the Census occupation categories for some key variables. Our final data show that young adult women in managerial and professional service and sales categories have lower late-teen BMI (23.7 and 23.9, respectively) than women in service (average late-teen BMI 24.7), administrative support (24.0), or blue collar (24.7) occupations. Also, in each occupation the proportion of women who were obese in their late teen is lowest in sales (4.8%) followed by managerial and professional service (6.6%). For men, the average late-teen BMI is lowest in sales category (25.8) followed by managerial and professional service (25.9). Similar to women, 12.9% and 15.6% of men in managerial and professional service and sales, respectively, were obese in late teen whereas 18– 21% of men in other occupation categories were obese in late teen. Further, we categorize occupations by whether they require social interaction with customers or colleagues and define the employment and occupation outcome by the following two categories: (1) employed in occupations requiring social interactions and (2) employed in occupations not requiring social interactions. 0We use the DOT to identify occupations requiring social interactions with customers or colleagues. The DOT was developed for standardizing occupational information by the U.S. Employment Service. Each DOT code has information on the relationship to people, which has nine categories: mentoring, negotiating, instructing, supervising, diverting, persuading, speaking-signaling, serving, and taking instructions-helping (Office of Administrative Law Judges Law Library, 1999). To help understand the coding system of the relationship to people in the DOT code, consider two examples associated with each category (shown in parentheses): attorney, doctor (mentoring); property manager, solicitor (negotiating); program director or sales supervisor (supervising); coach or faculty in college or university (instructing); comedian or tap dancer (diverting); broker or pharmaceutical detailer (persuading);

mathematician or economist (speaking-signaling); hair stylist or waiter (serving); clerk or boiler operator (taking instructions-helping). We include all categories except the taking instructions-helping category as indicators that interpersonal interaction is an important aspect of an occupation, and thus, social interactions with customers or colleagues are required following previous studies (Hamermesh and Biddle, 1994; Han et al., 2009). The proportion of occupations requiring social interactions in the early career is much higher for women (70%) than men (58%) (Table 1). Not shown in the tables, the level of late-teen BMI and obesity are similar in occupations requiring social interactions and those not requiring social interactions for both women and men. The respondents in occupations requiring social interactions earned higher wages ($13.9 for women and $17.3 for men) than people in occupations not requiring social interactions ($12.6 for women and $14.9 for men). The majority of all white collar occupation categories based on the Census occupational categories are classified as occupations requiring social interactions (approximately 70% of service, administrative support or clerical occupations, and management and professional services and 84% of sales), whereas only 44% of blue collar occupations are classified as occupations requiring social interactions in our final sample. Our final data also show that both women and men in occupations requiring social interactions are slightly taller, have higher level of education, higher AFQT scores, and higher level of self-esteem than their counterparts in occupations not requiring social interactions. The variable of primary interest is BMI, defined as selfreported weight in kilograms divided by self-reported height in meters squared. NLSY79 provides the respondents’ height information in 1981, 1982, and 1985. Because the respondents were between 20 and 28 years old in 1985, height in 1985 was used as the respondents’ adult height (Cawley, 2004). To define late-teen BMI, we took an average of a person’s BMI between the ages of 16 and 20 years old. The average late-teen BMI is 23.8 for women and 26.2 for men with a range of 16–58 (see Table 1). For current BMI, we averaged reported BMI between ages 30 and 34, which is averaged at 26.6 for women and 27.2 for men in our sample. We further classified the average lateteen BMI and current BMI into two clinical weight categories, which include obese (BMI  30) and non-obese (BMI < 30). In our sample, 8% of women and 18% of men are obese in their late teens whereas 24% and 22% of women and men, respectively, are obese in their early thirties. Other covariates include age, race, education level (years of schooling), marital status (married versus nonmarried), the number of children, the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, the duration between the late-teen BMI measurement and the early adult wage measurement, the average age at which the late-teen BMI is measured (only when late-teen BMI is used), self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

