The specific aims of this proposal are to estimate the effects of cigarette advertising bans and advertising expenditures on smoking. This proposal outlines four projects which will examine the effects of cigarette advertising. Projects 1 and 2 use international data sets and projects 3 and 4 use US data sets. New empirical studies of cigarettes advertising should employ advertising data that have a large variation since this is more likely to be associated with a significant variation in consumption. Two empirical strategies designed to generate a significant variation in the advertising data will be employed. The first strategy is to construct data sets which include both places with cigarette advertising and places which ban cigarette advertising. This strategy will be implement with two international data sets. The first project employs an international micro data set. The special advantages of this data set are the extensive cigarette health knowledge questions and the freedom from endogeneity problems between advertising and consumption. The second project employs an aggregate international data set. The advantages of this data set is that it includes 22 countries with very diverse advertising policies. The second empirical strategy is to use data sets with a significant amount of cross sectional variation in the advertising data. This strategy will be implemented with two US data sets. The third project employs a data set of high school seniors who were reinterviewed every other year for a period of 10 years. The special advantages of this data set is its panel nature which allows for estimating lagged effects of advertising and rational addiction models. The data set employed in the fourth project is a large scale micro data set from the 1987 National Health Interview Survey: Cancer Control Study. The advantage of this data set is its size and the age distribution of the subjects. The NHIS is also large enough to extract reliable subsamples of women and minorities. One advantage of using four very diverse data sets is that consistent results are mutually validating and inconsistent results can be helpful in identifying data and econometric problems.
|Saffer, H; Chaloupka, F (2000) The effect of tobacco advertising bans on tobacco consumption. J Health Econ 19:1117-37|