The New Palgrave Dictionary of Economics

Living Edition
| Editors: Matias Vernengo, Esteban Perez Caldentey, Barkley J. Rosser Jr


  • George Loewenstein
  • Scott Rick
Living reference work entry


Research on addiction had already yielded a wide range of interesting and important findings when economists first arrived on the scene. The economic study of addiction was initiated by a seminal paper by Becker and Murphy (1988) which challenged the prevailing view of addiction as self-destructive, proposing instead a ‘rational account of addiction’. Although some empirical research has confirmed the model’s critical prediction that anticipated increases in future prices will decrease current demand for a drug, more recent research by economists, stimulated by the prior work from other disciplines, has challenged some of the rational account’s assumption and predictions.


Addiction, rational account of Becker, G. Drugs Excise tax Forward price elasticity Rational behaviour Sin taxes Time consistency 

JEL Classifications


Economists were latecomers to the study of addiction, a concept which researchers in other disciplines usually define as including a loss of self-control, continuation of behaviour despite adverse consequences, and preoccupation or obsession with the substance or activity one is addicted to. Economists came late to the subject perhaps because the first two of these characteristics seem inconsistent with economists’ rational choice paradigm.

This may be exactly what spurred Gary Becker, along with coauthor Kevin Murphy, to propose, in 1988, a ‘rational account of addiction’, which stimulated much subsequent research and theorizing by economists. Although not the first economic account of addiction, Becker and Murphy’s model (referred to henceforth as B&M) was certainly the most influential, and has spawned a very lively line of research, theorizing and debate about addiction by economists.

Contributions of Disciplines Other than Economics

Prior to B&M, scientists in a range of disciplines had already developed a rich tradition of research on addiction. For example, early studies by psychopharmacologists identified the actions of addictive drugs in the brain, and subsequent research by neuroscientists has uncovered the neural pathways through which addictive activities derive their motivational power (see, for example, Gardner and James 1999; Lyvers 2000). Sociologists have also been major contributors, conducting ethnographic and life-course studies of drug users that have identified many of the social influences on drug use. Psychologists have studied the widest range of different facets of drug abuse, including biological underpinnings and social, cognitive and emotional dimensions, and have also been in the forefront when it comes to treatment. Psychologists, as well as other health professionals, have tested a great diversity of treatments for addiction, including residential treatment, counselling, psychotherapy, drug therapies such as methadone, nicotine patches and antidepressants, aversive conditioning, and hypnosis. Taken together, these diverse lines of research have yielded a number of important, and often counter-intuitive, findings.

  • Historic use of different types of drugs exhibits ‘fads’, rising then falling in popularity, sometimes repeatedly for a specific drug.

  • Most drug users do not just use a single drug, but many different drugs.

  • Many if not most drug abusers also suffer from other psychiatric conditions, such as anxiety or mood disorders, schizophrenia or antisocial personality disorder.

  • Much if not most quitting occurs outside of treatment.

  • It is not short-term withdrawal from drugs (for example, for a few days) that most addicts find difficult, but long-term abstinence, which tends to be punctuated by episodes of ‘craving’ which create an almost overwhelming motivation for drug use.

  • Episodes of craving are often triggered by ‘cues’ – people or other stimuli that the addict associates with drug use.

  • While approximately 20 per cent of a sample of veterans reported being addicted to heroin in Vietnam, and 45 per cent reported narcotic use, only one per cent remained addicted, and two per cent reported using narcotics after returning home (Robins 1973); this finding radically changed prevailing views of the incidence of recovery from heroin addiction.

  • Humans and other mammals voluntarily self-administer most of the same chemical compounds. (Hallucinogens, which some humans seek out but most animals avoid, are a major exception.)

  • Although a small number of intense users account for a large fraction of drug use, most drug users consume at moderate or low rates, and do not become addicted in the sense of losing control, suffering adverse consequences or becoming obsessed with drug-taking.

  • Many of the adverse health effects of illicit drugs, such as opiates, do not stem from physical effects of the drugs themselves, but from the difficulty of financing an illegal, and hence typically expensive, habit.

  • Most addictions begin when people are in their teens or early twenties, and addicts often ‘mature out’ – quitting when they reach middle age. People rarely become addicted for the first time in middle or old age.

In addition to generating a wide range of interesting and important findings, researchers in disciplines other than economics have proposed a variety of theoretical perspectives on addiction. Some perspectives place great importance on the pleasure of drug-taking, the pain of withdrawal, or the motivational force of ‘cue-conditioned’ craving, while others view drug use as a form of self-medication for psychiatric conditions such as depression.