training between late teen and early thirties, and regional variables. Approximately 60% of the sample was nonHispanic White, and a quarter was non-Hispanic Black for both genders. More than one-half of the sample (60% of women and 61% of men) was married at the time of interview. The majority lived in urban areas. The regional variables include the following variables: urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), and unemployment rate in the residential area (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference). The unemployment rate in the residential area is at the metropolitan area level for the respondents residing in a metropolitan area. Otherwise, the unemployment rate is the state unemployment rate (not including Census metropolitan areas) in which the respondent resides (U.S. Department of Labor, 2004). We also control for contextual macroeconomic

387

conditions by including the following state-level covariates: total number of private businesses per 10,000 capita, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. The state-level variables were obtained from the Census County and City Data Books by the U.S. Census Bureau (U.S. Census Bureau and U.S. Department of Commerce, 2000) (see Table 1). 5. Results We present regression results of hourly wages in the early thirties on current BMI and other controls including education and occupation outcome (Table 2). In Models 1– 4, we present regression results of hourly wages in the early thirties on current BMI with gradually adding the two indirect pathways. A one-unit increase of BMI is associated with a decrease in hourly wages of 0.78% for women (Model 1). However, the extent of the direct BMI wage

Table 2 OLS estimation results of log hourly wages in the early career on BMI in the late-teenage years, the stock of education, and occupation outcome in early career. Sibling fixed effects

OLS

Women (N = 906) Current BMI

Model 1

Model 2

Model 3

Model 4

Model 1

Model 2

Model 3

Model 4

0.0078*** (0.0029)

0.0067** (0.0029) 0.0472*** (0.0122)

0.0053* (0.0027)

0.0051* (0.0027) 0.0198 (0.0130) 0.1530*** (0.0535) 0.4051*** (0.0710) 0.1487* (0.0807) 0.0899* (0.0519) 0.0532 (0.0429)

0.0163 (0.0115)

0.0165 (0.0115) 0.0259 (0.0344)

0.0184* (0.0108)

0.0183* (0.0108) 0.0129 (0.0347) 0.1662 (0.1433) 0.4652*** (0.1571) 0.0211 (0.1759) 0.1327 (0.1519) 0.0276 (0.1047)

0.0038 (0.0037) 0.0337*** (0.0116) 0.1233** (0.0530) 0.1958*** (0.0543) 0.0590 (0.0756) 0.0606 (0.0635) 0.0676* (0.0381)

0.0030 (0.0095)

Highest grade completed

0.1810*** (0.0499) 0.4079*** (0.0708) 0.1469* (0.0811) 0.0944* (0.0525) 0.0556 (0.0432)

Managerial or professional specialties Service Sales Administrative support/clerical Occupation requiring social interaction Men (N = 1068) Current BMI Highest grade completed Managerial or professional specialties Service Sales Administrative support/clerical Occupation requiring social interaction

0.0020 (0.0037)

0.0039 (0.0037) 0.0462*** (0.0107)

0.0027 (0.0037)

0.1763*** (0.0480) 0.1930*** (0.0555) 0.0784 (0.0744) 0.0368 (0.0602) 0.0689* (0.0380)

0.1549 (0.1394) 0.4610*** (0.1560) 0.0140 (0.1741) 0.1293 (0.1510) 0.0254 (0.1041) 0.0041 (0.0097) 0.0165 (0.0254)

0.0019 (0.0097)

0.0512 (0.1047) 0.0854 (0.1462) 0.0053 (0.1514) 0.0809 (0.1559) 0.0511 (0.0807)

0.0032 (0.0098) 0.0264 (0.0280) 0.0926 (0.1136) 0.1154 (0.1496) 0.0246 (0.1528) 0.0936 (0.1566) 0.0544 (0.0808)

Note: All models control for the following covariates: age, race, marital status (married versus non-married), the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job training between late teen and early thirties, and regional variables, which include urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), unemployment rate in the residential unit (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference), total number of private businesses per 10,000 capita at the state level, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. Standard errors for the marginal effects are in the parentheses. Unit of observation is person. * p < 0.1. ** p < 0.05. *** p < 0.01.

