| This dissertation examines the determinants and consequences of prenatal maternal smoking. The first chapter studies the infant health consequences of maternal smoking, motivated in part by previous research which has reported surprisingly large impacts on infant health. I suggest that these results may be driven by the interaction of methodology and poor data quality. In particular, I suggest that, on birth records, maternal smoking status may be poorly (self-) reported, in which case standard instrumental variables (IV) estimates are biased. To overcome the problems generated by misclassification of a dummy endogenous variable, I propose a GMM approach based on recently developed parametric methods for misclassified binary dependent variables. The approach allows me to recover consistent estimates of the second stage coefficients, as well as of the misclassification probabilities. Accounting for misclassification yields causal estimates which are considerably smaller in magnitude than those delivered by linear IV and are more consistent with experimental estimates.;The second chapter is concerned with maternal smoking behavior itself. Since the seminal work of Becker & Murphy [1988], the theory of rational addiction has dominated economists' discussions of addictive behavior. The central empirical prediction of the rational addiction model is that future prices and behaviors affect current decisions. This chapter re-examines the best evidence to date on forward-looking behavior in cigarette smoking, provided by Gruber & Köszegi [2001]. They report robust evidence that, conditional on current and past taxes, announced but not yet effective state cigarette tax increases reduce maternal smoking. In contrast. I find estimates that are sensitive to specification, as well as to assumptions surrounding when, over the course of their pregnancies, women are responding to changes in taxes. I then consider the public policy implications of addictive behavior which isn't fully forward-looking. I develop a model of addiction to match the weaker evidence of forward-looking behavior documented here, whereby addicts make systematic errors in predicting future utility by projecting part of their current well-being onto their future selves. By not fully internalizing the intertemporal complementarities that distinguish addiction, an intra-personal externality is generated. Even in the absence health and societal externalities associated with many addictions, there is scope for government taxation of addictive goods, with the optimal tax being the one that forces addicts onto the optimal consumption trajectory of the rational addict version of themselves. |
