David R. Henderson  

False Externalities

PRINT
Paul Krugman on jobs and GDP... The new left...
However, increased consumption among youths leads to negative externalities. Raising the share of adults registered as medical marijuana patients by one percentage point increases the prevalence of recent marijuana use among adolescents and young adults by 5-6% and generates negative externalities in the form of increased traffic fatalities (7%) and alcohol poisoning deaths (4%).
This is from the abstract of "The Kids Aren't Alright but Older Adults are Just Fine: Effect of Marijuana Market Growth on Substance Use and Abuse," by Rosanna Smart, a Ph.D. student in economics at UCLA.

In case you think there is no connection between the first sentence and the second, there is. In the paper for which this is the abstract, she does claim that people who use marijuana and then cause traffic accidents and/or die from alcohol poisoning are creating a negative externality.

Let's consider the two alleged externalities in turn.

First, traffic accidents and fatalities. There certainly is a strong case to be made that people who use marijuana and then cause traffic accidents and fatalities are creating a negative externality. But the case must be made that the accidents hurt other people, damage other people's property, or kill other people. She does not do that. Moreover, she does not appear to put any weight on the distinction between killing oneself and killing others. If I, through some foolish act, kill myself but don't damage anyone else's property or injure or kill anyone else, I have not created a negative externality.

Second, alcohol poisoning. Here the reasoning is even more straightforward. If using marijuana causes me to die from alcohol poisoning, presumably because marijuana and alcohol are complements, I have not created a negative externality. I have done something destructive to myself, but there is no negative externality.

I looked elsewhere in Ms. Smart's paper for some hint that she was aware that she needed to make a case for her externality claim. This, on page 18, is the closest she came:

If individuals are rational and fully anticipate the potential negative consequences of marijuana consumption on future utility, then any increase in use induced by medical marijuana availability increases consumer welfare (Becker and Murphy, 1988). If, however, individuals underestimate the potential negative consequences (e.g., risk of dependence) or make mistakes in their consumption choices triggered by environmental cues, this increased marijuana use may decrease social welfare (Laibson, 1997; Bernheim and Rangel, 2004). Consumption by children is often assumed to be suboptimal prima facie, as reflected by the range of policies designed to guide their consumption decisions (e.g., the minimum legal drinking age, compulsory schooling laws, and debt contract age limits). Under any theory of individual decision-making, if marijuana use generates negative externalities, then the socially optimal level of consumption is below the individually optimal level of consumption achieved under a free market regime.

In other words, she jumps from entirely plausible claim that young people may make bad decisions that are not in their long-run self interest to the non sequitur that those decisions create negative externalities. But recall that a negative externality is a cost imposed on someone else. If I kill myself, I kill my self.

HT to Tyler Cowen.


Comments and Sharing


CATEGORIES: Political Economy




COMMENTS (37 to date)
James Hanley writes:

That last quote also leaves me uncertain whether she realizes that externalities can result from rational choices.

But to play devil's advocate here, if you kill yourself and those who love you experience great emotional pain asca consequence, could that be considered an externality? You may kill only your self, but others are negatively affected. (And here's hoping that our deaths don't create too many positive externalities!)

Greg G writes:

>---"If I kill myself, I kill my self."

If you kill yourself there may be many devastating externalities on those who depend on you in various ways - especially on those who love you. I am not arguing in favor of the War on Drugs which carries many devastating externalities of its own. I'm just saying this is a very simplistic view of externalities.

Philo writes:

I think some suicides do create negative externalities, by making others dispose of the body without leaving resources sufficient to pay for the disposal.

David R. Henderson writes:

@James Hanley,
But to play devil's advocate here, if you kill yourself and those who love you experience great emotional pain as [a] consequence, could that be considered an externality?
Yes, but notice that she didn’t make that claim. And notice that her first sentence I quote in the second quote forecloses that option in her view of the world.

Philo writes:

It seems that Smart deals only with the consequences in the sort run of legalizing marijuana. In the long run society would adjust in various ways to legalization. The long-run consequence are more important in evaluating public policies; of course, they are harder to discern.

Glen writes:

When she quotes Laibson and Bernheim & Rangel, she is implicitly invoking "internalities," i.e., externalities imposed by one's (present) self on another (future) self. This is another variant of behavioral paternalism -- which, as you know, is highly problematic.