388

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

penalty for women decreases to 0.67 and 0.53%, respectively, after controlling for the stock of education (Model 2) and occupation outcome (Model 3). Finally, the magnitude of the direct BMI wage penalty for women falls to 0.51% when we control for both the stock of education and occupation outcome (Model 4). These effects for women are statistically significant. Indeed, there is a relatively large association between wages and the stock of education and occupation outcomes for both genders. Having one additional year of schooling is associated with an increase in hourly wages by 2.0 and 3.4% for women and men, respectively, although the results are statistically significant only for men. We also find that women and men, respectively, earn 15.3 and 12.3% higher hourly wages in managerial or professional specialty occupations compared to blue-collar occupations (the reference), whereas service occupations are associated with 40.1 and 19.6% lower hourly wages for women and men, respectively.

Controlling for sibling fixed effects greatly increases the standard errors by a factor of about three to four. Point estimates in the fixed effects models for women are larger in absolute value than in the OLS models and similarly the results are statistically significant only at the 10% level. As with the OLS models, fixed effects models for men find no relationship between current weight status and log wages (see Table 2). We also show regression results from wage equations using clinical weight classifications instead of a linear measure of BMI in Table 3. For women, we show that obesity is associated with 12.5% lower hourly wages in young adulthood when education and occupation outcome are not controlled for (Model 1). Similar to the linear measure of BMI, the magnitude of the direct obesity wage penalty decreases to 11.1 and 8.9% if we control for education (Model 2) or occupation outcome (Model 3), respectively. Controlling for both education and occupation outcomes (Model 4), the direct obesity wage penalty is

Table 3 OLS estimation results of log hourly wages in the early career on obesity in the late-teenage years, the stock of education, and occupation outcome in early career. OLS

Women (N = 906) Currently obese

Sibling fixed effects

Model 1

Model 2

Model 3

Model 4

Model 1

Model 2

Model 3

Model 4

0.1246*** (0.0426)

0.1110*** (0.0426) 0.0474*** (0.0122)

0.0888** (0.0401)

0.0859** (0.0402) 0.0199 (0.0130) 0.1533*** (0.0536) 0.4042*** (0.0708) 0.1521* (0.0807) 0.0893* (0.0517) 0.0524 (0.0427)

0.1472 (0.1623)

0.1444 (0.1627) 0.0240 (0.0346)

0.1647 (0.1524)

0.1673 (0.1531) 0.0156 (0.0350) 0.1821 (0.1443) 0.4457*** (0.1579) 0.0035 (0.1772) 0.1223 (0.1529) 0.0465 (0.1048)

0.0067 (0.0383) 0.0329*** (0.0116) 0.1228** (0.0531) 0.1968*** (0.0547) 0.0587 (0.0755) 0.0623 (0.0631) 0.0672* (0.0381)

0.1016 (0.1012)

Highest grade completed

0.1816*** (0.0500) 0.4070*** (0.0705) 0.1506* (0.0811) 0.0940* (0.0522) 0.0548 (0.0430)

Managerial or professional specialties Service Sales Administrative support/clerical Occupation requiring social interaction Men (N = 1068) Currently obese Highest grade completed Managerial or professional specialties Service Sales Administrative support/clerical Occupation requiring social interaction

0.0126 (0.0384)

0.0022 (0.0391) 0.0452*** (0.0107)

0.0015 (0.0378)

0.1747*** (0.0481) 0.1941*** (0.0557) 0.0775 (0.0743) 0.0381 (0.0598) 0.0686* (0.0380)

0.1685 (0.1405) 0.4405*** (0.1569) 0.0050 (0.1755) 0.1179 (0.1520) 0.0440 (0.1042) 0.0970 (0.1019) 0.0124 (0.0250)

