Alcohol advertising bans and alcohol abuse

Henry Saffer [Saffer (1991) Journal of Health Economics 10, 65-79] concludes that bans on broadcast advertising for alcoholic beverages reduce total alcohol consumption, motor vehicle fatalities, and cirrhosis deaths. A reexamination of his data and procedures reveals a number of flaws. First, there is evidence of reverse causation: countries with low consumption/death rates tend to adopt advertising bans, creating a (spurious) negative correlation between bans and consumption/death rates. Second, even this correlation largely disappears when the estimates are corrected for serial correlation. Third, estimates based on the components of consumption--spirits, beer and wine--mostly indicate that bans are associated with increased consumption.