Contagion at work: Occupations, industries and human contact

Abstract

Using nationally representative micro panel data on flu incidence from the Medical Expenditure Panel Survey in the United States, we show that employed individuals are on average 35.3% more likely to be infected with the flu virus. Our results are robust to individual characteristics including vaccinations, health insurance and individual fixed effects. Within the employed, we find significant differences in flu incidence by occupation (e.g., sales occupations show 40.5% higher probability of infection than farmers) and by industry (e.g., education, health and social services show 52.2% higher probability of infection than mining). Further, we show that the interaction between occupations and industries is important to understand contagion. Indeed, cross-industry differences in flu incidence cannot be fully explained by differences in the within-industry occupation structure. As a potential mechanism for contagion, we study how flu incidence varies with the extent of human contact interaction at work—with an occupation-industry-specific score that we construct based on O’NET occupational characteristics. We find that the higher the human contact at work, the greater are the odds of infection. Our results are larger in years of high aggregate flu incidence and robust to firm size, a number of jobs and hours worked.