New York University

"Should We Ban E-cigarettes? Theory and Evidence."

Electronic cigarettes are one of the most controversial new products of the past decade, due to uncertainty about their health e ects and whether they are primarily a quit aid or a gateway drug for traditional combustible cigarettes. We show that demographics with stronger latent demand for e-cigarettes had steady relative declines in cigarette consumption over 2004-2018 that did not change as e-cigarettes were introduced, suggesting that e-cigarettes had negligible e ects on cigarette consumption. We formalize this intuition in a shift-share econometric strategy that con rms that e-cigarettes are neither complements nor substitutes on average; our con dence intervals rule out that e-cigarettes a ected the 2004-2018 aggregate smoking decline by more than 10 percent in either direction. We carry out a new expert survey to aggregate the state of knowledge about the health e ects of e-cigarettes. There is material disagreement; the average expert believes that e-cigarettes impose externalities and \internalities" (harms to users that they do not correctly perceive) that are 48 and 115 percent as large as those of combustible cigarettes, respectively. We embed these empirical ndings in a behavioral welfare analysis with Monte Carlo simulations to account for uncertainty. Successfully banning e-cigarettes increases expected social welfare, although it decreases welfare in more than 40 percent of parameter draws.