In addition, she says this: "Under any theory of individual decision-making, if marijuana use generates negative externalities, then the socially optimal level of consumption is below the individually optimal level of consumption achieved under a free market regime." This is not necessarily so. First, if there are also positive externalities or monopoly power present, these may partially or completely offset the effective of negative externalities. Second, if there are any existing taxes, they may offset the effect of negative externalities. Third, if there are any existing punishments for the negative consequences -- e.g., fines for accidents, jail time for inebriated driving and manslaughter, damage payments for liability claims -- these may have already internalized the externality.

David R. Henderson writes:

@Philo,
Good point. Of course, the short run is all she has to work with, but she should then be more careful than she is about extrapolating.
@Glen,
Thanks.

Justin K. writes:

The only feasible argument I could see for a negative externalize caused by one's death without harming others or property is if that person dies without a will and without resources to provide for their interment and the disposal of their property. This should pretty clearly provide a negative externality. Otherwise great post.

Philo writes:

Don't grammarians still frown on "alright" (as opposed to "all right").

Rusty writes:

I guess I would assume that increasing the number of fatal accidents would most likely involve other people. Perhaps she figured this was obvious. I agree that she doesn't explain well the externalities she is asserting resulting from ones own death.

jon writes:
But the case must be made that the accidents hurt other people, damage other people's property, or kill other people.

I'm trying to imagine a scenario where a fatal traffic accident doesn't involve some harm to other people or their property. The best I can come up with is a single person, with no dependents, that crashes into their own fully-paid, non-insured, home.

Hazel Meade writes:

In addition, in order to be an "externality" the costs imposed on others must be uncompensated. We already have liability insurance for car accidents. Unless marijuana smoking somehow makes people drive without insurance, then an increase in accidents isn't an externality.

Matt Moore writes:

"@James Hanley,
But to play devil's advocate here, if you kill yourself and those who love you experience great emotional pain as [a] consequence, could that be considered an externality?
Yes, but notice that she didn’t make that claim. And notice that her first sentence I quote in the second quote forecloses that option in her view of the world."


Depends entirely on if you take that cost into account or not. I expect most people do. In fact, I expect that taking into account other's feelings maybe the leading preventer of suicide / self-destructive behaviour.

hanmeng writes:

@ Philo, Justin K.,
Can't people donate their bodies for education or research to institutions that will eventually cremate and dispose of the remains at no cost to the donor? In cases where the body is not suitable for research, cremation costs as little as $250.

I don't understand the concept of negative externality well enough to know whether the financial and emotional impact of someone's death is not considered one.

Matt Moore writes:

Unintentionally raises a somewhat interesting point though.

There are various definitions of externalities that I have seen, but two archetypes:

1) A cost that is not considered by the decision maker

2) A cost that affects someone other than the decision maker.

It is obvious you can have 1 but not 2 (some rationality failure), and also 2 but not 1 (empathy).

I'm not sure how to parse this. Does "external" mean an external agent or external to the decision?

I guess 2 but not 1 isn't really an externality, since it enters the agent's utility.

I suppose the only reasonable solution is to require 1 and 2 always. But I have sympathy for 1 but not 2 as an externality. It's not analytically different, although it is philosophically.

Just thinking out loud

Hazel Meade writes:

@Matt Moore

There are various definitions of externalities that I have seen, but two archetypes:

1) A cost that is not considered by the decision maker

I don't think whether the decision maker anticipates the cost or not matters at all. A polluter make be fully cognizant of the fact that he is imposing costs on others and still be incentivized to do it if he or she is not paying for those costs. That is the classic example of an externality.

Only (2) counts. An externality is an uncompensated cost that is imposed on others.


James Hanley writes:

Matt Moore,

If I am your neighbor amd know you have to get up early for work but go ahead and host a loud party that I know will keep you from sleeping, we have 2 but not 1, and a classic example of an externality.

Matt Moore writes:

Sorry, by 'considered' I meant, fully enters into the utility function.

That is, perfect empathy precludes externality.

Jake from Kent writes:

David Friedman's explanation of externalities in "Law's Order," I think, would clear up a lot of confusion by all parties.

Standard definitions of externalities don't capture most of what economists mean when they call a cost an "externality."

Most textbooks might define negative externalities it as "costs imposed on third parties," but that does not quite define externalities for policymakers. There are many examples of costs imposed on third parties that are *not* externalities.

Way back in medieval england (so I'm told), an enterprising teacher opened up his own private school and charged students a fee for his services. Like most markets, potential competitors observed his earning economic rent, and so one other educator opened up his own school to compete with the pre-existing school. Predictably, this drove down the price of schooling and the revenue earned by the first headmaster.

