Association for Public Policy Analysis & Management

In the U.S., smokers’ wages are 4-11% lower than non-smokers (Levine et al., 1997; Viscusi and Hersch, 2001; Grafova and Stafford, 2009). There are at least four competing explanations for this empirical regularity. One, smoking could directly reduce productivity via diminished health. Two, smokers may face unequal or unfair treatment. Three, the decision to smoke might be endogenous to wages: only those who (expect to) earn lower wages choose to smoke. Four, and the focus of this paper, wages for smokers (in the U.S.) should account for any additional cost of providing employer-sponsored health insurance (ESI).

The cost of ESI matters because ESI is experience-rated at the firm level. Therefore, the cost of ESI for any given firm will track the medical expenditures of the firms’ workers. ESI therefore creates a cost-wedge between workers with varying medical expenditures unless employers can adjust wages to compensate. For that reason, this paper examines how the Affordable Care Act’s employer mandate affected the wages of smokers relative to non-smokers. The mandate, announced in 2010, requires firms with more than 50 full-time equivalent employees (FTEs) to provide ESI to full-time workers from 2014. Because the mandate makes some workers relatively more expensive to employ, such as smokers, it provides an excellent source of identifying variation.

To estimate if smokers’ wages are lower due to ESI (and therefore if smokers “pay” for their own medical expenditures), the paper uses data on income, ESI coverage, smoking status, and firm size from the 2006 to 2014 waves of the Medical Expenditure Panel Survey (MEPS) in a difference-in-difference framework. The paper’s findings suggest that smokers and non-smokers who will receive ESI due to the employer mandate tend to bear the cost of that coverage via relatively lower wages. Compared to non-smokers, estimates suggest that smokers experience an additional mandate-related wage offset of between 19 and 33 cents per hour. The size of this effect aligns well with the actual medical expenditure differences between employed smokers and non-smokers. The estimates have a causal interpretation if nothing else affects smokers’ and non-smokers’ wages differently in the 2006 to 2014 period.

For credibility, the study also examines the same estimates for two “control” groups: (1) workers at firms who are not affected by the mandate because they have fewer than 50 FTEs and (2) workers who are already offered ESI by their employer (without being required to do so by the mandate). In both cases, there is no differential effect on the wages of smokers.