For better or for worse, economists’ focus on addiction has been much narrower, at both the theoretical and the empirical levels. Most empirical work has involved estimating price elasticities of demand for drugs (often using aggregate consumption data), and most theoretical work has involved some type of generalization of Becker and Murphy’s perspective.

Becker and Murphy’s Model

In Becker and Murphy’s rational model of addiction, utility from an addictive good, c(t), is assumed to depend on consumption of that good and on the degree of addiction S(t). S(t) changes according to the function \( \dot{S}(t)= c(t)-\delta S(t), \) where the first term represents the impact of engaging in the addictive good on one’s level of addiction, and the second represents the natural decline in addictedness when one desists. The individual is assumed to trade off consumption of the addictive good against consumption of other (non-addictive) goods, discounting for time delay in the conventional (exponential) fashion. The central insight of B&M is that people treat addictive goods no differently from the way they treat any good whose utility depends on consumption over time, trading them off against other goods based on current and future (anticipated) prices.

This model can accommodate a number of features of classical addiction, such as that being addicted lowers instantaneous utility u s < 0, that it increases the instantaneous marginal utility of taking the drug u cS > 0. Solving the model yields a number of implications, most importantly that it can be rational for an individual to maintain a positive rate of consumption of an addictive good.

Empirical tests of B&M have focused on the strong prediction that anticipated changes in future prices affect the current behaviour of addicts, which is counter- intuitive given that addicts are commonly seen as behaving myopically. The model is therefore typically tested by estimating what could be called the ‘forward price elasticity’ of various addictive substances. Consistent with Becker and Murphy’s model, negative forward price elasticities have been found for alcohol, cigarettes, marijuana, opium, heroin and cocaine (for a review, see Pacula and Chaloupka 2001), although the effect appears to be more consistent for adults than for youth.

Moving Beyond Becker and Murphy

In proposing their rational account of addiction, Becker and Murphy initiated the study of addiction among economists, and made the key point that it is useful to think of addicts as solving a forward-looking optimization problem. However, the B&M model fails to incorporate a number of important features of addiction, and is either inconsistent with or fails to predict many salient features of addiction, including some of the stylized facts listed above. Responding to these limitations, economists have built upon the B&M model by relaxing some of its most extreme assumptions or incorporating more realistic assumptions that are often inspired by research in other disciplines.

One important generalization has been to examine the implications of relaxing the assumption of exponential time discounting. Gruber and Koszegi (2001, 2004), for example, propose a model in which time-inconsistent addicts have self-control problems: they would like to quit using but cannot force themselves to do so (see also O’Donoghue and Rabin 1997). As in B&M, Gruber and Koszegi’s model predicts that a rise in current or anticipated excise taxes will reduce use of addictive substances. However, although the models make similar behavioural predictions, they interpret the hedonic consequences of altered usage behaviour differently. B&M predicts that taxes on addictive substances – ‘sin taxes’ – make addicts worse off since the price of a good that they enjoy has risen. Gruber and Koszegi’s model, on the other hand, predicts that the tax makes time-inconsistent addicts better off since it provides a valuable self-control device.

Since behavioural data cannot distinguish between the models, Gruber and Mullainathan (2005) bypassed the standard practice of measuring the impact of policy interventions by estimating price elasticities in favour of directly examining the impact of these interventions on subjective well-being. They did so by matching cigarette excise taxation data to surveys from the United States and Canada that contain data on self-reported happiness. Consistent with Gruber and Koszegi’s model, Gruber and Mullainathan (2005) found that excise taxes on cigarettes make smokers happier.

Another implication of time inconsistency involves purchasing patterns. The B&M model predicts that addicts will behave in a time-consistent fashion and hence will buy in bulk to save time and money in satisfying their anticipated long-term habit. Wertenbroch (1998, 2003), however, found that consumers – even those who are not liquidity-constrained – often purchase ‘vice’ items, such as cigarettes, in small quantities in an attempt to control their intake of the harmful substance.

Other research has questioned the assumption that addicts begin drug taking with full knowledge of the consequences. For example, Slovic (2000a, b) has argued that people take up cigarette smoking in part because they underestimate the health risks, although Viscusi (2000) counters that any error is actually in the opposite direction – that smokers overestimate the health risks of smoking. Pointing to a somewhat different type of underestimation, Loewenstein (1999) has argued, based on a wide range of evidence, that potential drug users underestimate their own proneness to addiction because they underestimate the motivational force of drug craving.

