Cut out paper people representing discrimination against overweight people
VoxEU Column Gender Health Economics Labour Markets

The weight of discrimination: How obesity affects wages and why it matters

A large body of economics research has consistently found a negative association between obesity and wages, particularly among women. This column uses survey data on workers in the US and detailed info on their job characteristics to examine whether this is driven by statistical discrimination or prejudice and, if so, what are its roots. The authors find that obese women suffer a significant wage penalty, particularly those in public- and client-facing roles, while the evidence is much weaker for obese men. These findings point to obesity-based wage discrimination expressing biases and prejudices that affect women’s labour market outcomes more than those of men.

Obesity has evolved into a global pandemic, raising concerns about its impact on individual health and societal costs (Sawers and Dobbs 2014). The association between obesity and a range of health problems is well-established in the medical literature. Equally well-recognised is the strain it places on health insurance and social security systems literature (Cawley and Meyerhoefer 2012). The implications are clear: obesity matters, and it matters a lot. But what about the connection between obesity and economic wellbeing? In particular, how does obesity influence wages, and why is it important to understand this relationship? There is a large body of research in economics which consistently finds a negative association between obesity and (hourly) wages, particularly among women (e.g. Tekin 2007, Wada and Tekin 2010, Cawley 2015) which is not explained by (observable) productivity differences among workers. Nonetheless, there is no well-established consensus on either the specific channels underlying this negative correlation or the roots of potential labour-market discrimination against obese people.

Uncovering the hidden factors behind obesity-based wage discrimination

To uncover the hidden factors behind the existence of (unexplained) wage penalties due to obesity, in a recent paper (Dolado et al. 2023) we use very rich longitudinal information from ten waves (2001-11) of the 1997 National Longitudinal Survey of Youth (NLSY97) on young (17-31) white male and female workers in the US. The panel has a total of around 9,600 person-year observations for men and 8,800 for women. Our objective is to analyse whether the above-mentioned negative correlation between obesity and wages could be attributed to statistical and/or taste-based workplace discrimination (i.e. prejudice), once observable productivity differences are used as covariates. Further, by classifying different types of jobs according to their specific characteristics (such as the extent of oral communication involved or the importance of dealing with customers and employers), we seek to identify which of these traits are more consistent with potential discrimination being exerted by either co-workers, employers, customers, or agents outside the company. The latter information, drawn from the O*Net Online database and mapped to the Census Occupation Codes 2002, facilitates achieving these goals. Specifically, to identify sources of potential obesity wage discrimination due to prejudice, we consider a wider set of occupational characteristics than in previous closely related studies on this topic (e.g. Baum and Ford 2004, DeBeaumont 2009, Han et al. 2009) which inform about direct contact with clients, employers, or other economic agents outside the company. Some of these detailed job characteristics are useful to rationalise some previously unexplained results in this literature.

In the empirical analysis, we control for a wide range of observable productivity-related characteristics and interpret the surviving (residual) impact of body fat percentage (BFP, defined as the ratio between body fat and body weight, both in kg) – which is a better measure of obesity than the popular BMI – on hourly wages as likely attributable to potentially discriminatory practices. In our NLSY97 sample, female respondents have a higher average BFP than men (0.33 vs. 0.24), while they receive a lower (real) hourly wage ($11.7 vs. $14.2) despite having a higher educational attainment. According to the World Health Organization (WHO), a BFP above 0.25 (0.33) defines obesity for men (resp. women) aged 20-39, while those within the range 0.21-0.25 (0.31-0.33) are borderline cases. Further, we tackle the potential endogeneity of obesity (i.e. the possibility that low wages cause obesity through poor diets) by instrumenting the respondent’s body weight measure with that of a close biological relative – the mother and the closest sibling – following, among others, Cawley (2004) and Brunello and D’Hombres (2007). Finally, by means of Oster’s (2019) methodology to test for the potential relevance of omitted unobserved components (unrelated to the observed controls) in biasing the estimates, we find that this problem is minor in our setting. 

Key findings

Baseline penalty

Our study yields intriguing results. While we find little evidence of either type of discrimination against obese men, the story is quite different for women. Obese women face a significant wage penalty. More precisely, we show that an increase of one standard deviation in BFP (around 0.067) is associated on average with a reduction of 2.1 log points in female wages, which can even reach 2.7 log points once some job traits are accounted for (see below).

The role of occupational characteristics

To pinpoint which occupational characteristics exacerbate or reduce the wage penalty experienced by obese workers, we interact them with BFP (see Figure 1). We find some small wage penalties for obese male workers in those occupations involving intense direct contact with the public, consumers, and external communication. In stark contrast, the results are much stronger for women: the penalty is highly statistically significant in occupations involving close direct contact with the public, clients, frequent oral communication, and where mistakes are punished by employers. Particularly noteworthy is the penalisation of obese women who have to speak in public and deal with clients. This evidence points to a specific niche of job environments where prejudice against obese women takes place, being more closely related to employers’ and customers’ negative reactions. Overall, we interpret this evidence as supporting that clients, employers and other internal agents of the firm are likely to be the main roots of taste-based discrimination against obese women.

