The influence of advertising on tobacco consumption: some problems with Chetwynd et al.'s analysis

In a recent study of the relationship between cigarette advertising and the aggregate consumption of cigarettes in New Zealand between 1973 and 1985, Chetwynd et al. (1988) argue that quarterly data suggest that advertising affects overall consumption of cigarettes with an elasticity of +0.07. In addition, they argue that advertising has a 'carry over' effect of about four quarters on current consumption. These results are potentially important for two reasons. Although the evidence is mixed, the conventional view is that cigarette advertising affects brand choice among smokers but not aggregate demand for cigarettes. Chetwynd et al.'s results, (hereafter, Chetwynd) contradict this traditional view. Secondly, if advertising does increase the aggregate demand for cigarettes, then a public policy banning cigarette advertising might reduce aggregate demand for cigarettes. Unfortunately, the Chetwynd, study is sufficiently flawed with conceptual and econometric problems that their inference that advertising increases cigarette demand is questionable. Certainly cigarette advertising may increase or decrease cigarette consumption. The point we wish to make is that, any inference one way or the other based on the results of Chetwynd, cannot be viewed as well-grounded in either scientific methodology or statistical principles.