Thursday 3 October 2013

The addictive choice

Maybe the Becker-Murphy rational-addition model isn't so unrealistic after all.

Becker and Murphy's model shows that even rational, forward-looking, fully informed individuals can choose to consume addictive substances, knowing and accepting that it will result in addiction. It isn't meant to describe each and every addict, but if we can show that addiction can be consistent with rational choice, we cannot presume that all addicts are irrational. Further a lot of the subsequent empirical work finds results consistent with rational addiction. After a price change, the rational addiction model predicts larger long-term than short-term consumption changes; this prediction has been confirmed in the literature. It's a sharp contrast to some policy work around addiction that assumes agency away.

Sally Satel at The Atlantic surveys recent work by Carl Hart:
...Carl Hart, a neuroscientist at Columbia University, who has been showing that cocaine and methamphetamine addicts have a lot in common with Powell. When Hart’s subjects are given a good enough reason to refuse drugs—in this case, cash—they do so too.
The basic experiment goes like this. Hart recruits addicts who have no interest in quitting but who are willing to stay in a hospital research ward for two weeks for testing. Each day, Hart offers them a sample dose of either crack cocaine or methamphetamine, depending upon the drug they use regularly. Later in the day, they are given a choice between the same amount of drugs, a voucher for $5 of store merchandise, or $5 cash. They collect their reward when they’re discharged two weeks later.
More often than not, subjects choose the $5 voucher or cash over the drug, except that, when offered a higher dose, they go for the drug. But when Hart ups the value of the reward to $20, addicts chose the money every time.
In his new book, High PriceA Neuroscientist’s Journey of Self-Discovery That Challenges Everything You Know About Drugs and SocietyHart reports that he was surprised by his findings. Wasn’t addiction a dopamine-driven compulsion “that ’hijacked’ the brain and took control of the will?” he asks. As a graduate student Hart was taught that. It's understood that recovered addicts eschew substances for fear that even a small amount could set off an irresistible craving for more. 
Satel also describes the array of self-control mechanisms available to help addicts avoid consumption urges when it's important to avoid consuming.
Yet there’s room for deliberate action in the form of “self-binding,” a practice by which addicts can erect obstacles between themselves and their drugs. Examples include avoiding people, places, or things associated with drug use; directly depositing paychecks or tearing up ATM cards to keep ready (drug) cash out of one’s pockets; or avoiding boredom, a common source of vulnerability to drug use.
Her discussion here reminded me of Jon Elster's older work on the topic.


  1. Addicts respond to incentives and make choices, but that has (or should have been) known for a long time. My impression is that economists back before Becker thought of addiction as a "change in preferences." That makes it hard to compare welfare, since you have to decide whether to use the non-addicted or addicted set of preferences and so on. Even if you rationally took the risk of addiction into account, though, you wouldn't be a Beckerian rational addict:

    The new thing in rational addiction theory was that Becker and Murphy derived unstable consumption from stable preferences under full information (later work by others introduced uncertainty), which they did by introducing an "addictive stock" that summarizes past consumption history and which the addict manages rationally. Their (to my mind not very credible) claim is that addicts show increasing consumption as their addiction takes hold because they are implementing a forward-looking (not necessarily conscious) investment plan that takes the stock dynamics into account. The point is that this optimal plan is all the addiction is.

    So: people taking risk of addiction into account seems reasonable, addicts still able to respond to incentives and still attempting to make the best of their situation seems reasonable. Smokers and heroin junkies displaying unstable consumption (of the kind we call addictions) because they have solved and are implementing an optimal forward looking consumption-and-investment plan - not so much...

  2. Yup. I spend a week on the Becker-Murphy model and draw out the graph in my 200-level class. Main point is that evidence of addiction isn't necessarily evidence of irrationality, not that all addicts are necessarily behaving that way.