0.1059 (0.1028)

0.0574 (0.1038) 0.0960 (0.1448) 0.0192 (0.1505) 0.0882 (0.1552) 0.0553 (0.0801)

0.0989 (0.1033) 0.0229 (0.0277) 0.0944 (0.1132) 0.1236 (0.1488) 0.0370 (0.1522) 0.1000 (0.1560) 0.0588 (0.0803)

Note: All models control for the following covariates: age, race, marital status (married versus non-married), the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job training between late teen and early thirties, and regional variables, which include urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), unemployment rate in the residential unit (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference), total number of private businesses per 10,000 capita at the state level, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. Standard errors for the marginal effects are in the parentheses. Unit of observation is person. * p < 0.1. ** p < 0.05. *** p < 0.01.

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

389

Table 4 The marginal effect of the BMI/obesity at the late-teenage years on the years of schooling completed in early career in OLS. Dependent variable: years of schooling completed

Women (N = 906) Model 1

Men (N = 1068) Model 2

Model 1

0.0232* (0.0137)

Late-teen BMI

Model 2

0.0235* (0.0140) 0.3576** (0.1180)

Obese in late teens

0.1408 (0.1606)

Note: All models control for the following covariates: age, race, marital status (married versus non-married), the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job training between late teen and early thirties, and regional variables, which include urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), unemployment rate in the residential unit (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference), total number of private businesses per 10,000 capita at the state level, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. Standard errors for the marginal effects are in the parentheses. Unit of observation is person. * p < 0.1. ** p < 0.05.

associated with 8.6% lower wages. However, such statistically significant associations of the likelihood of obesity with wages lose statistical significance when we control for sibling fixed effects. No significant direct obesity wage penalty is found for men regardless of model specifications (see Table 3). Next in Table 4, we show that higher late-teen BMI is associated with statistically significant fewer years of schooling acquired by the early thirties for both genders. A one-unit increase in late-teen BMI decreases the highest

grade completed in the early thirties by 0.023 and 0.024 units for women and men, respectively. Re-estimation of years of schooling with clinical weight classification shows that late-teen obesity is associated with less years of schooling by 0.4 units for women. For young adult men, late-teen obesity is not statistically significantly associated with years of schooling (see Table 4). As for occupation in the early thirties, the results in Table 5 show that the effect of late-teen body weight status does not have a statistically significant effect on the

Table 5 The marginal effect of BMI/obesity at the late-teenage years on occupation outcome based on the Census occupational classification in early career in a multinomial logit model. Dependent variable: occupation outcome

Service

Model 1 Women (N = 906) Late-teen BMI

0.0033 (0.0031)

Obese in late teens Education Men (N = 1608) Late-teen BMI

0.0196** (0.0086)

0.0089** (0.0052)

Sales

Model 1

Model 1

Model 2

0.0640 (0.0492) 0.0191** (0.0072)

0.0782*** (0.0079)

0.0056 (0.0726) 0.0784*** (0.0073)

0.0031 (0.0032) 0.0151 (0.0294) 0.0088 (0.0072)

0.0638*** (0.0073)

Model 2

0.0033 (0.0026)

0.0018 (0.0043)

0.0016 (0.0029)

Obese in late teens Education

Model 2

Managerial/ professional specialty

0.0051 (0.0058)

-0.0003 (0.0048)

Blue-collar

Model 1

Model 1

Model 2

0.0031 (0.0522) 0.0054 (0.0066)

0.0396*** (0.0088)

0.0036 (0.0577) 0.0397*** (0.0073)

0.0063 (0.0051)

0.0141* (0.0088)

0.0578 (0.0430) 0.0142** (0.0075)

0.0096** (0.0041)

0.0019 (0.0019) 0.0249 (0.0231) 0.0001 (0.0058)

Model 2

0.0062** (0.0021)

0.0014 (0.0038)

