Wednesday, June 3, 2009

Marijuana: The "estimated" cost savings from legalization

I have been writing on the legalization debate because I think the topic is interesting and because the marijuana debate is so messy. My comments have added value to the marijuana debate (which gets emotional) because I write as someone who just wants to weigh the merits of each side.

As I'm still trying to make up my mind on the issue, I have been reading various perspectives on the topic. From my readings, I conclude that it is hard to know what to trust. The problem is that none of the available information in this debate is objective, even the purported "scientific studies." This applies to both sides of the debate. When an onlooker like me wants a statement of the facts, he finds a report produced by an academic. That's what I do in this article.

In this post, I analyze the report, "The Budgetary Implications of Marijuana Prohibition," the report that claims that legalizing and taxing marijuana would net the U.S. Government nearly $14 billion in savings and revenue. When it was published in 2005 by Harvard economist Jeffery Miron, the report was endorsed by 500 economists. As a budding economist, I wanted know how Miron obtained these numbers, so I read the report. I'm nerdy like that.

I'm glad I read the report because after looking at its methodology (and data), I can definitively say that I don't believe the cost estimates. Not one bit. Let me explain why.

Cost methodology. The report generated the cost estimates by taking two steps (see Section II, page 5, paragraph 2):

1. Estimate the annual cost of marijuana prohibition (categories: incarceration, policing, courts).
2. Take the entire current cost of prohibition as the cost savings if marijuana were legalized.

As an economist, I object. The second step implies that dealing with legal marijuana has a budgetary cost of zero. Marijuana may cost less to control when it is legal, but there's no way it costs nothing. I read the whole section on estimating costs, discovering nothing deeper in the report's methodology. This flaw severely overstates the cost savings of legalization. It troubles me that there is no serious discussion of the costs of regulating legal marijuana.

The most direct mention of these other costs is the following dismissive statement (page 2 of the report):

If marijuana were legal, enforcement costs would be negligible and the government could levy taxes on the production and sale of marijuana.
In a decriminalized system with a legal age, the cost of controlling underage use is significant, not negligible. If you do not believe this, just look at the resources devoted to policing minors in possession of tobacco or alcohol. Clearly, these costs are not zero and they can be quite high.
Furthermore, it is not costless to deal with the marijuana problem in a less abrasive way: even Portugal's decriminalized system has budgetary costs. In Portugal, when someone is found with marijuana, they’re referred to a panel of medical experts, which recommends a variety of options. Portugal’s policy appears to be effective, but it is not zero cost. And, the whole point of Miron’s report is that the government can save money by legalizing marijuana.

To sum up, a reliable estimate of cost savings would subtract post-legalization budgetary cost from our current budgetary cost. The Miron report trivializes the after-legalization expenditures, and as a result, produces an unreasonably high estimate of the cost savings.

Why might the budgetary cost of legal marijuana be high? More minors will use marijuana in a legal scheme with an age cutoff because there is less chance of being punished for dealing to minors under a legal scheme (at least minors will use more intensively, see a more detailed argument here or here). With more marijuana use among minors, we should expect the expenditure on policing minors in possession of marijuana to increase, not decrease.

Depending on the fraction of users who are minors, this effect could either increase or decrease how much we spend on controlling marijuana upon legalization for adults. After all, any legalization bill on the radar prohibits marijuana use for someone under some cutoff age. Realistically, we're not talking about legalizing pot for first graders.

I am not sure whether to expect an increase or a decrease in expenditures, but one thing is for sure: the estimates in this report make no mention of these effects. Even worse, as I stated above, the report does not budget for any expenditures post-legalization. And, that omission concerns me.

Next Wednesday, I will discuss other lingering issues I have with the report. In particular, I have yet to talk about the tax estimates, but there are some important statistical/data issues.

Note: I modified this post from its original version. In particular, the original version made claims about Miron's motives. As those claims were unnecessary and detracted from my point, I took them out. My main point deserves to stand on its own -- i.e., we really should account for the costs of a legal scheme.


  1. Your entire premise is based on your assumption that "More minors will use marijuana in a legal scheme with an age cutoff because there is less chance of being punished for dealing to minors under a legal scheme". This is faulty logic.

    Your assumption is false. Fewer minors will use marijuana when it is age-restricted. It will be harder for them to obtain, plus it will lose its' "forbidden fruit" image that youth looking to rebel desire.

