Adult smoking prevalence has been falling in the United States, down from 42.4 percent in 1965 to 24.1 percent in 1998, a 43 percent reduction. The percentage of adults who never smoked increased from 44 percent in the mid-1960s to 55 percent in 1997.1 It is not surprising then that owners of restaurants have responded to changing customer preferences by changing how they allocate their air space between smoking and non-smoking use. Restaurants without non-smoking seating sections have become exceptions. These provisions are not isolated to locations where state or local laws restrict or ban smoking and indicate an active private market in accommodation.

It has been shown that areas with lower adult smoking rates have more smoking laws and bans. It is important to note, however, that such laws have been introduced without much benefit of research on how the private market has dealt with the issue of accommodation of both smokers and non-smokers. Proponents of smoking restrictions often argue that smoking exerts a negative externality on non-smokers and that governments should control smoking through laws and bans. [Campaign for Tobacco Free Kids, 2002] Even if such externalities exist, it would appear to be useful first to ask whether, in the absence of laws, private markets work toward internalizing the externalities. An active market in accommodation might indicate that laws and bans are simply overturning, partially or totally, actions of owners that had successfully dealt with smoking-related externalities.

This paper examines the diversity of private market accommodations in roughly 1,000 restaurants and bars in Wisconsin. We develop a model that predicts which factors determine the share of seating devoted to non-smoking use within the setting of profit-maximization. While it is important to note that the data are limited to Wisconsin, this is likely a reasonable case study for a number of reasons. First, the state has few stringent smoking laws, providing a rich data set to explore how voluntary choices of owners are related to various factors. This analysis examines laws each owner may be subject to and considers this information in the model. Wisconsin is also not an outlier because it does not ban smoking everywhere, as does California, nor does it fail to control its use in restaurants, as does Nevada. Finally, the adult smoking rate in Wisconsin is 23.4 percent, which is well within two standard deviations from the median smoking rate of 22.9 percent of all states

The paper begins with a literature review summarizing previous studies that address smoking policies. This is followed by a description of the data. An economic model then indicates which factors underlie smoking policies chosen by owners, and determines which factors are important to smoking policies in the absence of government restrictions and which factors are no longer important following the imposition of smoking laws. The conclusion discusses implications about the private market in accommodation and the effects of smoking bans and restrictions on individual owners.

LITERATURE REVIEW

Only limited research has been conducted on the private market in accommodation. Two studies directly examine seating allocations within restaurants and bars. Boyes and Marlow [1996] examine survey data in San Luis Obispo, one of the first cities to institute a ban on smoking in restaurants and bars. Sixty-four restaurants and bars were surveyed-roughly 65 percent of all affected businesses. The authors suggest that an active market in private accommodation existed prior to the ban because 62 percent of owners had non-smoking sections.

Logit models of support for bans in San Luis Obispo were estimated using data from a random survey of 764 individuals, 84 percent of whom were non-smokers. The empirical evidence indicates that the odds of supporting the smoking ban are significantly lower for respondents who smoke and are male. Age, education, and whether or not a respondent resided in the local community did not significantly influence odds of supporting the ban. Despite widespread support for the bans a significant percentage of the non-smokers (62 percent) also believed that smoking/non-smoking sections dealt effectively with smoking prior to the ban. Apparently, while a majority of non-smokers believed that the private market in accommodation was effective, they nonetheless favored outright bans.

Dunham and Marlow [2000a] also study private market accommodation. They examine national survey data drawn in 1996 from owners/managers of 1,300 randomly selected restaurants (650) and bars/taverns (650) across the United States. Samples were drawn in a statistically random manner and were applicable to all such establishments with a maximum sampling error of approximately plus or minus 4 percentage points. The survey found that, for restaurants, on average, 54 percent of seating was allocated to non-smoking, and for bars/taverns 5 percent was thus allocated. Both restaurants and bars/taverns exhibited cases where smoking was entirely prohibited as well as allowed throughout establishments. The authors conclude that state smoking laws do not influence allocations of non-smoking seating and suggest that laws are passed in states with relatively few smokers and therefore owners had already allocated relatively many seats to non-smoking use prior to the passage of laws. This hypothesis is supported by their finding that states with relatively many people involved in the tobacco manufacturing industry are significantly less likely to pass smoking laws than those states with relatively few workers. They also find that the percentage of seating allocated to non-smoking is negatively related to the percentage of smokers in the population, is significantly higher for owners affiliated with chains, and is significantly lower for older businesses and for bars