0.0029 (0.0024) 0.0470 (0.0355) 0.0637*** (0.0078)

Administrative support/clerical

0.0304 (0.0201) 0.0064 (0.0050)

0.0609*** (0.0097)

0.1174** (0.0465) 0.0611*** (0.0097)

Note: All models control for the following covariates: age, race, marital status (married versus non-married), the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job training between late teen and early thirties, and regional variables, which include urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), unemployment rate in the residential unit (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference), total number of private businesses per 10,000 capita at the state level, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. Standard errors for the marginal effects are in the parentheses. Unit of observation is person. * p < 0.1. ** p < 0.05. *** p < 0.01.

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

390

Table 6 The marginal effect of BMI/obesity at the late-teenage years on occupation outcome in the early career by the requirement of social interactions with customers or colleagues in a logit model. Dependent variable: employed at occupations requiring social interactions

Women (N = 906) Model 1

Late-teen BMI

Model 2

Model 1

*

Model 2

0.0055 (0.0044)

0.0012 (0.0007) **

Obese in late teens Education

Men (N = 1068)

0.0473 (0.0185) 0.0128 (0.0092)

0.0126 (0.0097)

0.0123* (0.007)

0.0456 (0.0542) 0.0127 (0.0096)

Note: All models control for the following covariates: age, race, marital status (married versus non-married), the elapsed time from the latest pregnancy to the time of interview (indicators for being pregnant within 2 years, 4 years, and 6 years from the time of interview with not being pregnant within 6 years from the time of interview as the reference), highest grade completed by parents, height in meters, AFQT scores, self-esteem (strongly agree or agree on a question of whether having a positive attitude), years of employment, whether participated in any on-the-job training between late teen and early thirties, and regional variables, which include urban/rural status of the respondents’ residential area, four regional areas in the US (South, Midwest, West, and Northeast as the reference), unemployment rate in the residential unit (larger than 12%, between 9 and 12%, between 6 and 9% with less than 6% as the reference), total number of private businesses per 10,000 capita at the state level, per capita average income in $1000 deflated by yearly GDP by state, and the Consumer Price Index. Standard errors for the marginal effects are in the parentheses. Unit of observation is person. * p < 0.1. ** p < 0.05.

occupation outcome based on the Census occupational classification in the early thirties overall. We find that a one-unit increase in late-teen BMI reduces the likelihood of being in a blue-collar occupation by 0.62 percentage points for women, whereas the association is positive to the similar extent (11.7 percentage points) for men. The results also suggest that the lasting wage penalty in the early career for late-teen body weight status operates through education. A one year increase in the years of schooling completed raises the likelihood of having managerial or professional specialty occupations by 7.8 and 6.4 percentage points for women and men, respectively. An additional year of education completed reduces the likelihood of being in service occupations (by 2.0 and 0.9 percentage points for women and men, respectively),

administrative support or clerical occupations (by 4.0 percentage points, women only), and blue collar occupations (by 1.4 and 6.1 percentage points for women and men, respectively) (see Table 5). Late-teen BMI has a negative effect on the probability of having occupations requiring social interactions for women to the extent that a one-unit increase of late-teen BMI reduces the probability of being in occupations requiring social interactions by 0.12 percentage points. Estimating the model with clinical weight classification also shows that obese women in their late-teenage years are 4.7 percentage points less likely to have occupations requiring social interactions compared to their non-obese counterparts. No statistically significant association was found for men (see Table 6).

Table 7 Direct and Indirect weight status wage penalty, for BMI and obesity, by gender. Women (N = 906)

Indirect pathways Education Occupation outcome: Census occupation categories Occupational characteristics of requiring social interactions with customers or colleagues Education through occupation outcome measured as the Census occupation categories Education through occupational characteristics of requiring social interactions with customers or colleagues Total indirect wage penalty Total direct wage penalty Standard errors, calculated from bootstrapping, are in the parentheses. * p < 0.1. ** p < 0.05.