The first headmaster sued the second in court, claiming that the latter imposed a cost on the former.

Predictably, the plaintiff lost. The reason that he lost is because his own loss was at least exactly equal to the gain by students and his competitors.

This is not an externality, but it is an example of a cost being imposed on third parties.

The impact on social welfare, then, should be an essential part of what exactly we mean by "negative externality." Negative externalities are costs imposed on third parties that unambiguously reduce social welfare.

Brandon Berg writes:

But the case must be made that the accidents hurt other people, damage other people's property, or kill other people.

Is it really necessary to make this case? It seems pretty obvious to me. Sure, sometimes someone will just crash into a tree or drive off a ledge (and even then, taxpayers usually get stuck paying for the cleanup), but it would be very surprising if none of the additional traffic accidents were two- or multi-car collisions.

Hazel Meade writes:

The impact on social welfare, then, should be an essential part of what exactly we mean by "negative externality." Negative externalities are costs imposed on third parties that unambiguously reduce social welfare.

I don't think this works either. What counts as "social welfare" is far too subjective. Moreover the classic externalities mentioned above may not have a significant impact on "social welfare" because they may only affect a few people. That does not mean they don't count as externalities. It would not do to have a definition of externalities that only counts when large numbers of people are affected, or when some powerful group of people feel that some social value they hold dear has been impugned.

A better approach is to make some determination as to what sorts of harms must be compensated for, a determination which is grounded in rights theory. For example, in your teacher example, suppose we were to say that everyone has the equal right to enter the teaching profession. In that case, while the first teacher has been harmed, he has no right to demand compensation. His rights have not been violated because he had no exclusive right to be a teacher.

Externalities are thus uncompensated harms that violate people's rights. The question then becomes how to properly define those rights. Should everyone have the right to enter the teaching profession, or only some people?
To get back to the original post, do people have a right to prevent loved ones from making poor life decisions if those bad decisions cause emotional pain to themselves? If not, then then the harms caused by increased marijuana consumption do not count as externalities.

andy writes:

Sure, sometimes someone will just crash into a tree or drive off a ledge (and even then, taxpayers usually get stuck paying for the cleanup), but it would be very surprising if none of the additional traffic accidents were two- or multi-car collisions.

I was under the impression that marihuana makes people work slower, so actually the impact on traffic accidents should be very muted as these people would drive slower and thus the accidents tend to be very mild. Is that wrong impression?

ThomasH writes:

Definitions are whatever people agree on, but I'm not a fan of using "externality" for non-market actions. If I pollute the air and someone dies as a result, that is an externality, it I shoot that person, that is not an externality. And "externality" does not mean any behavior that might in principle be regulated to produce a better outcome.

Ralph writes:

More generally, since human beings are hyper-social, don't almost any individual actions have externalities, positive or negative?

Anonymous writes:

A few year ago, an editorial appeared in a wildlife magazine that alleged to point to a "market failure." It seems that some sportsmen choose to purchase hunting rifle that are more expensive than are others, and the editorial author suggested this was a "market failure" since those sportsmen could have saved some money by purchasing less expensive rifles that were just as serviceable - in the author's opinion - as were the more expensive models.

My response in a letter to the editor was rejected.

David R. Henderson writes:

@ThomasH,
If I pollute the air and someone dies as a result, that is an externality
We agree on this part of your statement. And notice that your polluting the air could easily be a non-market transaction. You burn would in your fireplace and people downwind suffer from pollution. There is no market transaction here.

David R. Henderson writes:

@Brandon Berg,
but it would be very surprising if none of the additional traffic accidents were two- or multi-car collisions.
I would be surprised too. What’s your point?

Hazel Meade writes:

@ThomasH

I'll add an amendment. Externalities are incidental costs imposed on others, which violate their rights, and which are not otherwise compensated for.

Tom West writes:

This touches a sort point for me, so excuse my rant...

I'll admit that it's somewhat odd having an economics Ph.D. student misuse the term externality.

However, having said that, it's patently obvious to any human that "externality" barely captures any aspect of the impact that a suicide has upon humans, society, and in fact, the local economy.

I have no problem with this, economics is studying an incredibly narrow aspect of the human condition. In a shooting, we don't think that a physicist should analyze more than the direction and momentum of the bullet.

But somehow, economics is thought to be a sound basis on which to advocate or at least offer comment on certain policies.

I'm happy to agree with David on his definition of externality. But then clearly externalities as so defined have next to nothing to offer with respect to arguments about suicide or legalization.

