Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Many states and localities have either recently enacted or are considering bans on smoking in restaurants, making this one of the most active areas of tobacco control policy. The bans are intended to reduce secondhand smoke exposure for both the customers and employees of restaurants. Restaurant smoking bans might also help encourage smokers to cut down or quit entirely. However, in addition to their public health benefits, the bans might change the restaurant-food consumption patterns of smokers and nonsmokers. Previous research explores the economic impact of smoking bans on the restaurant industry, such as taxable restaurant sales. Much less is known about the impact on consumer behavior. For example, some previous empirical studies find that smoking bans have approximately zero net effect on restaurant sales. This finding could mean that most consumers simply do not respond to a ban one way or the other. Alternatively, the same finding could result from strong but offsetting consumer responses by smokers and nonsmokers. We estimate an econometric model of the impact of restaurant smoking bans on the demand for restaurant food by smokers and non-smokers. We use data from multiple waves of the Simmons National Consumer Survey (NCS) and the Consumer Expenditure Survey (CES) from 2000 – 2004. The combined waves of the NCS provide measures on the number of visits to family and fast food restaurants for a sample of over 110,000 adults. The combined waves of the CES provide a measure of expenditures on food away from home for a nationally representative sample of over 37,000 households. We use geocode information to merge the individual-level NCS and CES data with data on restaurant smoking bans in the respondents’ state and locality. We improve upon most previous studies of smoking bans because we take into account local bans and we take into account the bans’ strictness. With these improvements, our measures of restaurant smoking bans show substantial variation across states and over time in our sample period. For example, the fraction of the NCS sample subject to any state-level smoking ban, regardless of strictness, stays roughly constant at about 64% from 2000 – 2004. In contrast, the fraction of the NCS sample subject to strict ban at either the state- or municipal-level increases from about 16% in 2000 to 35% in 2004. A complete policy analysis of restaurant smoking bans requires a better understanding of the bans’ consequences for the consumers of restaurant food and the businesses that serve them. A few highly publicized studies find that restaurant smoking bans have a positive impact on restaurant sales. This is surprising because in this situation profit-maximizing restaurant owners should voluntarily accommodate customers’ preferences for a ban. However, restaurant owners may be in a prisoner’s dilemma type game where each owner loses if she individually enacts a ban, but the industry as a whole gains through the coordinated action of a uniform smoking ban. Because our results quantify the impact of the bans on smokers’ and non-smokers’ demand for restaurant food, we shed new light on the plausibility of the prisoner’s dilemma scenario. Furthermore, our results shed new light on the public health benefits of the bans for the smokers themselves: the bans’ consequences for smokers depend on whether the bans prompt them to change their smoking behavior or their restaurant-going behavior.