| The first essay of this thesis provides the first empirical study of cigarette demand using detailed individual level data that includes consumers' cigarette brand choice. By estimating and comparing the effects of advertising and price on cigarette demand for different age groups, I find that young adults are less responsive to cigarette advertising but more sensitive to price than older adults. In addition, cigarette advertising is more effective at reducing the existing smokers' incentive to quit and inducing former users to resume use than it is at attracting young people to smoke and encouraging brand switching among current users. My results raise questions about the claim of many anti-smoking groups that cigarette advertising mainly attracts young smokers.; The second essay develops an empirical dynamic oligopoly model of the cigarette industry to study the response of firms to various anti-smoking policies and to estimate the implications for policy efficacy. Modeling firms as coordinating in price and competing in advertising, this dynamic model captures two crucial features of the cigarette industry: (1) dynamic demand due to addiction; and (2) differences between young smokers and adult smokers in their responses to price and advertising. The structural parameters are estimated using a combination of micro and macro level data and firms' optimal price and advertising spending are solved as a Markov Perfect Nash Equilibrium. I find that the estimated dynamic model predicts industry prices and advertising expenditures extremely well. Moreover, the simulation results show that increasing the tobacco tax reduces both the overall smoking rate and the youth smoking rate, while advertising restrictions may backfire and actually increase the youth smoking rate. Furthermore, the long run effects of anti-smoking policies on reducing smoking rates tend to be smaller than the short run effects, as firms have stronger incentives to exploit current smokers in the short run. In addition to adding to the policy debate, this paper extends the existing class of dynamic oligopoly models to include demand-side dynamics brought about by switching costs and to allow for continuous state variables. |
