Addiction and Health

Policy Implications Of Addiction Perspectives

The extent to which people operate in the perfectly rational framework of Becker and Murphy has important normative implications that impact policy. Under the assumptions of the perfectly rational framework, people consume addictive goods according to their individual preferences and policy interventions are welfare improving only to the extent that they account for externalities associated with addictive consumption. For example, policy to reduce alcohol consumption is only welfare improving to the extent that it reduces externalities (involuntary benefits or, here, costs imposed on third parties), such as traffic accidents and violent crime.

However, even small departures from perfect rationality may imply a greater role for policy (Laux, 2000; Suranovic et al., 1999). Policy intervention can be welfare enhancing when people have incorrect or insufficient information, or if the decision-making process is in part driven by irrational behavior such that ‘internalities’ (costs a person imposes on their future self as a result of irrational behavior) result. However, the specific type of policy intervention that should be implemented depends to a large extent on the model of consumer behavior. Further, it should be cautioned that policies designed to correct internalities are by definition paternalistic and hence controversial (Viscusi, 2002).


One oft-suggested tool for intervention policy is taxation. There are sound reasons to tax addictive goods that do not hinge on their addictive property. The external costs of some addictive goods, such as second-hand smoke from cigarettes, can and should be internalized with taxes. Generating government revenue by taxing inelastically demanded goods creates fewer market distortions than taxing goods with elastic demand. Therefore, addictive goods with inelastic demand should be heavily taxed for revenue creation. These arguments do not rest on improving the welfare of potential addicts per se.

Addiction itself has no clear-cut implication for tax policy because different models generate different optimal tax policies. For example, if people have time-inconsistent preferences , such as in hyperbolic discounting models, or incorrectly forecast utility with a present bias, then the optimal tax will be higher than those predicted by perfectly rational models of addiction with only externalities (Gruber and Koszegi, 2001; Levy, in press). Present bias and utility-projection bias mean that people place too little importance on, or systematically misjudge, how current behavior will impact their future selves. Therefore, taxes on addictive goods can enhance welfare by forcing people to internalize the impact of their current behavior on their future selves. In a simulation of their hyperbolic discounting model, Gruber and Koszegi (2001) estimate that a tax of at least US$1.00 per pack of cigarettes should be applied to correct the present bias in discounting. With both a utility projection bias and a present bias in discounting, Levy (in press) estimates that an optimal corrective tax should be set considerably higher.

Not all economic models of addiction imply corrective tax policy to improve the well-being of addicts and potential addicts. In the temptation model of Gul and Pesendorfer (2007), individuals optimally consume the addictive good given the temptation they face and their ability to commit to future consumption. In this framework, tax policy, when used alone, is always welfare reducing: A tax increases the cost of consuming the addictive good but does not remove or reduce temptation. Likewise, in the cue-triggered decision-making model, Bernheim and Rangel (2004) find that taxation of addictive goods may be harmful, as it may do little to change the consumption behavior of addicts and instead crowds out consumption of nonaddictive goods. Even if taxation is beneficial, Bernheim and Rangel find that banning consumption of the addictive good may be a superior policy to taxation.


Bans and restrictions are perhaps the most commonly used policy intervention with respect to addictive substances. Many models of imperfect rationality and irrational behavior predict that bans can be welfare improving. Gul and Pesendorfer (2007) show that prohibitive policies are always welfare improving because they limit the opportunity to make addictive consumption choices, thereby reducing temptation. A partial ban, say in the workplace but not at home, is considered by de Bartolome and Irvine (in press) who model the short-run behavior of an addict. The addict likes higher overall consumption of the addictive good but dislikes variance in consumption throughout the day. The workplace ban reduces daily consumption through the addict’s dislike of variance; reductions in workplace consumption of the addictive good are not fully reallocated to consumption at home. These models, however, are not designed to evaluate the overall implications of prohibitions and, in particular, do not attempt to assess unintended consequences of prohibitions (Miron and Zweibel, 1995) nor the operation of black markets (Lee, 1993), so these policy implications must be considered as only part of a much larger story.

Partial bans in the form of controlled distribution offer another policy instrument. In a cue-triggered model of behavior these can be used to improve welfare. Specifically, when distribution is controlled in such a way that addicts are forced to ‘stock-up’ in cold states, rather than make purchases as hot states arise, they will choose the optimal level of consumption for their future selves. Partial bans allow the cold state decision maker to commit to hot state consumption. Such policy could potentially be achieved through the use of prescriptions or time-specific restrictions on sales.

Information And Insurance

When people lack information about their susceptibility to addiction, public provision of accurate information about addictive goods can enhance welfare (Orphanides and Zervos, 1995). Further, continued research and dissemination of information on the assessment of individual risk with respect to addiction can be welfare enhancing, even when people know the true distribution of risk across the population. Such efforts will assist people in better assessing their uncertain susceptibility to addiction. The need for accurate information also means that there is a welfare case to be made for restricting misleading advertising campaigns (Orphanides and Zervos, 1995). Similarly, limiting cue use in advertising for addictive goods is potentially beneficial (Bernheim and Rangel, 2004).

When uncertainty exists about susceptibility to addiction or the environmental cues which an individual will face in the future, there is an opportunity for a welfare-enhancing policy intervention through insurance provision. This insurance may come in the form of subsidization for rehabilitation and withdrawal treatment. It should be noted that the moral hazard and asymmetric information problems that accompany this market are nontrivial (Orphanides and Zervos, 1995).

Finally, Tomer (2001) argues that even with full information addicts may incorrectly weigh the costs and benefits associated with their behavior, placing too little weight on the negative consequences of their actions. In this way, continued addiction may be the result of systematic mental accounting of errors in which the addict places too little weight on the potential loss of family, friends, and other forms of social capital, and too much weight on the immediate cravings associated with addition. In this case, interventions by family and friends, to make the benefits of abstinence salient, will be welfare improving. Such interventions are commonly used in cases of severe addiction.

Abortion and Health
Advertising as a Determinant of Health