Standard economic analysis would suggest that the proportion of bars (or restaurants… but let’s stick to bars) in a given region that adopt smoking vs non-smoking policies is a simple function of consumer demand. If most bars permit smoking, on this logic, it must be because that reflects the preferences of bargoers. If very many people wanted a nonsmoking environment in which to drink (and/or smokers didn’t much mind having to go outside), the argument runs, we’d see many more non-smoking bars. Therefore, we should expect laws banning smoking in bars to be economically harmful, as marginal carousers opt not to stick around for that extra drink—or not to head out to the bar in the first instance.
Certainly, there’s plenty of anecdotal evidence to support this model. I’ve witnessed at least one cafe go under for precisely this reason, as the chain-smoking hipsters who’d once kept it bustling and lively abandoned it, until even the non-smoking patrons no longer found it a vibrant place to go and chat or listen to music. But there are also solid seeming studies that fail to find this effect in the aggregate. Why wouldn’t we see this happen? What’s going on there?
One possibility is that the standard model ignores strategic interaction. The market for bars only presents a parametric choice situation if you’re a fucking drunk. The rest of us go out drinking with people, and, moreover, go out in permutations that shift from evening to evening (or within an evening). On plausible assumptions—smokers generally want a smoking bar, non-smokers a non-smoking, but both would rather give on that preference than break up the carousing group—you’ve got your standard ballet/boxing game. Except, again, you often don’t know the players in advance & don’t want to have to renegotiate which rule is operative every time the group changes, so it’s useful if everyone has some simple norm for what they do in mixed smoker/nonsmoker groups.
Either a “defer to the smoker” or a “defer to the non-smoker” norm is a Schelling point. Start out with a high enough number of smokers (with intense enough preferences) and people converge on a “defer to the smoker” norm that doesn’t get renegotiated, even as the balances begin to shift. The deference-to-the-smoker rule is pretty well internalized. Impose a ban, and the norm may simply change, with the desire for group cohesion overriding the individual preference. Of course, this is still likely to be severely suboptimal, but it does raise the interesting possibility that a temporary ban repealed after, say, a couple of years, might conceivably help overcome a kind of market failure. Not that I’m advocating any such thing (not by a longshot)… just an interesting hypothetical.