Thursday, 28 January 2010

Smoking bans and offsetting effects

Ban smoking in restaurants and bars and children wind up being more exposed to second hand smoke.

We evaluate the effect of smoking bans and excise taxes on the exposure to tobacco smoke of nonsmokers, and we show their unintended consequences on children. Smoking bans perversely increase nonsmokers' exposure by displacing smokers to private places where they contaminate nonsmokers. We exploit data on bio-samples of cotinine, time use, and smoking cessation, as well as state and time variation in anti-smoking policies across US states. We find that higher taxes are an efficient way to decrease exposure to tobacco smoke.
Except, of course, that smokers pay far more in tobacco taxes than they ever cost the health system; efficiency isn't the healthists' goal.

HT: Modeled Behavior, who also notes that the ideal citizen (who produces large positive externalities) will be a smoker...


  1. So in a utilitarian sense - leaving aside the children for a second - the smoking ban and raised taxation are working perfectly well. You remove smokers from public places, the larger proportion of society that doesn't smoke is happier and the state gets a greater amount of revenue from smokers who take less from healthcare than they put in. The state can then spend that money on things beneficial to society as a whole.

    In fact, the only people who are significantly affected are children and those who choose to be in the company of smokers. And if you choose to be in the company of smokers, then that is, after all, your choice.

    As for the children - well, that needs to be up to the adults, who should perhaps be reminded that the ideal first principle is do no harm.

  2. Depends on the costs to smokers of shifting their smoking, the benefits to non-smokers of having the smoking be shifted, and the ownership status of the venue in which the smoking would otherwise have taken place. In bars and restaurants, I'm convinced by the Coasean argument that the efficient amount of smoking would otherwise prevail; banning it then has to push to an inferior outcome.

  3. I seem to recall an Irish study that found that the smoking ban in that country led to a decrease in the prevalence of smoking in the home. There's also the partially tested idea that if smoking is displaced to the home then the alcohol industry still benefits because people buy alcohol to drink in the home instead. Those smokers then benefit because they save money on other external costs related to drinking out.

    Of course, all this talk of the children misses the point of the ban in the first place - in most countries they're the result of labour legislation and worker protections from hazardous substances, not protection of the general public.

  4. Don't know the Irish one, but it's plausible that it goes either way: there'll be a substitution effect towards the home and away from banned places, and a kind of an income effect where the inconvenience of being a smoker causes some to quit. Relative magnitudes of the two effects then matters, but in theory it could go either way.

    Don't think the pub owners much like the shift to drinking at home :>

    Are bans on specific environmental pollutants a necessary consequence of workplace safety legislation? I would have thought it would have come either out of tort, where a worker would sue the employer on basis that tobacco smoke concentrations above X ppm constituted a health hazard, or a regulatory ruling that concentrations above X constituted a health hazard. Then pubs would have the option of ventilation systems or smoking bans. At least, that's a version consistent with it really being about worker protection similar to worker protection against other workplace risks. The blanket ban is consistent with something else entirely.

  5. Well, in the UK the legislation is the Health Act 2006, enforced by regulations in each of the four countries. Pubs and clubs fall under 2(2)(a) of that act, being places of work. Ireland defines it in the same manner: that pubs and clubs are not public spaces but they are workplaces. The EU take a similar line - they're all about workplaces because no-one has a right of access to establishments.

    NZ, I've found, is slightly different, because the law stems from the Smoke-free Environments Act 1990 but that act specifically defines pubs and clubs as a workplace because they're not public spaces.

    It would be a tort - in fact, it is, it's a statutory tort because established medical fact at this point is that it's harmful in the same way as other types of exposures. But it won't ever be as specific as a certain concentration of a substance. Say, for example, in NZ, the Resource Management Act doesn't call for a certain level of particle presence as recorded for smoking chimneys, but merely that the smoke would cause a nuisance or that the odour would be objectionable. Because tobacco smoke contains concentrations of substances known to be hazardous and appears on the IARC group 1 carcinogen list, any exposure in the course of employment would mean that you'd probably have to have the same kind of safety considerations as you would for other routine exposures during the course of employment.

  6. None of that tells me why good ventilation systems that reduce exposure to negligible levels aren't sufficient. As for the carcinogen list...oral contraceptives are on the same list. Are prostitutes working in brothels forbidden from using them as complement to condoms? No?

  7. No, but that's their choice and no-one else is subject to their choice. :) You don't spray oral contraceptives into the air for everyone else in the place to breathe. Tanning beds are also on the same list, but tanning salon employees aren't forced to be exposed to the beds.

  8. But the Coasean point is that the property owner could decide to ban or not ban smoking from the premises, and workers can decide whether to work there or elsewhere, and customers can decide whether to drink there or elsewhere. Compensating differentials combined with preference-based sortition can work wonders....