This set of evidence paints a compelling case that many decisions, and in particular smoking, are not time consistent. This is particularly true when one acknowledges that there is absolutely no evidence, experimental or otherwise, for time consistent preferences as a better explanation for laboratory or real world phenomenon than are time inconsistent preferences.
But none of this evidence meets the gold standard for economics testing, which is to test the hypothesis through revealed behavior in response to a real change in the underlying economic environment. Unfortunately, that may be a standard that is almost unreachable given the similarities between the models. Thus, whether higher cigarette taxes will make smokers better or worse off remains an open question.
Understanding the impacts of cigarette taxes on well-being is not simply a matter of intellectual curiosity; these different models also have radically different implications for government policy. Under the rational addiction model, there is no rationale for government regulation of addictive bads other than interpersonal externalities. Just as the government has no cause, absent market failures, for interfering with revealed preference in the realm of non-addictive goods, there is no reason to take addictiveness per se as a call to government action, if individuals are pursuing these activities rationally.
It is this framework that implicitly underlies the well-known efforts of Manning et al. and others to measure the external costs of cigarette and alcohol consumption. These estimates, which are frequently cited and influential in debates over excise taxation, suggest that the optimal tax rate for cigarettes in particular is fairly low, since the net external costs of smoking are small. In particular, most estimates of the externalities from smoking are well below the existing average level of excise taxation (Gruber and Koszegi, 2002). birthcontroltab.com
Gruber and Koszegi explore in detail the implications for government policy of a introducing time inconsistent, quasi-hyperbolically discounted preferences into the Becker-Murphy framework. In the Gruber and Koszegi model, the optimal excise tax is greater than zero even absent externalities, due to the self-control benefits to time inconsistent agents. Calibrations show that this point is not a theoretical curiosity, since the “internalities” (damage to the smoker himself) of smoking are so large, at over $35 per pack when accounting for mortality effects alone (calculated using the impacts of smoking on length of life and standard estimates of the value of a life from Viscusi, 1992). They find that the optimal tax in their model, even with very modest time inconsistency, is well over $1 per pack, above and beyond externalities.
It is important to note that not all alternatives to the rational addiction model deliver the prediction that smokers will be made better off by higher cigarette taxes. For example, in the temptation models of Bernheim and Rangel, agents do not behave in a rational time consistent fashion; they have different preferences over “tempted” and “untempted” states. But there is by definition no price elasticity in the “tempted” state, so that higher prices serve no selfcontrol purpose; thus, higher prices only make them worse off.
Similarly, in the model of Gul and Pesendorfer, there is a direct disutility from being tempted; but, so long as the agent can afford the product which is tempting them, there is no reduction in this disutility from higher prices. Even in the model of Gruber and Koszegi, time inconsistent but naive consumers, who have a self-control problem but don’t recognize its existence, would not be made better off in their own eyes by higher taxation; such consumers view themselves as time consistent, so that by the same logic as above they would feel worse off from a tax-induced price rise. Social welfare, discounted exponentially, may rise when taxes increase on naive hyperbolic smokers, but their own perceived welfare will not increase.