THE DURATION OF THE SMOKING HABIT
| Autor: | Stratford Douglas |
|---|---|
| Rok vydání: | 1998 |
| Předmět: | |
| Zdroj: | Economic Inquiry. 36:49-64 |
| ISSN: | 1465-7295 0095-2583 |
| DOI: | 10.1111/j.1465-7295.1998.tb01695.x |
| Popis: | I. INTRODUCTION Most smokers smoke out of habit. Once the habit is ended, many of smoking's most harmful effects decline rather quickly. Preventing and ending cigarette addiction is, therefore, the straightest path toward preventing the health damage done by cigarette smoking. Consequently, the related questions of why people start smoking and why they quit are central to formulating and evaluating effective antismoking policies. Duration, or hazard, models are designed to measure the probability of transition between states, and are thus ideal for investigating the structural determinants of the transition between addiction and non-addiction. Nevertheless, most previous work in the economics literature uses the aggregate smoking participation rate (defined as the percentage of people who currently smoke) or an individual's probability of participation, rather than a hazard rate, to measure the attractiveness of habitual smoking. This paper models the hazard rates of both starting and quitting the cigarette smoking habit. The starting hazard rate is the probability that an individual will start smoking in a given time period, conditional on being a nonsmoker until then. The quitting hazard rate is the probability of quitting smoking in a given time period, conditional on being a smoker since the starting date. Both of these rates should rise and fall in response to changes in the attractiveness of smoking. The two rates will respond differently to different stimuli, since the starting hazard reflects the point of view of nonsmokers, and the quitting hazard reflects the point of view of smokers. Hazard or duration models possess several advantages over smoking participation models. Most obviously, smoking participation rates cannot directly distinguish between increased starting rates and decreased quitting rates, but the hazard models can. Also, in a participation rate model it is difficult to control for the effect of the duration of the habit, whereas duration models are designed specifically to allow for and measure the dependence of the hazard rate on duration. Another advantage is that the participation rate at any one time will oversample the number of long-term smokers relative to those who had a short-term habit, whereas a duration model is designed to avoid this source of bias. One drawback, however, is that because of their inherent panel nature, the data requirements of duration models are somewhat more difficult to satisfy, as shall be seen. The only previous work in the economics literature that estimates the hazard rate of starting smoking is Douglas and Hariharan [1994]. My paper improves upon Douglas and Hariharan in at least three important ways. First, it estimates the hazard rate of quitting smoking, as well as that of starting. Second, it takes into account variation in state-level regulation. Third, its time-varying covariate model allows hazard rates to respond dynamically to changes in cigarette price, regulation, and information variables. The use of time-varying covariates is important because it brings the empirical model into much closer accord with theory. The decision whether or not to start (or quit) smoking is an ongoing decision, made "one day at a time," on the basis of current information. The hazard of starting is greatest when the cost of the habit appears lowest and its benefit appears highest. This comparison of cost and benefit changes through time as price, information, income, and life cycle variables change. While Douglas and Hariharan [1994] used price information from the potential smoker's most vulnerable age on the theory that its marginal effect is greatest at that time, the model used in this paper allows each individual's hazard rate to respond to changes in price and public information as they occur. The leading theoretical model of the decision to smoke is the rational addiction model of Becker and Murphy [1988], which assumes that consumers of addictive goods begin smoking only after fully weighing the costs and benefits of the habit. … |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|