Finally, a recent line of theoretical models, while also building on the insights of Becker and Murphy, has incorporated evidence from the psychological literature on cue-conditioned craving and from neuroscience. For example, Laibson (2001) proposes a model of addiction that incorporates the role of cue-conditioned craving. In his model, environmental cues that become associated with drug use, when encountered by an ex-addict, produce surges of craving (like sudden changes in S(t) in B&M). Bernheim and Rangel (2004) develop a model of addiction that is particularly closely grounded in neuroscience research and that is perhaps the most radical departure from B&M. Their model is based on the idea that repeated experience with drugs sensitizes individuals to environmental cues that trigger mistaken usage.

So far, economists are still playing catch-up with researchers in other disciplines when it comes to their understanding of addiction or their influence on policy. Thus, a large fraction of empirical research on drug use by economists has focused on price elasticities. While price is one determinant of drug use, it is arguably not the most important, or even the most amenable to manipulation through the instruments of policy. Nevertheless, economic models of addiction have made great strides, building on Becker and Murphy’s seminal contribution with new models that incorporate many of the insights and findings generated by research in other disciplines.

See Also



We thank Caroline Acker, Ted O’Donoghue and Antonio Rangel for helpful suggestions.


  1. Becker, G., and K. Murphy. 1988. A theory of rational addiction. Journal of Political Economy 96: 675–700.CrossRefGoogle Scholar
  2. Bernheim, B., and A. Rangel. 2004. Addiction and cue-triggered decision processes. American Economic Review 94: 1558–1590.CrossRefGoogle Scholar
  3. Gardner, E., and D. James. 1999. The neurobiology of chemical addiction. In Getting hooked: Rationality and addiction, ed. J. Elster and O.-J. Skog. Cambridge: Cambridge University Press.Google Scholar
  4. Gruber, J., and B. Koszegi. 2001. Is addiction ‘rational?’ Theory and evidence. Quarterly Journal of Economics 116: 1261–1305.CrossRefGoogle Scholar
  5. Gruber, J., and B. Koszegi. 2004. A theory of government regulation of addictive bads: Tax levels and tax incidence for cigarette excise taxation. Journal of Public Economics 88: 1959–1987.CrossRefGoogle Scholar
  6. Gruber, J., and S. Mullainathan. 2005. Do cigarette taxes make smokers happier? Advances in Economic Analysis & Policy 5: 1–43.Google Scholar
  7. Laibson, D. 2001. A cue-theory of consumption. Quarterly Journal of Economics 116: 81–119.CrossRefGoogle Scholar
  8. Loewenstein, G. 1999. A visceral account of addiction. In Getting hooked: Rationality and addiction, ed. J. Elster and O.-J. Skog. Cambridge: Cambridge University Press.Google Scholar
  9. Lyvers, M. 2000. ‘Loss of control’ in alcoholism and drug addiction: a neuroscientific interpretation. Experimental and Clinical Psychopharmacology 8: 225–245.CrossRefGoogle Scholar
  10. O’Donoghue, T., and M. Rabin. 1997. Addiction and self control. In Addiction: Entries and exits, ed. J. Elster. New York: Russell Sage Foundation.Google Scholar
  11. Pacula, R., and F. Chaloupka. 2001. The effects of macro-level interventions on addictive behavior. Substance Use and Misuse 36: 1901–1922.CrossRefGoogle Scholar
  12. Robins, L. 1973. The Vietnam drug user returns. Washington, DC: U.S. Government Printing Office.Google Scholar
  13. Slovic, P. 2000a. What does it mean to know a cumulative risk? Adolescents’ perceptions of short-term and long-term consequences of smoking. Journal of Behavioral Decision Making 13: 259–266.CrossRefGoogle Scholar
  14. Slovic, P. 2000b. Rejoinder: The perils of Viscusi’s analyses of smoking risk perceptions. Journal of Behavioral Decision Making 13: 273–276.CrossRefGoogle Scholar
  15. Viscusi, W. 2000. Comment: The perils of qualitative smoking risk measures. Journal of Behavioral Decision Making 13: 267–271.CrossRefGoogle Scholar
  16. Wertenbroch, K. 1998. Consumption self-control via purchase quantity rationing of virtue and vice. Marketing Science 17: 317–337.CrossRefGoogle Scholar
  17. Wertenbroch, K. 2003. Self-rationing: Self-control in consumer choice. In Time and decision: Economic and psychological perspectives on intertemporal choice, ed. G. Loewenstein, D. Read, and R. Baumeister. New York: Russell Sage Foundation.Google Scholar

Copyright information

© The Author(s) 2008

Authors and Affiliations

  • George Loewenstein
    • 1
  • Scott Rick
    • 1
  1. 1.