Tourism and Hotel Revenues Before and After Passage of Smoke-Free Restaurant Ordinances FREE

As the evidence that secondhand tobacco smoke endangers nonsmokers has accumulated,^1- 2 more and more communities have eliminated smoking in public places and workplaces. As of September 1998, 212 communities and 3 states had enacted laws mandating smoke-free restaurants³ and 1 state (California⁴) and 31 communities³ had enacted local ordinances requiring smoke-free bars. These ordinances not only protect nonsmokers from the toxins in secondhand smoke, they also create an environment that encourages smokers to quit.⁵

The tobacco industry vigorously opposes these public health measures to protect its sales. During the debates over these laws, it is common for the tobacco industry (acting directly or through front groups^6- 8) to claim that these ordinances create severe economic problems for the restaurants and bars. After Glantz and Smith^9- 10 published their study demonstrating that smoke-free restaurant ordinances have had no effect on restaurant revenues in the first 15 cities to pass such ordinances, the tobacco industry's claims of economic chaos lost credibility, particularly in California and Colorado, where the cities were located. Glantz and Smith^11- 12 later updated this study and extended it to include smoke-free bars. Subsequent work by other researchers yielded similar findings for smoke-free restaurant ordinances in 89 cities in 6 states.^13- 19 Despite tobacco industry protestations to the contrary, all the empirical evidence supports the proposition that smoke-free restaurant ordinances do not hurt the restaurant business.²⁰

As the tobacco industry's claims of adverse effects on the restaurant and bar business have lost credibility, it has advanced a new economic argument against passing smoke-free restaurant ordinances: these ordinances will adversely affect tourism. In some places, the industry has claimed that tourism from countries such as Japan and Germany will be particularly affected. There is only 1 study of 1 city on the effects of a smoke-free ordinance on tourism.¹⁸ We identified 3 states and 6 cities in which opponents of clean indoor air ordinances specifically advanced claims that the ordinance would adversely affect tourism (Table 1^21- 35) and obtained data on tourism from the local authorities. Contrary to industry claims, these ordinances were not associated with significant drops in tourism. Quite the contrary, in several locales the ordinances were associated with significant increases in tourism.

We searched newspaper databases and publications by tobacco industry groups (such as the National Smokers' Alliance that was created for Philip Morris Incorporated³⁶) and contacted tobacco control advocates in voluntary health agencies, nonsmokers' rights groups, and health departments to identify localities in which the issue of effect on tourism was raised in the debate over clean indoor air ordinances.

We then identified those local ordinances and state laws that required 100% smoke-free restaurants. (An exemption for the bar area of a restaurant did not disqualify a smoke-free restaurant ordinance from our study, so long as the eating areas were smoke-free.) Smoke-free restaurant ordinances and their effective dates were confirmed with local health department officials. This process yielded the 3 states and 6 cities that met the criteria for inclusion in the study outlined earlier. Because hotel revenue data for Aspen, Colo, were not available predating passage of its ordinance in 1985, we could not include it, leaving 6 cities for analysis (Table 1).

We used revenues from hotel rooms as our measure of tourism. Data on hotel revenues were obtained from the appropriate authorities (Table 2^37- 46). We analyzed the hotel revenues directly and in constant 1997 dollars using the appropriate seasonally unadjusted all-items consumer price index.

We also analyzed hotel revenues as a fraction of total retail sales, similar to the analysis we did in our studies of restaurant revenues.^9- 12 Analyzing hotel revenues as a fraction of total retail sales accounts for underlying economic conditions and inflation.

In our earlier studies,^9- 12 we compared restaurant revenues in similar control cities that did not have 100% smoke-free restaurant ordinances. Rather than doing a locality-by-locality matching, in this study our comparisons against control are done by comparing hotel revenues in the study localities with hotel revenues for the entire United States. We followed this approach because, unlike our earlier study, there was often not a natural match to the study cities and states or, when there may have been a logical match, the "control" locality did not have available data or had a smoking-restriction ordinance in place that prevented it from qualifying as a control locality. Comparing revenues in the study localities with the United States as a whole controlled for the overall health of the tourist industry.

The issue of impact of smoke-free ordinances on international tourism was raised in California, Utah, and New York City (Table 1). We obtained data on the numbers of international tourists for California, Utah, and New York City (Table 2) and analyzed the effects of the ordinance on the number of tourists over time. The dependent variable was the hotel revenues in the study locality divided by total US hotel revenues for the same year. To facilitate comparisons between localities, this ratio was normalized by 1989 population for each locality (Table 1) divided by the US population (248,709,873) from the 1990 census.²¹ Data were analyzed with linear regression: y=b₀+b_tt+b_L(t−t_law)L+Σb_iS_i where y indicates the dependent variables in Table 3 and Table 4; t, time to represent the underlying secular trend; L, a dummy variable that indicates whether a smoke-free restaurant law is in force; and t_law, the time the law went into force. The dummy variable L quantifies the presence of a smoke-free restaurant ordinance according to L=0 if no ordinance and L=1 if an ordinance is in effect. For the period in which the ordinance goes into effect, L is set to a value between 0 and 1 that corresponds to the fraction of the period that the ordinance was in force. The term b_L (t−t_law)L models the effect of the smoke-free law as a change in the slope of tourism revenues or volume over time. This approach differs from our earlier work, which modeled the effect of the ordinance as a simple intercept change. We found that modeling it as a slope change consistently gave better fits to the data than an intercept change model; the results obtained with an intercept change model were qualitatively similar to those presented in this article using the model above. For locations where data were available more frequently than annually (ie, quarterly or monthly), we also included a dummy variable, S_i, to allow for seasonal variability. The estimate of the coefficient b_t quantifies the annual rate of increase (or decrease) in the dependent variable y each year. The coefficient b_L quantifies the magnitude of the effect of the ordinance on the rate of change over time of the dependent variable.