Men (N = 1068)

Model 1

Model 2

Model 1

Model 2

(BMI)

(Obese)

(BMI)

(Obese)

0.000300 (0.00186) 0.000633 (0.00256) 0.000033 (0.00123) 0.000689 (0.00070) 0.000008 (0.00008) 0.00105 (0.00312) 0.0183* (0.0108)

0.00559* (0.00332) 0.0291 (0.0468) 0.00220 (0.0155) 0.00990* (0.00509) 0.000213 (0.00139) 0.0354* (0.0187) 0.1673 (0.1531)

0.000621 (0.00116) 0.000249 (0.00189) 0.000301 (0.00103) 0.000149 (0.00041) 0.000016 (0.00008) 0.00009 (0.00247) 0.0032 (0.0098)

0.00559 (0.0372) 0.0291* (0.0168) 0.00220 (0.0155) 0.00990 (0.0118) 0.000213 (0.00139) 0.0354** (0.0069) 0.0989 (0.1033)

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

Finally, we show the extent of the direct and indirect wage penalty associated with BMI and obesity in Table 7. For both women and men, we do not find a statistically significant indirect wage penalty when the body weight status is measured by linear BMI in late-teenage years, although we find statistically significant 1.83% direct BMI wage penalty for women. Estimations using clinical weight classification reveal that an indirect BMI wage penalty through education and occupation outcomes occurs to a larger extent at the upper tail of the BMI distribution for both men and women. Late-teen obesity is indirectly associated with 3.5% lower wages for both women and men. The indirect obesity wage penalty among women mainly comes from education (0.56% less hourly wages) and occupation outcomes based on the Census occupational classification operating through education (0.99% less hourly wages), whereas for men it comes mainly through Census occupation classification (2.9% less hourly wages) (see Table 7). 6. Conclusions Our paper is motivated by the idea that an obesity wage penalty may operate through choices earlier in life, namely through education and occupation choice. If correct, then the total wage penalty for high weight status could be larger than has been measured in previous studies. We extend the existing literature by showing that in addition to the potential direct wage penalty in a given job with regard to body weight, additional penalties may stem from the indirect effect of late-teen body weight, particularly, at the upper tail of the BMI distribution, on outcomes related to both education and occupation in one’s early career. The results from our education and occupation regressions show that higher late-teen BMI (women and men) and late-teen obesity (women only) is related to lower levels of accumulated education in the early thirties. Lateteen body weight status is associated with occupation outcomes based on Census occupational classification in the early thirties and indirectly through investments in education. Higher late-teen BMI or obesity (women only) leads to occupations that do not require social interaction with customers or colleagues. These results show that it is important to examine these potential long-run effects that may result in lower wages over time. Assessing the extent of these indirect effects on wages, we do not find a statistically significant indirect wage penalty measured by BMI. However, late-teen obesity is indirectly associated with 3.5% lower wages for both women and men. This indirect obesity wage penalty mainly comes from education and occupation through education for women and from occupation classification for men. We show that previous studies that estimate the BMI wage penalty conditional on education and occupation may underestimate the penalty for both genders. For men, although no statistically significant association of current weight status with wages is found as in the previous literature (Cawley, 2004; Han et al., 2009; Averett and Korenman, 1996; Norton and Han, 2008), our study finds an indirect obesity wage penalty of 3.5% lower wages via occupation. For women, our results show 3.5% lower wages

391

resulting from late-teen obesity via the indirect pathways of education and occupation. These study results suggest that in order to fully capture the relationship between body weight and labor market outcomes researchers need to account for the early part of the life-cycle when individuals make important decisions related to human capital formation. These results highlight the existence of persistent negative externalities from youth obesity prevalence which support the importance of interventions to reduce childhood obesity.