(Not that I've seen David attempts to do so, but I've seen plenty of others use the sort of "motte-and-bailey" argument of argument where externality is calculated very narrowly, then that same figure is used in policy debates where externality implies the sum total of the cost of that policy upon others.)

Hazel Meade writes:

@Tom West
I also find it irritating when people argue under the apparent assumption that policies should be decided on the basis of broad harm statistics, whether or not we call them externalities.

Should people have the right to do risky things to themselves? Should people have the right to be protected from emotional harms? Those are the questions that should decide marijuana policy, not aggregate statistics about car accidents and alcohol poisoning. The latter is just a kind of crude utilitarianism.

jon writes:
  • @Brandon Berg, but it would be very surprising if none of the additional traffic accidents were two- or multi-car collisions.
I would be surprised too. What’s your point?

Not Brandon, but I think he is making a similar point to mine:

Most fatal traffic accidents are the result of running into something, killing you and causing damage to the 'something.' In such cases, that 'something' is someone and/or their property. How is that not an externality?

David R. Henderson writes:

@jon,
Most fatal traffic accidents are the result of running into something, killing you and causing damage to the 'something.' In such cases, that 'something' is someone and/or their property. How is that not an externality?
It is. That’s why I specified the traffic fatalities as I did: to not be an externality, they would have to meet the criteria I set out.

Nathan W writes:

There could be some issues about correlation versus causation happening here.

Tom West writes:

Hazel, an argument for legalization based on fundamental freedom is one I can accept and respect. I don't agree with it, but I think it's an entirely valid approach.

(I do agree with legalization, but mostly on the basis of crude utilitarianism.)

Jake from Kent writes:

@Hazel from Meade

"What counts as "social welfare" is far too subjective."

Social welfare in most principles of economics courses describe welfare as an aggregate summation of consumer and producer surplus across an economy, which is almost purely descriptive and makes no comment on rights. By this definition, then, we have a method to determine whether parties ought to receive compensation for harms: would the two parties, absent any bargaining or transaction costs, agree to eliminate or reduce the externality?

If reducing the externality increases total surplus, then the net winners should, in principle, be able to compensate the net losers to such an extent that the losers agree to eliminating or keeping the externality.

I'm sure you are aware of the following simple example, but I feel that repeating it can help both of us.

Imagine that there are two businesses, a steel mill and a tourist resort, both of which use the same river. I'm going to use four possible worlds to show how we arrive at the efficient outcome through bargaining.

The steel mill pollutes the river. The resort loses $10,000 worth of business due to this pollution, but it would cost the steel mill $5,000.

Now let's see what happens if the steel mill has property rights to the river. The resort can pay the steel mill at least $5,000 (and $10,000 at most) to stop polluting, and both parties would, in total, be better off by $5,000.

Now let's see what happens if the resort has property rights: they can sue the steel mill and win in court, which means that the steel mill will stop polluting. In both cases, total welfare goes up by $5,000. No matter who has property rights, the efficient outcome is achieved.

Now let's say that the efficient outcome is to pollute: the resort loses only $5,000 from pollution, but it would cost $10,000 for the steel mill to stop polluting. If the resort has rights to the river, the steel mill can pay the resort at least $5,000 (at most $10,000) to pollute, and total welfare will increase by $5,000.

Now let's say that the steel mill has rights to the river: it will pollute, and total welfare is, again, increased by $5,000.

Again, absent transaction costs, the efficient outcome is achieved.

So how do we know social welfare is reduced? We ask what would happen if there were no transaction costs. If the externality would be eliminated, then the externality unambiguously reduces welfare. If the externality would not be eliminated, then the externality cannot be said to reduce welfare.

The Original CC writes:

@DH

It is. That’s why I specified the traffic fatalities as I did: to not be an externality, they would have to meet the criteria I set out.

DH, I think the commenters are referring to this part of your original post:

But the case must be made that the accidents hurt other people, damage other people's property, or kill other people. She does not do that.

The other are arguing (as I would) that it is patently obvious that most traffic accidents hurt other people and it'd be a little silly for the author to spell that out.

Hazel Meade writes:

@Jake from Kent

OK, I see where you are going with that. Marijuana legalization should always increase social welfare because it allows people to bargain over any externalities created. Banning marijuana would be analagous to having a blanket environmental regulation in your example banning any river pollution, regardless of whether the resort would accept a payment allowing some pollution.

Comments for this entry have been closed
Return to top