For hotel revenues as a fraction of retail sales and normalized locality hotel revenues divided by total US revenues, we also conducted a pooled analysis with the equation above by adding effects-coded dummy variables to code for between-locality effects. The pooled analysis was done using annual data for all localities. A change is considered statistically significant when P<.05.

Table 3 and Figure 1 present the results for total hotel revenues over time before and after implementation of the law. In terms of constant 1997 dollars, the smoke-free law was associated with a significant increase in the rate of growth of hotel revenues in 4 localities, no significant change in 4 localities, and a significant slowing in the rate of increase of hotel revenues in 1 city (Flagstaff) where revenues tended to flatten out. Analysis of hotel revenues in current dollars or as a fraction of total retail sales (Table 3) yielded similar results. Pooled across all localities, there was no significant change in the fraction of hotel revenues as total retail sales (P=.16).

The smoke-free law was associated with no significant change in the rate of growth of hotel revenues compared with the United States as a whole in 5 localities, a significant speeding in 2, and a significant slowing in 2 localities (Table 3). Pooled across all localities, there was no significant change in the rate of change of hotel revenues compared with the United States as a whole (P=.93).

Figure 2 and Table 4 show the changes in the number of tourists from Japan (or Asia) and Germany (or Europe) associated with implementation of the California, Utah, and New York City smoke-free restaurant ordinances. The implementation of the ordinances was associated with a significant increase in the rate of change of tourists from Japan to California and from Europe to New York City. The other trends were not significantly affected by the ordinances.

The regressions for Flagstaff and Mesa, Ariz, exhibited significant Durbin-Watson statistics, indicating the presence of serial correlations in the residuals. We attempted a variety of alternate models using functions of time, changes in the intercept term associated with the ordinance, or interactions between the seasonal variables and the presence of the ordinance. None of these approaches substantially changed the value of the Durbin-Watson statistics. Figure 1 suggests that the significant Durbin-Watson statistic for Flagstaff is due to a period of rapid hotel building between 1989 and 1993; the rate of change in hotel revenues before and after this period (which includes the time covered by the smoke-free ordinance) were similar. For Mesa, the significant Durbin-Watson statistic is due to the disproportionate seasonal increase in business following implementation of the smoke-free ordinance.

This study debunks the tobacco industry allegation that smoke-free restaurant laws adversely affect tourism, including international tourism. Quite the contrary, implementation of these laws is often associated with an increase in the rate of growth of tourism revenues. In the pooled analysis, the ordinances had no significant effect, one way or the other, on tourist revenues as a fraction of total retail sales or compared with the rate of change in the United States as a whole. The cities and states included in this study represent a wide range of geographic locations and types of tourist destinations, a fact that increases the confidence one can have in the generality of the results.

The result that smoke-free restaurant ordinances did not hurt, and may have helped, international tourism was surprising because of the commonly held belief that Europeans are more willing to tolerate secondhand smoke and less supportive of clean indoor air regulations than are Americans. Secret research conducted for Philip Morris Incorporated in 1989, however, shows that this belief is incorrect.⁴⁶ Philip Morris polled 1000 people in each of 10 European countries and found that smokers were more accepting of smoke-free restaurant ordinances than were Americans (Figure 3).

In our analysis of smoke-free restaurant ordinances, we include Boulder, Colo, which permits the construction of a separately ventilated smoking room. While the Boulder Environmental Enforcement Office has not done a formal survey, they reported that "actual use" of such separate smoking rooms is rare. We also included Flagstaff and Mesa, cities that allowed for the application of hardship exemptions or exceptions. The Flagstaff County Health Department reported that no such hardship exemptions have been granted. As of August 1998, the City of Mesa Code Compliance Office cited 73 (3.5%) of 2080 businesses (including smoke shops) that were granted such exceptions. Our results are based on aggregate data, not results from individual businesses. As a result, we cannot exclude the possibility that some establishments experienced gains in business that exactly offset losses in other businesses. At the same time, no data have ever been published to support this possibility. In any event, it is the aggregate data that are necessary to test the tobacco industry's hypothesis that business is severely depressed by these laws.

Food-service workers enjoy the least protection from secondhand tobacco smoke of any employee group.⁴⁷ Legislators and government officials can enact such health and safety requirements to protect patrons and employees⁴⁸ in restaurants from the toxins in secondhand tobacco smoke without the fear of adverse effects on tourism. Indeed, these ordinances may even be beneficial for business.