Industrial organization
Econometrics

Econometrics
Applied Econometrics
Industrial Organization
Microeconomics
Health Economics

May, 1997 (Oral) Industrial Organization, Econometrics
May, 1996 (Written Comprehensive) Microeconomics and Macroeconomic Theory

Empirical Essays on Rational Cigarette Addiction

Professor Steve Berry
Professor Ariel Pakes
Professor John Rust

Summer 2001

M.Phil., Economics, Yale University, 1999
M.A., Economics, Yale University, 1996
M. Comm. (Honours), Economics, The University of Melbourne, 1995
B. Comm. (First Class Honours), Economics, The University of Melbourne 1994

Yale Dissertation Fellowship, 1999
Yale Graduate Fellowship, 1995-1998
The University of Melbourne Graduate Scholarship, 1994-1995
Desmond J. Cleary Prize, 1995

Teaching Assistant, Economics of Natural Resources, Yale 2000
Teaching Assistant, Introductory Statistics and Econometrics, Yale 1997-1998
Teaching Assistant, Intermediate Microeconomics, Yale 1998
Teaching Assistant, Intermediate Microeconomics, University of Melbourne 1994

Assistant to Professor Ariel Pakes and Professor Jenny Lanjouw, Yale, 1998–1999

• Worked on project modeling patent application and renewal decisions in multiple countries at the European Patents Office using data from the OECD.
• Developed modeling and estimation framework that uses information on the countries where patent protection is sought and the length of time patents are renewed to estimate patent value.
• Part of an ongoing OECD project of developing new measures of innovations.

Assistant to Professor Oliver Linton, Yale, 1996–1997

• Worked on a number of projects involving multivariate kernel estimation, resulted in two joint papers.
• Developed a multivariate nonparametric kernel estimation procedure that overcomes the curse of dimensionality using the properties of elliptically symmetric errors.

Computer consultant, Yale Statlab, 1998-current

• Worked as a software and statistics consultant at the Yale Statlab

• "Estimation of Multivariate Time Series Models with Elliptically Symmetric Errors" (with D. Hodgson and O. Linton), Yale University, 1999, mimeo.
• "Asymptotic Approximation and the curse of dimensionality in Kernel density Estimation" (with O. Linton), Yale University 1999, mimeo.
• "Adverse Selection in the Interregional Market for Slaves," 1998, Yale University, mimeo.
• "A Topological Test of Chaos," 1995, Masters thesis submitted at The University of Melbourne (appears in Nonlinear Economic Models edited by J. Creedy. and V. L. Martin).
• Estimating a Rational Model of Addictive Consumption: The Case of Cigarette Smoking" mimeo, Yale University 2000. (Job Market Paper)
• "Modeling Rational Addiction using the Euler Equation Approach" mimeo, Yale University 2000.

There is a long history of theoretical and empirical interest in modeling addiction. In the case of smoking, much of the empirical literature takes a reduced form approach, which predicts cigarette consumption of the smoker using observed individual characteristics, the history of observed behaviors, and regulatory factors that influence the smoker's well-being. While interesting and simple to implement, this approach only allows a restricted range of policy experiments usually in the form of price changes. The principal paper of my dissertation, Rational Addiction and Rational Cessation: A Dynamic Structural Model of Cigarette Consumption, builds and estimates a dynamic structural model of rational addiction and cessation using longitudinal data from a smoking cessation study. A new methodological framework is presented that explains smoking and quitting behavior, and allows interesting and pertinent policy experiments that existing models have difficulty analyzing. More importantly, it provides a practical and intuitive way of incorporating a stochastic health state into a structural model of rational addiction. The paper explicitly models the unobserved stochastic addiction and health process. The health generating process is identified using external medical data. This model allows us to consider the effects of regulating the level of nicotine in cigarettes, subsidizing quitting behavior, finding a cure for lung cancer, and temporal changes in prices.

The paper explicitly models smoking and quitting behavior using a utility maximizing framework with uncertainty, taking into account the effects the individual’s behavior has on the addiction and health process. Addiction is modeled as an unobserved stock that accumulates stochastically. Current consumption provides instantaneous utility and also adds to this unobserved stock, which affects the marginal utility from future consumption. Health is approximated by a discrete state variable. The individual is assumed to be in one of three possible states, either high, low or an absorbing terminal state. The evolution of this stochastic health state depends of the smoker’s characteristics, his current smoking decision, and an estimate of his health status that reflects (among other things) his smoking history. The model uses information on the individual’s smoking history together with data from external medical studies to estimate the individual’s health status. In effect, this estimated health status incorporates the morbidity of smoking addiction and together with endowed individual’s characteristics becomes a source of individual heterogeneity.

Smoking increases the probability that a worse health state occurs in the future and indirectly increases the likelihood that the individual enters the absorbing terminal state sooner. This cost where continual smoking brings the smoker closer to an undesired state is factored into the smoker’s decision each period. The event of a worse health state is meant to represents a lower state of well-being or the realization that the smoker is closer to the terminal state as a result of his addiction creating an incentive to quit. This incentive is weakened by the fact that the individual gets instantaneous utility from smoking and accumulating the addiction stock. The incentive to quit also varies according to the smoker’s characteristic. The paper demonstrates through Monte Carlo simulation that this addition to the rational model of addiction is important in explaining smoking and quitting behavior.

In Modeling Rational Addiction using the Euler Equation Approach, I consider an alternative estimation strategy that employs a perturbation approach to generate moment conditions for the model of rational addiction. The sizeable empirical literature that uses Becker and Murphy’s (1988) seminal model of rational addiction typically ignores the binding constraint that quitting behavior places on the first order conditions. This paper explicitly takes this constraint into account and relaxes the functional form restriction used in the preceding paper. In particular, it allows for addiction to have a tolerance effect on the smoker. Like the preceding paper, the entities of interest are the primitives of the model. This is work that is currently in progress.