Acknowledgement Han and Powell gratefully acknowledge research support from the Robert Wood Johnson Foundation through the Bridging the Gap program for the ImpacTeen project. References Averett, S., Korenman, S., 1996. The economic reality of the beauty myth. The Journal of Human Resources 31 (2), 304–330. Baum, C.L., Ford, W.F., 2004. The wage effects of obesity: a longitudinal study. Health Economics 13 (9), 885–899. Bhattacharya, J., Bundorf, M.K., 2009. The incidence of the healthcare costs of obesity. Journal of Health Economics 28 (3), 649–658. Blau, P.M., Gustad, J.W., Jessor, R., Parnes, H.S., Wilcock, R.C., 1956. Occupational choice: a conceptual framework. Industrial and Labor Relations Review 9 (4), 531–543. Borghans, L., Duckworth, A., Heckman, J.J., ter weel, B., 2008. The economics and psychology of personality traits, NBER Working Papers, Cambridge, Massachusetts. W13810. Borghans, L., ter weel, B., Weinberg, B.A., 2007. Interpersonal styles and labor market outcomes, NBER Working Papers, Cambridge, Massachusetts. W12846. Bourdieu, P., Passeron, J.C., 1977. Reproduction in Education Society and Culture. Sage, Beverley Hills, CA. Bradley, S., 1991. An empirical analysis of occupational expectations. Applied Economics 23, 1159–1174. Bynner, J., 1998. Education and family components of identity in the transition from school to work. International Journal of Behavioral Development 22 (1), 29–53. Carpenter, C.S., 2006. The effects of employment protection for obese people. Industrial Relations 45 (3), 393–415. Carr, D., Friedman, M.A., 2005. Is obesity stigmatizing? Body weight, perceived discrimination, and psychological well-being in the United States. Journal of Health and Social Behavior 46, 244–259. Cawley, J., 2000. An instrumental variables approach to measuring the effect of body weight on employment disability. Health Services Research 35 (5), 1159–1179. Cawley, J., 2004. The impact of obesity on wages. Journal of Human Resources 39 (2), 451–474. Conley, D., Glauber, R., 2005. Gender, body mass and economic status. National Bureau of Economic Research Working Paper Series No. 11343. Crosnoe, R., Muller, C., 2004. Body mass index, academic achievement, and school context: examining the educational experience of adolescents at risk of obesity. Journal of Health and Social Behavior 45 (4), 393–407. Daniels, J., 2005. Weight and weight concerns: are they associated with reported depressive symptoms in adolescents? Journal of Pediatric Health Care 19 (1), 33–41. Everett, M., 1990. Let an overweight person call on your best customers? Fat chance. Sales and Marketing Management 142, 66–70. Flegal, K.M., Carroll, M.D., Ogden, C.L., Curtin, L.R., 2010. Prevalence and trends in obesity among us adults, 1999–2008. JAMA: The Journal of the American Medical Association 303 (235), 241. Frieze, I.H., Olson, J.E., Good, D.C., 1990. Perceived and actual discrimination in the salaries of male and female managers. Journal of Applied Social Psychology 20 (1), 46–67. Garcia, J., Quintana-Domeque, C., 2007. Obesity, employment, and wages in Europe. In: Bolin, K., Cawley, J. (Eds.), The Economics of Obesity.

