Julian Sanchez header image 2

photos by Lara Shipley

Anecdotes and Aggregates

November 23rd, 2003 · No Comments

I figured it might be worth cross-posting some comments I made on this thread, where one poster argues that it’s dishonest or irrelevant to bring up anecdotal claims of plummeting business at specific bars due to smoking bans, because aggregate analyses in California and New York show no net economic impact. First, I can think of at least a few reasons off the top of my head why a negative effect would be more likely in Montgomery County than in a whole state that’s banned smoking—lower exit costs for smokers among them. But leave that aside for now.

I’ll suggest that anecdotes do have a specific, if limited, usefulness in cases like these precisely because it’s difficult isolate aggregate causation in these macro-analyses. Have you controlled for all possible relevant variables when asserting that the direction of change in revenues was or wasn’t attributable to a change in the law? When you’re looking at a whole city or state, I doubt that a very high level of certainty is possible. Bar owners on the ground seem at least somewhat more likely to have insight into the reasons for declining (or rising) revenue.

It’s important for a second, more salient reason as well: it establishes that the net effect (some non-smokers going out more, some smokers staying in more) is NOT evenly distributed across bars. Some, apparently, are doing better, others worse. Why does that matter? Because absent a uniform regulation, the business-improving option, banning smoking, is ALREADY available to bar/restaurant owners, many of whom would presumably have eventually implemented the policy that maximizes their revenues. As soon as you factor in dynamic individual-business policies, I think there’s a far more compelling case for an economic impact, even in the aggregate. NOT a decrease relative to present revenues, but the failure to attain the higher revenues that would be manifest as the proportion of smoking and non-smoking establishments approached equilibrium. In other words, assume the ban is repealed. The places reporting that the ban has increased their business because their patrons don’t like smoke continue the smoke-free policy, preserving (most of) their gains. The places reporting that their chain-smoking patrons are deserting them revert to a smoke-friendly policy and recover (a substantial portion of) their previous clientele. It should be fairly intuitive that this yields an aggregate increase in revenue for those bars. It’s then a short step to the conclusion that a ban is costing the difference between the status quo and that higher revenue level. (Ironically, as suggested earlier, passing and then repealing a ban may actually be more efficient than never having passed one, because it reveals to each bar or restaurant which policy is optimal for them, by showing them which way business trends under a smoke-free policy.)

Tags: Uncategorized