What smoking bans tell us about optimal product diversity

By Renaud Foucart

Is it possible that a competitive market fails to supply the type of good or service desired by a majority of people? There is at least one well-documented example where this has been the case: nonsmoking bars and restaurants. When the first smoking bans have been implemented in the nineties, a large majority of consumers were already non-smokers. Yet, the market hardly supplied nonsmoking premises. Even non-smoking areas were only the result of legal requirements, and not something the market naturally offered.

In “Group Consumption and Product Diversity: The Case of Smoking Bans”, published in the current issue of the Journal of Industrial Economics, I argue that this result is consistent with standard economic assumptions, and could apply to all markets where consumers buy in possibly heterogeneous groups, and have to bear some information costs to experience a product.

Smokers and nonsmokers belong to three types of groups: smokers only, nonsmokers only, and mixed groups. Assume that for historical reasons all restaurants used to allow smoking. Suppose also that a majority of consumers do not smoke. If a few restaurants decide to ban smoking, they become in principle more attractive to nonsmokers. As there is little competition on the nonsmoking market, it is however in the interest of nonsmoking restaurants to charge a higher price for a given quality. Hence, a mixed group finds better deals on the market for smoking restaurants and never patronizes a nonsmoking one. There thus exists a market equilibrium in which nonsmoking restaurants are niche products that cater to some nonsmoking consumers only, while smoking restaurants are mainstream and competitive, and attract both smokers and nonsmokers.

In this world, smoking bans may not sound very attractive: few people enjoy the existing nonsmoking restaurants and are willing to force the ones they like to become similar to the nonsmoking ones. However, once a smoking ban is in place, the market for nonsmoking restaurants becomes more competitive, and it is in the interest of their owners to cater to everyone. This explains another well-documented fact: the support for smoking bans increases in the presence of bans. The bans are not very popular before being enforced but few people question them once they are implemented, as long as they apply to all premises.

The model also makes the following prediction: the worse possible policy is to ban smoking in specific places only. Such a partial ban does not work without strong enforcement, hurts the profit of owners, and remains unpopular, as it does not modify the nature of the market equilibrium. Either governments should ban smoking (almost) everywhere, or not ban it at all.

More generally, the model shows that, in the presence of group consumption, a “wrong” product diversity in which the product preferred by a majority is consumed by a minority is a likely market outcome. Hence, it is not always true that the mainstream product actually corresponds to the mainstream taste.

Read the full article (pay-walled) Group Consumption and Product Diversity: The Case of Smoking Bans, Journal of Industrial Economics, September 2017 or the last working paper version.