392

E. Han et al. / Economics and Human Biology 9 (2011) 381–392

Advances in Health Economics and Health Services Research, vol. 17. pp. 187–218. Gortmaker, S.L., Must, A., Perrin, J.M., Sobol, A.M., Dietz, W.H., 1993. Social and economic consequences of overweight in adolescence and young adulthood. The New England Journal of Medicine 329 (14), 1008–1012. Ham, R., Junankar, P.N., Wells, R., 2009. Occupational Choice: Personality Matters. IZA DP No. 4105. Hamermesh, D.S., Biddle, J.E., 1994. Beauty and the labor market. The American Economic Review 84 (5), 1174–1194. Han, E., Norton, E.C., Stearns, S.C., 2009. Weight and wages: fat versus lean paychecks. Health Economics 18 (5), 535–548. Harper, B., 2000. Beauty, structure and the labour market: a British cohort study. Oxford Bulletin of Economics and Statistics 62, 771–800. Heckman, J.J., Stixrud, J., Urzua, S., 2006. The effects of cognitive and noncognitive abilities on labor market outcomes and social behavior. Journal of Labor Economics 24 (3), 411–482. Holland, J.L., 1985. Making Vocational Choices, 2nd ed. Prentice Hall, Englewoods Cliffs, NJ. Kaestner, R., Grossman, M., Yarnoff, B., 2009. Effects of weight on adolescent educational attainment. National Bureau of Economic Research Working Paper Series No. 14994. Krueger, A.B., Schkade, D., 2005. Sorting in the labor market: do gregarious workers flock to interactive jobs? Working Paper, Princeton, NJ. Lundborg, P., et al., 2007. Obesity and occupational attainment among the 50+. In: Bolin, K., Cawley, J. (Eds.). The Economics of Obesity, Advances in Health Economics and Health Services Research. Martel, L.F., Biller, H.B., 1987. Statues and Stigma: The Biopsychosocial Development of Short Males. Lexington Books, Lexignton, MA. Mincer, J., 1974. Schooling, Experience, and Earnings. NBER, Columbia Press. Moan N.H., Tekin E., 2009. Obesity, Self-esteem and Wages. NBER Working Papers 15101. Mobius, M.M., Rosenblat, T.S., 2006. Why beauty matters. American Economic Review 96 (1), 222–235. Morris, S., 2007. The impact of obesity on employment. Labour Economics 14, 413–433.

Mueller, G., Plug, E., 2006. Estimating the effect of personality on male– female earnings. Industrial and Labor Relations Review 60 (1), 3–22. Norton, E.C., Han, E., 2008. Using genetic information to identify causal effects of obesity. Health Economics 17, 1089–1104. Office of Administrative Law Judges Law Library, 1999. U.S. Department of Labor. Dictionary of Ocupational Titles. Pagan, J.A., Davila, A., 1997. Obesity, occupational attainment, and earnings. Social Science Quarterly (University of Texas Press) 78 (3), 756–770. Paraponaris, A., Saliba, B., Ventelou, B., 2005. Obesity, weight status and employability: empirical evidence from a French national survey. Economics & Human Biology 3 (2), 241–258. Persico, N., Postlewaite, A., Silverman, D., 2004. The effect of adolescent experience on labor market outcomes: the case of height. Journal of Political Economy 112 (5), 1019–1053. Puhl, R., Brownell, K.D., 2001. Bias, discrimination, and obesity. Obesity Research 9, 788–805. Ruhm, C.J., 2000. Are recessions good for your health? Quarterly Journal of Economics 115 (2), 617–650. Sabia, J.J., 2007. The effect of body weight on adolescent academic performance. Southern Economic Journal 73 (4), 871–900. Sarlio-Lahteenkorva, S., Lahelma, E., 1999. The association of body mass index with social and economic disadvantage in women and men. International Journal of Epidemiology 28 (3), 445–449. Schoon, I., Parsons, S., 2002. Teenage aspirations for future careers and occupational outcomes. Journal of Vocational Behavior 60, 262–288. Schoon, I., 2001. Teenage job aspirations and career attainment in adulthood: a 17-year follow-up study of teenagers who aspired to become scientists, health professionals, or engineers. International Journal of Behavioral Development 25 (2), 124–132. Tunceli, K., Li, K., Williams, L.K., 2006. Long-term effects of obesity on employment and work limitations among U.S. adults, 1986 to 1999. Obesity 14 (9), 1637–1646. U.S. Census Bureau, 1996. Current Population Survey. U.S. Census Bureau, U.S. Department of Commerce, 2000. County and City Data Book. U.S. Department of Labor, 2004. NLSY79 Users’ Guide.