THE IMPACT OF BODY WEIGHT ON OCCUPATIONAL MOBILITY AND CAREER DEVELOPMENT

• Published date: 01 May 2019
• DOI: http://doi.org/10.1111/iere.12364
• Date: 01 May 2019

INTERNATIONAL ECONOMIC REVIEW

Vol. 60, No. 2, May 2019 DOI: 10.1111/iere.12364

THE IMPACT OF BODY WEIGHT ON OCCUPATIONAL MOBILITY AND CAREER

DEVELOPMENT∗

BYMATTHEW C. HARRIS1

University of Tennessee, U.S.A.

This article examines the relationship between individuals’ weight and employment decisions over the life

cycle. I estimate a dynamic stochastic model of individuals’ annual choices of occupation, hours worked, and

schooling. Evidence suggests that heavier individuals face higher switching costs when transitioning into white-

collar occupations, earn lower returns to experience in white-collar occupations, and earn lower wages in socially

intensive jobs. I simulate a hypothetical antidiscrimination policy treating obese workers as a protected class.

Although such a policy would reduce gaps in occupational attainment, it would have little effect on the observed

divergence in wages between obese and nonobese workers.

1. INTRODUCTION

How does body weight affect employment behavior and wages over the life cycle? We know

obesity yields high costs in the workplace.2Estimates place annual workplace productivity costs

of obesity between $12 and $30 billion. Although obese workers miss 15%–50% more work

time than healthy weight workers, two-thirds of these productivity costs are due to decreased

at-work performance. Reduced productivity not only affects contemporaneous wages and em-

ployment decisions, but also decreases subsequent pay increases and employment opportunities

(Holmstrom, 1999). Body weight today affects the expected present discounted value of em-

ployment decisions not only by affecting contemporaneous wages and utility but also expected

future wages and labor market opportunities.

The workplace costs of high body weight are inherently dynamic and vary by occupation.

Studies have shown that obesity leads to difficulty managing professional interpersonal rela-

tionships and reduces stamina when performing physical tasks.3Although lower productivity

affects wages, difficulties with certain job requirements may yield additional nonmonetary costs

that also influence occupational choices. An individual’s body weight may also provide a signal

about that individual’s self-discipline or work ethic, the value of which may differ between

occupations. Such a negative signal would lead to decreased occupational mobility for heavier

individuals. Differences between occupations in the expected costs of high body weight pro-

vide additional motivation for modeling these costs as a part of forward-looking individuals’

employment decisions. When an individual chooses an occupation, he accrues human capital

that is not perfectly transferable to other occupations (Kambourov and Manovskii, 2009). Thus,

∗Manuscript received January 2015; revised May 2018.

1I would like to thank Donna Gilleskie, Brian McManus, David Guilkey, Helen Tauchen, Clement Joubert, Steven

Stern, Maarten Lindeboom, Ahmed Khwaja, William Neilson, Carl Sanders, Chris Cronin, Michael Darden, participants

in the UNC Applied Microeconomics Workshop, Triangle Health Economics Workshop, Annual Health Economics

Conference, and the 6th Biennial Conference of the American Society of Health Economists. I also thank my editor,

Hanming Fang and two anonymous referees for their invaluable comments and stewardship. All errors are mine. Please

address correspondence to: Matthew C. Harris, Department of Economics and Center for Business and Economic

Research, 722 Stokely Management Center, 916 Volunteer Boulevard, University of Tennessee, Knoxville, TN 37996.

Phone: 865-974-5591. E-mail: mharris@utk.edu.

2See, for example, Ricci and Chee (2005) and Andryeva (2014).

3See Pronk et al. (2004), Johar and Katayama (2012), Hamermesh and Biddle (1994), DeBeaumont (2009), and Han

et al. (2009).

