Definition
The empirical and causal relationship between body mass index (BMI), particularly morbid obesity (BMI ≥ 40), and labor force non-participation. The central question is how much of the rise in prime-age men's non-employment over the past three decades is attributable to the parallel rise in obesity, and whether the obesity-employment relationship itself has changed over time.
Key Ideas
Compositional vs. Propensity Change
The rise in obesity can affect aggregate non-employment through two channels:
- Compositional channel: More people in a persistently high-risk category (more morbidly obese people → more non-employed people, even if the obesity penalty is constant)
- Propensity channel: The relationship between obesity and non-employment has changed (the obese are now relatively more likely to not work than before)
Butcher and Park (2008) find evidence for the first channel and not the second. The morbidly obese have consistently had worse employment outcomes, but the gap between morbidly obese and normal-weight men in reporting musculoskeletal conditions, routine needs disability, and Social Security Disability Insurance (SSDI) application has not statistically significantly widened from 1984 to 2005.
Supply Side vs. Demand Side
Non-employment among the obese can arise from:
- Supply side: Obesity impairs health and physical capacity; rising SSDI generosity relative to low-skill wages makes non-work more attractive
- Demand side: Employers discriminate against obese workers or avoid them to contain health insurance costs
These mechanisms are difficult to distinguish empirically. If health propensities were stable but employment gaps widened, that would indicate demand side changes. Butcher and Park find consistent null findings across health, disability, and employment outcomes, making the supply-side/demand-side decomposition indeterminate.
Oaxaca-Blinder Decomposition of Non-Employment
Butcher and Park apply Oaxaca-Blinder decomposition comparing 1984–85 to 2004–05 non-employment rates:
- Obesity alone: 3–13% of the 2.2 pp rise explained
- Obesity + age: 14–32%
- Obesity + age + race/ethnicity (with interactions): 34–47%
- The amount explained is similar under early vs. late period "returns to characteristics," consistent with a stable underlying relationship
This compares to Lakdawalla, Bhattacharya, and Goldman (2004), who found obesity explains 50% of disability rate increases for 18–29 year olds and nearly all of the increase for 50–59 year olds (1984–1996) using a three-component decomposition.
SSDI Diagnosis Composition
Coinciding with obesity trends, SSDI primary diagnoses shifted away from easily-verified conditions toward subjective ones (1981–2003):
- Musculoskeletal disorders: 17.0% → 26.3% of all awards
- Mental disorders: 10.5% → 25.4%
- Heart disease: 24.9% → 11.4%
- Cancer: 16.3% → 9.4%
The post-1984 change in eligibility rules (more weight on self-reported pain and functional capacity) and rising wage inequality (making SSDI more attractive relative to low-skill wages) likely drove these compositional shifts.
Broader Economic Context
- Economic framework for obesity (Cawley and Ruhm 2011): The economics-of-risky-health-behaviors literature frames obesity as a negative investment in health capital (Grossman model) with documented labor-market consequences for wages and employment. It also establishes the steep socioeconomic gradient — college graduates have ≈8.7 pp lower obesity than high school dropouts — and documents that U.S. obesity more than doubled (≈15% in the early 1970s to ≈34% by 2003–2006). These trends supply the population-level backdrop against which obesity's contribution to non-employment is assessed.
- Income does not cause weight, at least for the elderly (Cawley, Moran, and Simon 2009): Using the Social Security benefits "Notch" (a policy accident giving 1915–1917 cohorts permanently ~7% higher benefits) as an instrument for unearned income, the authors find no significant effect of income on body mass index or any clinical weight category, with tight bounds ruling out economically meaningful effects. This bears on the reverse-causality concern below: a pure income → weight channel is an unlikely explanation for the obesity–non-employment correlation, pointing instead toward health, productivity, and selection mechanisms.
Why It Matters
- Male non-employment puzzle: Rising prime-age male non-employment against a backdrop of generally improving mortality-based health metrics. Obesity is a key health dimension that has not improved and may explain part of the paradox.
- Policy targeting: If the obesity-non-employment relationship is stable, targeting specific periods for intervention is less important than reducing the prevalence of morbid obesity itself.
- Disability Insurance (DI) growth drivers: The obesity explanation is a complement to (not substitute for) the programmatic-incentives explanation (Autor and Duggan 2003). Rising obesity rates may have filled the pool of SSDI-eligible applicants that expanded incentives converted into beneficiaries.
Open Questions
- Causal identification: Without a valid instrument for BMI, the extent to which obesity causes non-employment (vs. both sharing common determinants) remains unclear.
- Reverse causality: Sedentary work and non-employment may themselves contribute to obesity, complicating interpretation.
- Recent trends: Post-2008 recession and post-2020 declines in SSDI enrollment — how much does continued obesity growth offset the enrollment decline?
Related
Sources