Figure 1 Wage effects of interactions between BFP and occupational characteristics

Figure 1 Wage effects of interactions between BFP and occupational characteristics

Note:  This figure reports estimates as well as 90% confidence intervals obtained from RE-GLS estimation with lnW (logged hourly wage) as the dependent variable. Separate regressions are run for each interaction term by gender. All columns include demographic, human capital, health status, occupational characteristic controls and industry and year dummies for each survey observation.  

Assessing the relative weights of statistical versus taste-based discrimination

To assess the relative importance of statistical versus taste-based discrimination, we interact of BFP with three discrimination-indicator proxies which are informative about asymmetric information on workers’ true productivity, namely, those captured by: (i) experiencing a job change during the year before the interview (JobCh), (ii) age (Age), and (iii) work seniority (Tenure). In line with Altonji and Pierret’s  (2001) approach, the insight for including these interaction terms is as follows. One one hand, if statistical discrimination is present, the conjecture is that those individuals who recently changed jobs would have less time than stayers to prove their true productivity to their new employers. Thus, the interaction coefficient of BFP with JobCh should become negative in this case. Conversely, if the estimate on this interaction term turns out to be positive, it should be interpreted as discrimination based on prejudice. On the other hand, older individuals who have accumulated longer tenure are likely to have provided solid information to employers about their true productivity, and therefore less likely to experience statistical discrimination. Accordingly, a positive interaction coefficient of BFP with Age and Tenure should be found, reducing the negative effect of BFP on hourly wages. Otherwise, the right interpretation would be discrimination due to prejudice. To address the potential endogeneity problem of Tenure and JobChange, we use ‘leave-one-out’ instruments, which qualitatively yield similar results to those obtained in the non-instrumented specification.

Our findings are as follows: while the estimates of the coefficients on the double interactions between BFP and both Age and Tenure are small and insignificant for men, they happen to be negative and significant for women, suggesting that the older the woman and the longer her job tenure, the greater the wage penalty for being obese. Thus, following the previous reasoning, this evidence goes against statistical discrimination while it is consistent with taste-based discrimination.

Relying of this evidence, our last exercise is to estimate Mincerian wage regressions which, besides levels and double interactions, now include triple interactions of BFP with the above-mentioned discrimination indicators and those occupational traits that turned to be more relevant in Figure 1. Our main finding is that, while most of these terms remain statistically insignificant for obese men, the (unexplained) wage penalty for obese women is particularly acute in jobs that involved direct contact with the public and consumers. For instance, evaluating tenure and the importance of dealing with customers at their female sample means, implies that an increase of one standard deviation in BFP is associated with a 2.7 log points reduction in hourly wages where the contribution of the triple interaction is about 30% of the total effect. Thus, it seems likely that the existence of gender-specific expectations on how physical appearance matters for men and women could explain gender differences in stereotypes. We argue that a potential reason for these differences could be that men are overrepresented in managerial positions (60% according to the US Bureau of Labor Statistics) and tend to discriminate more against obese workers of the opposite gender regarding image concerns. Likewise, male clients seem to have more prejudice against female attendants’ physical appearance than against male ones.

Policy implications and conclusions

The previous results offer valuable insights into the complex interplay between obesity and wage discrimination. First and foremost, we find that obesity-based wage discrimination is a tangible and substantial issue, particularly for women. It is not a matter of statistical disparities based on productivity differences, but rather an expression of deep-seated biases and prejudices that affect labour market outcomes. This finding underscores the urgency of addressing societal perceptions and prejudices regarding body weight in workplaces.

Furthermore, our research highlights that the impact of obesity on wages is not uniform across genders and job roles. Women, especially those engaged in professions with high levels of public interaction, are particularly vulnerable to wage penalties. These findings call for a targeted and nuanced approach to fighting obesity-based wage discrimination. Policy interventions and workplace initiatives should aim to raise awareness, challenge biases, and promote inclusivity, especially for older women in such roles.


Altonji, J G and C R Pierret (2001), “Employer learning and statistical discrimination”, The Quarterly Journal of Economics 116: 313–350.

Baum, C L and W F Ford (2004), “The wage effects of obesity: A longitudinal study”, Health Economics 13(9): 885-99.

Brunello, G and B D’Hombres (2007), “Does body weight affect wages? Evidence from Europe”, Economics and Human Biology 5: 1–19.

Cawley, J (2004), “The impact of obesity on wages”, Journal of Human Resources 39(2):451-74.

Cawley, J (2015). “An economy of scales: A selective review of obesity’s economic causes, consequences, and solutions”, Journal of Health Economics 43: 244–268.

Cawley, J and C Meyerhoefer (2012), “The medical care costs of obesity: an instrumental variables approach”, Journal of Health Economics 31(1): 219–230.

DeBeaumont, R (2009). “Occupational differences in the wage penalty for obese women”, The Journal of Socio-economics 38(2): 344-349.

Dolado, J, L Minale and A Guerra (2023), “Uncovering the roots of obesity-based wage discrimination: The role of job characteristics”, Labour Economics 83, 102425.

Han, E, E C Norton and S C Stearns (2009), “Weight and wages: Fat versus lean paychecks”, Health Economics 18(5): 535-48.

Oster, E (2019), “Unobservable selection and coefficient stability: Theory and evidence”, Journal of Business & Economic Statistics 37: 187-204.

Sawers, C and R Dobbs (2014), “Obesity: A global economic issue”,, 13 December.

Tekin, E (2007). “Obesity and wages: using body composition as an alternative to BMI”,, 18 December.