Marijuana creates abuse and dependence at higher rates than does alcohol.

Jonathan P. Caulkins writes at National Affairs,

Marijuana might better be described as a performance-degrading drug and, more dangerously, as a temptation commodity with habituating tendencies.

Trying marijuana is not dangerous, but using it is. Those who use marijuana on an ongoing basis face a much higher likelihood of becoming dependent than lifetime smokers do of developing lung cancer. Marijuana dependence is neither fatal nor as debilitating as alcoholism, but it is real, harmful, and far more common than is generally acknowledged.

Second, marijuana use is highly concentrated among the growing minority who use daily or near-daily. Adults who use fewer than ten times per month and who suffer no problems with substance abuse or dependence account for less than 5% of consumption. More than half of marijuana is consumed by someone who is under the influence more than half of all their waking hours. Most marijuana users are healthy; most marijuana use is not.

In the resulting confusion, advocates of legalization often argue (effectively) that "marijuana is safer than alcohol." It would be far more accurate to say "Marijuana is safer than alcohol, but it is also more likely to harm its users."

...it is essential that policymakers, public-health officials, and the larger public get beyond the lobbyists' talking points and gain a clear sense of the harms of marijuana.

...The custom is to define "current" users as those who report consumption within the last 30 days. The surveys estimate that 20 million Americans admit to past-month use. Another 13 million report use within the last year, but not the past month.

The surveys ask about problems users believe they have experienced because of their use. The Substance Abuse and Mental Health Services Administration consolidates the answers to estimate how many would meet the clinical criteria for dependence (2.8 million) or abuse (an additional 1.4 million) set down by the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (commonly referred to as the DSM-IV). SAMHSA makes these calculations for all substances in an effort to estimate the nationwide need for treatment for each substance.

Since 4.2 million people are estimated to meet the criteria for marijuana abuse or dependence, there is one such person for every 4.8 current users. Or, expressing the ratio the other way around, 21% of current users meet diagnostic criteria.

...It turns out that 137 million people self-report current alcohol use, and 17.3 million describe enough problems to meet the criteria for alcohol abuse or dependence, equivalent to one instance of alcohol abuse or dependence per 7.9 current users, or 13% of current users — significantly lower than the corresponding ratio for marijuana.

In all likelihood, both the 21% for marijuana and the 13% for alcohol understate the true extent of the problem, because denial is a hallmark of addiction. But if the rate of under-reporting problems among those who do admit to use is comparable for both substances — which is plausible because we're talking about reporting of problems by those who admit use, not the reporting of use in the first place — then the ratio of those ratios (0.21/0.13 = 1.62) may be a fair quantification of how much more common abuse and dependence is with marijuana than with alcohol. Thus, marijuana may generate about 62% more abuse and dependence per current user than alcohol does. If one focuses on just the more serious diagnosis of dependence, a little over 14% of past-month marijuana users meet the criterion for dependence, compared to only a bit under 6% of past-month alcohol users — meaning that marijuana appears to generate not just 62% but 133% more dependence per current user than alcohol.

...But what can be said empirically is that, within the context of aggregate use in the United States at this time, the best available data suggest that marijuana creates abuse and dependence at higher rates than does alcohol.

...While alcohol is more dangerous in terms of acute overdose risk, and also in terms of promoting violence and chronic organ failure, marijuana — at least as now used in the United States — creates higher rates of behavioral problems, including dependence, among all its users.

...Most people who use marijuana also drink alcohol, and the two are often used together. The surveys ask what other substances were consumed the last time the respondent drank. Among current marijuana users, 44% report using marijuana with alcohol the last time they drank, a proportion that rises to 67% among daily and near-daily marijuana users.

...Marijuana has four defining characteristics that make dealing with it difficult from a policy standpoint. First, it is a performance degrader. Advancing pharmacology may well challenge society in the not-too-distant future with the issue of what to do with cognitive enhancers — an Adderall that actually works — but marijuana is not such a substance.

Second, it is dependence-inducing. Marijuana is not crack, but marijuana dependence is nonetheless a real and not-uncommon consequence of prolonged use. This challenges the usual free-market presumption that consumers reliably maximize their welfare, particularly given that the vast majority of users start using before they are adults.

Third, marijuana is, for the most part, not directly harmful to third parties. While impaired driving and workplace use are concerns, and 1 million children live with a parent or guardian who meets diagnostic criteria for marijuana abuse and dependence, most of marijuana's direct harms fall on its users, and the families and friends who care about them.

And fourth, its health harms are, for the most part, minor. Yes, marijuana smoke contains carcinogens, and, despite vehement denials by legalization proponents, the evidence suggests that marijuana can trigger mental-health problems, including psychotic episodes. But the scale of those harms per unit of use does not distinguish it from other permitted recreations, including skiing and sky diving.

...Regardless of whether legalization is good or bad policy, it is certainly not a cause for jubilation. Borrowing again from Mark Kleiman, choosing legalization over prohibition or vice versa just trades one set of problems for another. Choosing prohibition means choosing black markets; choosing legalization means choosing greater drug dependence. It is trite but true: A country can choose what kind of drug problem it wants, but it cannot choose not to have a drug problem.

...With regard to drugs and other activities (gambling, for instance) that are not dangerous enough to merit banning outright, Mark Kleiman has argued for "grudging toleration." That means allowing adults access to some legally produced supply, hopefully on liberal enough terms to undermine the black market, but with restraints and hoops for users and suppliers to jump through that will be seen as features of the regulatory regime, not wrinkles to be ironed out. For example, Kleiman suggests that even adults should be required to pass a test to earn the right to use, the way one must earn the right to drive, rather than conferring that right universally upon reaching a certain age. And such licenses could be used to enforce limits on amounts purchased per week or per month.

Marijuana is likewise no ordinary commodity but a temptation good that society should tolerate grudgingly. There are many ways of putting that philosophy into practice. One way is to start by restricting production and distribution to non-profits or for-benefit corporations whose charters mandate that they merely meet existing demand, not pursue unfettered market growth to maximize shareholders' returns and owners' wealth. It would also be wise to require these organizations' boards to be dominated by public-health and child-welfare advocates. Furthermore, regulatory authority should be put in the hands of agencies like the FDA whose loyalties are to the public welfare, not industry, and who maintain a healthy suspicion toward industry motives and practices.

Read more here.