Abstract
Using data from a cross-section of countries, we find to varying degrees of significance that the probability governments adopt smoking bans and youth access restrictions depends upon per capita income, per capita cigarette consumption, life expectancy, government health care expenditures, economic freedom and population density. Because some results differ from studies of US anti-smoking policy, this highlights a limitation of using US results to draw inferences about other countries.
Notes
1 This argument is consistent with studies (e.g. Hitiris and Posnett, Citation1992; Crémieux et al., Citation1999) that find health outcomes improve with income. We expect governments in countries with higher income to face greater political pressure to adopt anti-smoking measures.
2 Interestingly, since the means of the anti-smoking policies in Table 1 are noticeably different from similar policies in the United States (Gallet et al., Citation2006), this lends support to the idea that it may be inappropriate to draw inferences about anti-smoking policy adoption in other countries from results that are to date based on US data.
Table 1 Variable names, definitions and descriptive statistics
3 Studies of US anti-smoking policies (e.g. Gallet et al., Citation2006; Shipan and Volden, Citation2006) use panel data to account for state and time variation in the decision to adopt anti-smoking policies. However, since consistent panel data is not available for anti-smoking policies across countries, we are restricted to a cross-section analysis. Nonetheless, since we use cross-section data, we did adjust the standard errors using a White procedure to control for potential heteroskedasticity.