    Legalization eliminates the profit in dealing the drug (and remember that it's the drug dealers who do not check ID).

    How about you cite some examples of these increased costs that you mention, and how they will be additive to current costs?

  2. Interesting point. I have two responses.

    First, my premise is that "if there are any remaining costs to policing underage use, the cost estimate in the report is an overstatement." That doesn't require that the cost of policing minors increases. It just requires that there is some cost.

    Second, I suspect both the monetary price of marijuana and its non-monetary cost (risk of punishment) to be lower in a legal scheme.

    The law of demand predicts that when the price (monetary plus non-monetary) falls, we see more use. This leads to two scenarios:

    One where minors use more. If this is the case, we will likely have to see some response by local law enforcement (because it would not be legal for underage users). That response has costs (more cops, more police cars, more K9 units, etc), and these costs increase with the use.

    In the second scenario, minors use less pot when it is legal. Presumably, this is because the government enacted some other intervention that actually made the price minors pay for pot increase. This intervention policy itself has costs.

    No matter where we look, it's costly to deal with marijuana use among high school kids.

    Lastly, I don't see how it will be harder for minors to obtain when pot is legal. Maybe someone can explain it to me. Suppose the legal age is 18. 18-year-old kids have younger friends, but they would be allowed to buy the drug legally and at a lower price.

    In a legal scheme, what prevents these kids from buying for their friends? What about fake IDs? These are just some of the ways that it will be easier to obtain.

  3. You say: "First, my premise is that "if there are any remaining costs to policing underage use, the cost estimate in the report is an overstatement." That doesn't require that the cost of policing minors increases. It just requires that there is some cost."

    This contradicts your problem with the statement "2. Take the entire current cost of prohibition as the cost savings if marijuana were legalized."

    If there are no increased policing cost, then the statement is true.

    Secondly, even if your assumption were true, you fail to consider the relevance of scale. Say these costs you speak of represent 1/1000th of a percent of the costs of prohibition. Would your point still be relevant?

    You also say "Lastly, I don't see how it will be harder for minors to obtain when pot is legal."

    You are failing to recognize the current reality. Under our current (failed) system of prohibition, it is easy for minors to get pot; easier in many cases than alcohol. According to national data compiled by the University of Michigan, 86 percent of 12th graders say that marijuana is ʺfairly easyʺ or ʺvery easy to get.ʺ

    "Suppose the legal age is 18."
    Bad supposition. The age would be 21, like alcohol.


  4. Dave,

    Thanks for your comment. Unfortunately, you misunderstand my point on #2.

    I'll restate the point to make sure it is clear:

    The report counts all major costs of marijuana prohibition: current police effort on marijuana, judicial costs, incarceration costs.

    BUT, there's a problem with saying that legalizing marijuana above some cutoff age saves on all of these costs. The problem is that there is still some enforcement and judicial costs for minors, even if we legalize.

    Even if we save on incarceration, not all of these costs go away. Yet, that is exactly what the report assumes in its calculations.

    On the second point, you're right that we should be worried about scale. I'm not sure what fraction of marijuana prohibition costs are borne prohibiting for the under 21 group, but it is worth looking into.

    Although I don't know the age distribution of use, let's think about the scale of these effects.

    Suppose that the under 21 group (who would be prohibited under your cutoff) currently makes up 40 percent of the enforcement and judicial costs of marijuana. Even in a world where we treat under 21 the same as we currently do, this means that we still have to pay 40 percent of the judicial and enforcement costs.

    In the context of the report, that means the report is an overestimate of about $2.4 billion dollars without increasing use among the under 21 group.

    Now, suppose that the price decreases by 50 percent. The report itself cites numbers that imply a 50 percent decrease in the price leads to 25-50 percent more use (elasticity of between -0.5 and -1 for you econ people).

    Then, if dollars of expenditure are proportional to the increase in use, we should expect another $600 million to $1.2 billion in enforcement and judicial costs.

    To be fair, this is a back of the envelope calculation using the numbers from the report. But the point is that these are not small numbers, but they were omitted from the report.

    That's an interesting point that marijuana is deemed "fairly easy" to "very easy to get." The real question, however, is whether it will become easier or harder to obtain if we legalize it.

    If someone (whether they be 21 or 18) can obtain the drug legally, I highly doubt it will be harder to obtain pot if it is legalized. Making the drug legal makes it easy to be a supplier to minors. That's why I think we'll see an increase in use among minors -- they'll have more willing suppliers (even if they get it from friends, someone has to get it initially, and that initial cost drops)

    And, if we really do make it harder for kids to obtain pot, that will take more enforcement, harsh penalties for dealing to minors, the whole bit. That just further adds to the enforcement costs.

    The point is that there is no free lunch.

    At the end of the day, it might be less costly to deal with marijuana legally than our current prohibiting it. But, that doesn't mean we shouldn't be careful and honest with the research that states this point.

    All in all, I appreciate the criticism.

  5. I decided to read through the actual report a bit, and from the perspective of a layman (I am not an economist by any means) you seem to be cherry-picking the one assumption that supports your position while simultaneously ignoring a great many other assumptions that would either a) lessen the significance of it or b) make it outright irrelevant.

    What do I mean? In reading through the study, I was astounded by a great many instances where the authors estimate on the conservative side of both the costs and tax revenues. If all of these under-estimates add up to greater than the sum of these theoretical costs you keep mentioning (and I see no reason to think that they wouldn't), then your point is moot.

    You mention the judicial costs. However, what you don't mention is that in the context of the report, the judicial costs refer solely to felony arrests. Now, if I had to guess at what percentage of these arrests were for minors, I would guess 0%. So, what you are really talking about is the percentage of judical costs only for people who are both over 18 AND under 21. Not quite so significant anymore...

    Also, in your discussion of supply and demand and how prices would affect usage, you completely neglect many factors of reality including the fact that a large percentage of the population does not want to use marijuana and would not use it even if it were both legal AND absolutely free!


  6. Dave,

    I am glad you read the report, but I'm sorry you feel that I am cherry-picking. My main point -- that there remain costs after legalization -- is not a cherry pick.

    On my other points, I can respond to your impression with a quote from the report:

    "There are other possible savings in government expenditure from legalization, but these are minor or difficult to estimate with existing data. The omission of these items biases the estimated savings downward."

    In other words, the report states that most of the times where he errs on the conservative side, it is minor (or he chose to not collect data adequate enough to estimate the effects, so he must not have thought it was important to the budgetary cost). Given that he thought these costs were not worth further investigation, I thought it wasn't worth my time to highlight them.

    On your second point, the report uses the fraction of felony arrests for marijuana as a "stand-in" for all arrests for marijuana. He then multiplies the entire judicial budget by the "stand-in" percentage. In the report's words,

    "A reasonable indicator of this percentage is the fraction of felony convictions in state courts for marijuana offenses."

    This is a standard econometric technique: the report's author really wants to estimate the entire judicial cost of marijuana (including misdemeanors, etc.), but he can only readily obtain data on the fraction of marijuana felonies. If I were stuck without data (but with a good "stand in"), I would consider doing what he did.

    More concretely, his calculation is:

    MJ Judicial cost = [% of all felonies that are MJ]*[Total judicial budget]

    For this calculation, he really wants to use [% of all prosecuted crimes that are MJ], but he doesn't have that information. Therefore, he assumes that the fraction of all felonies due to marijuana is about the same proportion as the fraction of all judicial proceedings due to marijuana. That's an additional assumption, but it is probably reasonable.

    It is not reasonable to interpret this exercise as meaning that felony arrests are the only costly arrests. He's still trying to understand the costs of non-felony arrests with this technique. He just doesn't have or use good data on the non-felonies.

    In fact, proponents of the "marijuana costs too much" argument would much rather point to all of the costs our government is spending on non-felony arrests and prosecution of marijuana -- the claim is that we spend boatloads of money to put "peaceful druggies" in jail. What's more peaceful than a misdemeanor? Miron would not, and did not, leave those costs out of the report.

    Last point, even in a world where no more people use marijuana if it is legal, we will see users of marijuana use it more intensively -- that's one result from my supply and demand analysis. Is this more costly to enforce? Perhaps. Consider an example:

    Suppose there is a town where 1000 minors use marijuana. When it is illegal, the cost of marijuana is such that each minor uses twice per month. When it is legal, the cost is such that each minor uses three times per month.

    In this example, the police have to contend with 1000 more infractions per month (an increase of 50 percent). 50 percent more offenses means more cost, even without drawing in additional users.

    Again, I appreciate the criticism. That's why I put this out there. My only goal is to learn more about this issue.

    As I stated in my post, I'm skeptical of both sides of the debate. I get better, more empassioned criticism if I take one side. Better criticism helps me learn more.