Monday, October 23, 2006

Restaurant Smoking Bans

Thomas A. Lambert, associate law professor at the University of Missouri School of Law, writes an op-ed in today's Washington Post against the proliferation across the country of restaurant smoking bans ("Against Restaurant Smoking Bans").

Lambert says:
Proponents of smoking bans contend that they address the "negative externality" smokers impose upon nonsmokers, shape individual preferences against smoking (thereby reducing the incidence of smoking), and alleviate the health risks associated with exposure to environmental tobacco smoke or "ETS." Considered closely, though, each of these arguments fails to justify the government-imposed restrictions.
He then argues that: (1) there are no widespread negative externalities (because if non-smokers or restaurant employees don't like smoky restaurants, they will go somewhere else, leaving the restaurant owner to take on the costs of the externalities); (2) the government has no business shaping individual preferences (and, besides, campaigns against smoking only make children want to smoke more as a symbol of rebellion against authority); and (3) the dangers of ETS are overblown.

I'll only look at Point 1 here. Point 2 is worth an entire debate. (Also, the point about childhood smoking is ridiculous, since it's not smoking per se that's the problem, but habitual smoking. It is clear that the proportion of the population that smokes regularly has decreased steadily over the years as smoking bans and Surgeons General warnings have come into play. To argue that smoking bans do not have an effect on habitual smoker preferences is to suggest that a smoker's demand for cigarettes is infinitely inelastic and they will smoke no matter how costly or inconvenient smoking becomes.) Finally, I think restaurant smoking bans revolve around Point 1 and not Point 3, at any rate.

My question is, if there are no negative externalities to restaurant smoking, then why have these bans proven so politically popular? I suggest that this fact alone demonstrates that there is some kind of market failure here. Non-smokers make up the vast majority of the U.S. population. According to the Centers for Disease Control, nearly 80 percent of the U.S. population does not smoke, up from only 57 percent in 1965. With this kind of discrepancy, you would think smoking in restaurants would be a non-issue. If non-smokers preferred non-smoking restaurants, the sheer weight of their numbers would make most restaurants non-smoking. On the other hand (and this is the interesting question), if non-smokers do not care about whether they share restaurants with smokers (and this might otherwise seem to be the case, since so few restaurants voluntarily become completely smoke-free), why are restaurant smoking bans so politically popular?

I suggest two possibilities: first, only a relatively small portion of non-smokers are hypersensitive to smoking in restaurants (and I don't mean "hypersensitive" in a negative way). This group is not large enough to have an economic impact on restaurants by themselves, but it is large enough to initiate a political ban. This hypothesis, however, begs the question why, if hypersensitive non-smokers are not numerous enough to have an economic effect on restaurants, why are they still numerous enough to defeat the political interests of smokers and restaurants, who presumably enjoy smoking in restaurants more (in aggregate terms) than hypersensitive non-smokers gain from a smoking ban.

My second suggestion is that in addition to this core group of hypersensitive non-smokers, there is a much larger group of "positionally sensitive" non-smokers who support such bans. "Positionally sensitive" (PS) non-smokers are willing to share a restaurant with smokers, provided the smokers are confined to a distinct part of the restaurant away from where they are (i.e., the "smoking section"). This situation is satisfactory perhaps 90 percent of the time. However, PS non-smokers run the risk (particularly during busy periods) of either having to sit directly adjacent to the smoking section or having to choose between taking a seat in the smoking section or having to wait for a table to open in the non-smoking section.

You can then see how this changes non-smoker political incentives, even if it does not change restaurant economic incentives. If 20 percent of the population wants to smoke in restaurants, but only 10 percent of the population is hypersensitive to smoking and won't frequent restaurants that allow smoking, most restaurants will allow smoking. If an additional 50 percent of the population are "positionally sensitive" non-smokers (with the remaining 20 percent of non-smokers having no preference whether they sit directly adjacent to a smoker during meals),* then most restaurants will have an economic incentive to have smoking and non-smoking sections. This seems to be the case now.

However, at any given time, 5 percent of the population (or one-tenth of PS non-smokers) are being inconvenienced by having to sit directly next to the smoking section or wait for a non-smoking table to open. This means that, at any given time, 15 percent of the general population is strongly in favor of a universal smoking ban. The size of this group is still not as large as the population that smokes, but the 5 percent of inconvenienced PS non-smokers is a rotating 5 percent. That means that, over the course of 10 restaurant visits, 65 percent of the population would support a universal ban.

The reason that this issue has shifted from the economic to the political realm, however, is lottery effect for positionally sensitive non-smokers. A 10 percent risk of being inconvenienced is not sufficent for PS non-smokers to stop frequenting a given restaurant. While they might otherwise forgo the benefit they receive from the restaurant if the risk of being inconvenienced was 100 percent, they are not willing to do this if the risk is only 10 percent. However, by supporting a legal ban on restaurant smoking, PS non-smokers can eliminate this 10 percent risk at little cost to themselves. (There is still some cost. A universal ban on restaurant smoking might present a slight risk to the viability of any given restaurant, and positionally sensitive non-smokers would suffer if their favorite restaurant went bankrupt.)

In other words, what restaurant smoking bans are really about is an attempt to force 20 percent of the population (smokers) to absorb the 10 percent risk they cause to 50 percent of the population (PS non-smokers) , plus the 100 percent risk they cause to 10 percent of the population (hypersensitive non-smokers). The market failure comes in because, even with these preferences, no individual restaurant can afford to ban smoking on its own, since only 15 percent of the population is ardently in favor of a smoking ban at any given time in his or her restaurant, versus 20 percent opposed. However, 65 percent of the population is harmed by this situation at one time or another.

Because of this, I suspect restaurant smoking bans will prove to have a negligible effect on restaurants. This situation has characteristics of an arms race. If any one restaurant adopts a ban, that restaurant loses out to the competition, since they can reliably expect to gain only 10 percent new customers versus a loss of 20 percent of existing customers. If all adopt a ban, however, none lose. (To argue otherwise is to suggest that smokers so enjoy smoking with dinner that they will stay home and cook themselves rather than go out. While possible, this seems unlikely. Smokers must forgo smoking in all sorts of places today, so they are likely used to having to wait a little to enjoy a smoke. Plus, how good is your own cooking likely to be if your sense of taste has been destroyed by tobacco smoke?)

Bars, on the other hand, may be a different story. It's unclear whether smoking makes people drink more (in which case bar owners have an incentive to allow smoking) or drinking makes people more inclined to smoke (in which case bars are currently being ripped off by the tobacco companies and should start demanding a cut).


*Other than the CDC figures, these percentages are fictional and merely used to demonstrate the idea.

No comments: