presents the WLS regression results. The positive and statistically significant coefficient for the smoke-free variable indicates that restaurants that operate under smoke-free regulations have a P/S ratio that is 0.059 ±0.021 (SE) higher than in comparable restaurants not in smoke-free locations. A restaurant in a smoke-free location sold for a higher price (thus the higher P/S ratio) than a restaurant with the same sales in a smoking location. This smoke-free premium indicates that businesses in smoke-free locales operate at a higher margin (i.e., more profits).
Determinates of the Price-to-Sales Ratio (P/S) (OLS)
Of the control variables, SDCF/S, GSP growth rate, unemployment, the Fast Food and the Family dummy variables and time were also significant. The positive and significant coefficients for GSP growth and fast food are not surprising because high growth rates indicate businesses are growing and thus more profitable, and fast food restaurants are low cost high turnover establishments that tend to be highly profitable. The negative coefficient for the unemployment rate suggests that as unemployment rises restaurant values decline, presumably because as more people become unemployed, fewer go out and spend money in restaurants. The negative coefficient for the family restaurant dummy suggests that family-style restaurants are less profitable than other restaurants. The negative coefficient for time suggests that real restaurant values have been falling over time, which indicates that the market has become more competitive over time. The coefficient on per capita GSP was not significant.
The authors further tested the robustness of the model by adding a quadratic time factor. The quadratic time coefficient was not significant, and its inclusion had no effect on the smoke-free coefficient.
The authors did a Monte Carlo simulation (20,000 iterations) to estimate the additional value a restaurant owner could expect from the enactment of a smoke-free ordinance. Price and P/S ratios were randomly drawn from the 490 smoking restaurants in the sample. A random value of the increase in P/S ratio, the coefficient for the smoke-free dummy variable in the regression analysis (), was drawn from a normal distribution with mean 0.059 and standard deviation 0.021. The ratio of the smoke-free coefficient to the P/S ratio represents the expected percentage increase in value the smoking restaurant would experience should a local smoke-free ordinance be enacted (). The authors multiplied this percentage increase in value by the corresponding sale price of the smoking restaurant to determine the absolute increase in value of the restaurant (in dollars) associated with the smoke-free law (). This procedure yielded a median percentage increase in the sale price restaurants in smoke-free communities of 16% (interquartile range of 11% to 25%), corresponding to a median dollar value increase of $15,300 (interquartile range of $9,000 to $27,000) for restaurants that were worth a median of $95,000 (interquartile range $50,000 to $157,000) where smoking is permitted.
FIGURE 1 Distribution of percentage change in sales price for restaurants in smoke-free communities compared to comparable restaurants in communities that permitted smoking. The median increase is 16% (interquartile range 11% to 25%). All values over 100% are (more ...)
FIGURE 2 Distribution of dollar value of changes in sale price of restaurants in communities with with smoke-free laws over comparable restaurants in communities with no restrictions. The median increase in value $15,300 (interquartile range of $9,000 to $27,000) (more ...)
A potential source of bias in the parameter estimates is that enforcement of the smoke-free ordinance may vary across the sample. Whereas studies of the enforcement of the ordinances tend to reject this hypothesis (Cancer Prevention and Control Program, 2001; Weber et al., 2003), unequal enforcement would bias the estimates of the effects of the smoke-free policy (smoke-free in ) toward the null. The average profitability of the restaurants included in the smoke-free sample is higher than that of the control group of restaurants that do not operate under a smoke-free ordinance. Incorrectly including a restaurant that ignores a smoke-free ordinance would then have the effect of lowering the average profitability of the sample of restaurants under smoke-free ordinances, thus reducing the estimates of smoke-free. Thus, even if enforcement does vary, it is unlikely that the present findings would be adversely effected.
