Despite its well known effects on health, smoking remains quite common in the United States. In order to devise anti-smoking measures it is essential to understand the long-term impact of current policies such as taxes. The goal of this research proposal is to evaluate possible explanations for the persistence over time of individual smoking behavior and to explore the resulting implications for policy analysis. One theory is that smoking is addictive and that persistence reflects behavioral or physical modification. Alternatively, persistence could be the result of individual-specific propensities to smoke: some people have characteristics which lead them to smoke while others have preferences to never smoke. Only if addiction is important will tax increases have significant effects on long-run smoking rates. We propose to develop a dynamic empirical model of smoking decisions which explicitly accounts for the impact of previous smoking behavior. After controlling for individual smoking differences in a very general manner, allowing both observed and unobserved individual differences, we will be able to estimate the empirical importance of smoking addiction. The framework will applied to two groups, teens and adults. Teens are important to study because this is the typical age of smoking initiation while adults may evidence strong addiction due to longer smoking histories. We will utilize the restricted use versions of the National Education Longitudinal Study of 1988 (NELS:88) and the Russia Longitudinal Monitoring Survey (RLMS) which contain data on the two groups. These are appropriate data-sets because they follow the same individuals over time, contain detailed smoking questions, and have a wealth of information about individual, family and community characteristics. The resulting parameter estimates will be used to simulate individual behavior following policy changes such as tax increases of various sizes and durations. To understand the importance of addiction, the estimates will also be used to simulate long-run behavior when all individuals are initially forced to smoke. Finally, this methodology will be extended to study the importance of alcohol addiction in the two data-sets. These results will be relevant for understanding the salience of public policy in reducing excessive drinking.
