smokingWharton health care management professor Mark Pauly looks at recent reaction to a health system’s decision not to hire new workers who smoke. Much of that discussion centered on the ethics of such a move. But Pauly, an economist, suggests a different framework for analyzing the issue. His comments, below, appeared earlier this week in a new blog — Voices@LDI — started by Penn’s Leonard Davis Institute of Health Economics.

The recent debate about the ethics of a health system’s decision to refuse to hire new workers who smoke was understandably couched in the framework of ethics. But hiring a worker — buying labor — is an economic transaction that occurs in a market, and so it may be useful to discuss the economic framework as well.

The very narrow framework presented was of a firm (a hospital, in this case) hiring a given number of workers in a labor market where more workers are willing to work at the wage-benefit package it offers than it needs to hire. Thus the ethical issue is framed as whether it is fair to “give” one of the jobs in short supply to a non-smoker rather than a smoker. But wages and demand for labor are determined in markets, both for labor and for products or services. Considering these market level effects, I argue that that some of the framing of the ethical problem is factually dubious and that the conclusions about fairness may need to be modified.

Wages, Jobs, Discrimination

In general labor economics, there are two well-known predictions: 1) the total compensation paid to a given worker in a given job will equal the marginal revenue the firm expects to bring in because it hired the worker, and 2) at that level of total compensation, labor markets will “clear” (all workers who want to work at that level of compensation will find employment and employers will not be looking to alter either employment or total compensation). How might smoking affect the equilibrium of total compensation, money wages, and employment levels in a competitive labor markets?

Smokers are less productive. The empirical evidence that smokers have lower productivity over working years is fairly strong. They are less productive because they may miss work more frequently (absenteeism) and because their smoking-related health conditions (shortness of breath, cardiovascular conditions) may affect their ability to perform certain kinds of jobs (presenteeism). Even if the firm does not pay sick leave, the disruptive effects of more frequent absences in jobs that require teamwork (such as hospital nursing) will probably lead initially to costs on employers. These cost effects may be minimized if wages can be varied for a given job, or, if wages cannot be varied, by hiring non-smokers.

Smokers have higher medical costs. Smokers have higher average health care costs than non-smokers, other things being equal. (They may have higher life insurance costs but lower pension costs, as well.) Almost all job-related insurance in hospitals is experience-rated or self-insured. Even if employee premiums for health and life insurance are not adjusted for smoking status and wages are uniform within firms, the fact that medical costs will increase for firms or occupations with larger shares of smokers means that there will be offsetting reductions in money wages.

Consumers and coworkers may prefer non-smokers. In face-to-face interaction with workers, buyers may have preferences about behaviors or appearance of workers; attributes that buyers regard as negative will depress total compensation and wages. In contrast, for occupations and industries where consumers see only the final product (a manufactured good, a piloted airplane) there should be no such discriminatory effects. However, if co-workers have similar preferences, they may require higher money wages to work with smokers.

Labor Outcomes at the Firm and Market Levels

The effects of smoking are easier to predict at the level of overall labor markets. Suppose that, across local labor markets, smoking rates vary for workers in a particular occupation or industry. The prediction is that smoking will lower average total compensation in those markets with larger proportions of smokers; whether compensation packages within markets are lower for smokers specifically or for all workers in high-smoking communities or occupations depends in part on what is administratively feasible and/or legal. If wage/premium discrimination is not possible, the compensation of all workers will be depressed, but aggregate data across markets will show lower total wages and compensation for smokers since they are more likely to be present in high-smoking markets. Higher benefits costs for smokers will further depress money wages unless employee-paid premiums can fully reflect higher benefits costs. Either way, take home pay will be lower in such markets.

If all firms in a market hire from the same labor pool with its mix of smokers and non-smokers, these effects will be uniform across firms. However, individual firms may  correctly estimate that, if they increase money wages above the prevailing level but then only hire non-smokers from the pool of applicants, the firm  may lower total labor cost. Note that behaving in such a discriminatory  fashion has a cost to the firm that follows this policy — it pays higher total compensation or hires less qualified workers — but its bottom line may still be better off.

This was one of Milton Friedman’s most perceptive arguments about discrimination, whether racial or habit-related: Whatever it does to workers, it costs employers money to discriminate. If all or most firms in a local market decide to discriminate, either money wages of smokers will fall (if wages can be discriminatory), or there will be an increase in unemployment among smokers if wage and/or insurance premium discrimination is not possible.

Ethical Arguments

Now, about those ethical arguments over not hiring smokers…

As an economist, I have no expertise in ethical judgments, but I can discuss how this economic framework might interact with the ethical arguments on both sides.

  • The employer who discriminates against smokers pays a price for doing so compared to not doing it — just what Friedman said about racial discrimination. The employer may offset a benefit it expects to get against this price, but it is making a sacrifice; discrimination is not free, and the worse it is, the more it costs the discriminator.
  • If all employers of a particular kind of labor (e.g., all hospitals) discriminate in a particular way against smokers, their wages are likely to fall. This means that, despite the point in the previous paragraph, employers may paradoxically, after the dust clears, end up paying little or no cost, though each employer would still lower its labor costs if it stopped discriminating. It all depends on alternative jobs available to smoking workers and      whether those jobs discriminate.
  • Any effects on health insurance costs can be handled with higher premiums for smokers — so that, if such smoker surcharges are in place, higher medical costs are no longer a reason to refuse to hire smokers; they pay their own way.
  • Discrimination for reasons other than medical costs will cut smoking workers’ job opportunities to a greater extent if wages cannot be varied by smoking status. If wages can be varied, jobs will still be there (although at lower wages, smokers may not want them) but both incentives to stop smoking and compensation to the employer for any distress to consumers will occur. The effect need not be large if smokers seek jobs in other industries, but it can happen.
  • The public health argument for discrimination is that increasing the cost to workers of smoking — you won’t be able to get a good job in our firm or industry — may cause workers to stop smoking or never start. This societal benefit should be offset against perceptions of ethical unfairness. Maybe, but this is a slippery slope. Think of other behaviors that society would like to discourage, such as teenage pregnancy or committing a felony. Should good jobs be closed to young mothers and felons who served their time? There is probably some calculus that can rationalize discrimination here, but it is distinctly unlovely — and there must be better and more effective ways to change behavior than this. Even without accepting the argument that there are kinder and better ways to get people to stop smoking, the “no job for you smokers” strategy seems like (pardon the expression) overkill.

Of course, the argument that discrimination is unethical because it penalizes smokers for an involuntary addiction also suffers by extension, because taxes on cigarettes do the same thing, and few would think that taxing smoking or drinking (alcoholic or sugar-sweetened) beverages is unethical because it discriminates against addicts. Plus, the incontrovertible fact that this addictive behavior responds to prices shows that it is not perfectly addictive — so do we need to decide how much is addiction and how much something else?

The most fundamental message is one you might expect from an economist: Whatever your ethical goals, it is better to achieve them by adjusting wages and prices than by rules about who gets a job. Using financial rewards and penalties can change behavior at least as effectively as discrimination, without adding excess harm to workers who cannot kick the habit. Like carbon taxes, these penalties may not seem to some to be sufficient punishment of the guilty or statement of virtuous behavior by the employer — if you are allowed to sin if you pay the price, aren’t we being nonserious about sin and letting sinners off too easily? But overall, these strategies may be better in the sense of being fairer and more effective at discouraging behavior whose costs outweigh its benefits.