A similar analysis performed for the 73 bars (including 5 that were in places with smoke-free laws) revealed a positive and significant effect of 100% smoke-free bar laws on the value of the bars. The coefficient for the smoke-free dummy variable was 0.24 ±0.12 (P=0.049). The ratio of this value and the median P/S ratio for the smoking bars of 0.38 suggests that the typical bar will experience an increase of 63% in value following the enactment of a smoke-free ordinance. The sample size for bar data set was small (only 5 smoke-free bars out of 78 transactions), so this result should be considered preliminary, only.
Comparison with Previous Literature
Scollo et al (Scollo M, Lal A, Hyland A and Glantz S, 2003
) reviewed 97 studies on the economic effects of smoke-free laws on the hospitality industry available as of 31 August 2002. They assessed the quality of the studies based on four criteria: use of objective data, inclusion of all data points after the law was implemented and several years before, use of regression or other statistical methods that control for secular trends and random fluctuation in the data, and appropriate control for overall economic trend. Of these 97 studies, 21 met all four quality criteria; all 21 studies concluded that smoke-free policies had no effect or a positive effect on the hospitality industry. Thirty-five of the studies concluded that smoke-free policies had a negative impact on the hospitality industry; all of them were funded by the tobacco industry or organizations affiliated with the tobacco industry. In studies concluding a negative impact, the odds of using a subjective outcome measure was 4.0 times (95% confidence interval (CI) 1.4 to 9.6; p = 0.007) and the odds of not being peer reviewed was 20 times (95% CI 2.6 to 166.7; p = 0.004) that of studies concluding no such negative impact. All of the best designed studies report no impact or a positive impact of smoke-free restaurant and bar laws on sales or employment.
One example of the low quality studies, as defined in Scollo et al (Scollo M, Lal A, Hyland A and Glantz S, 2003
), and funded by the tobacco industry from the economics literature is the work of Dunham and Marlow (John Dunahm and Michael L Marlow, 2000) previously published in this journal. Their paper was funded by Philip Morris (now Altria), the largest cigarette manufacturer in the US and coauthored by John Dunham, “manager of fiscal issues” at Philip Morris Management. The Dunham and Marlow study presents an economic model based upon a misapplication of the Coase theorem that is then “validated” through use of data from a poll that was funded by a close ally of the tobacco industry (American Tobacco, 1990). The economic model assumes that the “interested parties” are the smoking and nonsmoking patrons of the restaurant (John Dunahm and Michael L Marlow, 2000). This definition ignores the staff of the restaurant, which has a considerable interest in the smoke-free workplace due to the health risks of SHS. Dunham and Marlow assume that negotiation costs are low because the owner of the restaurant can act as an intermediary between these two groups. In contrast to the assumptions of the Coase theorem, smokers and nonsmokers are not two well-defined distinct groups, but are rather numerous individuals with varied tolerances for smoke and willingness to refrain from smoking or to go outside to smoke. Even if the staff of the restaurant is ignored, the number of interested parties is very large with greatly varied preferences in regard to the externality. The large number of interested parties would cause negotiation costs to be high, which violates the assumption of low costs in the Coase theorem. Furthermore, the externality is not well defined in the model, because its effect on the restaurant staff is ignored. The Coase theorem is therefore not applicable.
The data that Durham and Marlow used to validate the predictions of their “economic” model (John Dunahm and Michael L Marlow, 2000), is a poll that was conducted by Roper Starch for the National Licensed Beverage Association a regular political ally and recipient of money from the tobacco industry (American Tobacco, 1990). It is a survey of restaurant and bar owners’ predictions
of the outcome of smoking restrictions, rather than hard data on actual
economic outcomes. It is not surprising that the survey found that restaurant owners were fearful of smoking restrictions because it is well documented that the tobacco industry regularly feeds misinformation to the hospitality industry to fight smoke-free ordinances (Dearlove JV, Bialous SA and Glantz SA, 2002
In contrast to the work of Dunham and Marlow (John Dunahm and Michael L Marlow, 2000), the present study has a clearly defined and simple economic model (no change to revenues plus lower costs imply higher profits) and uses objective data to estimate the effects of smoke-free policies on restaurant profitability.