631

C

(2018) by the Economics Department of the University of Pennsylvania and the Osaka University Institute of Social

and Economic Research Association

632 HARRIS

contemporaneous occupational choice affects both expected future wages and future occupa-

tional decisions. Finally, an individual’s body weight is itself dynamic and maybe affected by

one’s choice of occupation and hours.

Despite the inherent dynamic relationship between body weight and employment outcomes,

the existing literature on the subject has largely relied on static approaches and abstracted from

either occupational choice or wages. I formulate and estimate a dynamic discrete choice model

where body weight affects both the distribution of wage offers and nonmonetary costs of each

employment alternative, and employment decisions subsequently affect weight. Although the

model therefore treats body weight as endogenous, it does so in a limited sense. Individuals

do not choose body weight per se, but their body weight evolves as a consequence of the

individuals’ history of employment decisions.4Both the model and empirics follow in the

occupational choice tradition of Keane and Wolpin (1997), Lee (2005), Lee and Wolpin (2006),

Kambourov and Manovskii (2009), Sullivan (2010), and Baird (2014). I construct indices of

the intensity of mental, physical, and social job requirements for each occupation to examine

how job requirements affect the wage and nonmonetary costs of body weight. I estimate the

parameters governing the individual’s decision-making process using data from the National

Longitudinal Survey of Youth, 1979 cohort.

This is the first article to examine differences in earnings and occupational attainment on the

basis of body weight in a dynamic discrete occupational choice framework. Dynamic stochastic

occupational choice models have been utilized to examine gender gaps in wages and occupa-

tional attainment (e.g., Altug and Miller, 1998; Flabbi, 2010; Eckstein and Lifshitz, 2011; Gayle

and Golan, 2012; Yamaguchi, 2013) and similar black–white gaps (e.g., Keane and Wolpin,

2000; Bowlus and Eckstein, 2002; Lehmann, 2013). However, much of the existing literature on

the relationship between body weight and employment outcomes has used static methods and

examined contemporaneous cross-occupation “wage penalties” for body weight (e.g., Pagan

and Davila, 1997; Cawley, 2004; Han et al., 2009; Johar and Katayama, 2012).5When occupa-

tion has been considered, it is usually as a control variable or mediating factor in the associative

relationship between weight and mean wages. To the extent that body weight increases fric-

tion when transitioning into certain occupations, treating occupation as a control excludes an

important mechanism by which body weight can affect wages.

This article also contributes to a growing literature where job requirements are incorporated

into dynamic models as a determinant of occupational choice (Sanders, 2010; Yamaguchi, 2012)

and the literature on how one’s employment behavior affects one’s health (e.g., King et al.,

2001; Lakdawalla and Philipson, 2002; Courtemanche, 2009; Kelly et al., 2014, Ravesteijn et al.,

2014). Additionally, this article incorporates prospective wage differentials into a single-agent

occupational choice framework. As Coate and Loury (1993) show, anticipated wage differentials

can affect the formation of human capital, which affects subsequent wages. Theoretically, the

model seeks to merge Mincer (1958), Ben-Porath (1967), and Becker (1957). The structure of

this model closely resembles Keane and Wolpin (1997) and Sullivan (2010), focusing on how

current and expected future monetary and nonmonetary costs affect individuals’ decisions over

the life cycle. There is also a small methodological contribution to the literature on dynamic

models of occupational choice regarding the distribution of unobserved wages. I estimate the

full distribution of wages inside the model using conditional density estimation (CDE; Gilleskie

and Mroz, 2004) instead of imposing a parametric distribution on wages. Additionally, CDE

allows the marginal effect of explanatory variables to vary over the support of the dependent

variable. This feature is useful given that Kline and Tobias (2008) and Johar and Katayama

4The purpose of this article is not to investigate the effects of employment decisions on weight, but rather the

opposite. The model permits employment decisions to affect body weight, but through a feedback mechanism instead

of modeling change in body weight as a choice.

5Notable exceptions to the lack of dynamic modeling include Gilleskie et al. (2011) and Tosini (2008), but neither

study models occupational choice.