In yesterday’s post about Eric Garner, I wrote:
Suppose you are a typical street vendor of an illegal product, such as, oh, say, untaxed cigarettes.
Suppose the police make a habit of harassing such vendors, by confiscating their products, smacking them around, hauling them off to jail, and perhaps occasionally killing a few.
I have good news: The police can’t hurt you.
…
Here’s why: Street vending can never be substantially more rewarding than, say, carwashing. If it were, car washers would become street vendors, driving down profits until the rewards are equalized. If car washers were happier than street vendors, we’d see the same process in reverse. (The key observation here is that it’s very easy to move back and forth between street vending and other occupations that require little in the way of special training or special skills.)
Because police harassment of street vendors has no effect on the happiness of car washers, and because street vendors are always just as happy as car washers, it follows that police harassment has no effect on the happiness of street vendors.
…
So if you’re a street vendor, the police can’t hurt you. On the other hand, when the police go around putting people in deadly chokeholds, they’re clearly hurting someone. So the question is: Who?
Answer: Not the vendors, but their customers. Fewer vendors means higher prices. That hurts consumers, and the sum total of that harm adds up to the harm that we see in the viral videos.
Several commenters jumped in to question the claims that:
- If you’re a street vendor, the police can’t hurt you.
- The costs of police harassment ultimately fall on consumers.
I’d like to thank those commenters — particularly David Sloan, Keshav Srinivasan and Eric — for keeping me honest and for persisting when I was initially too quick to dismiss their questions.
With regard to the first point, what I actually should have said was:
- If you’re a street vendor, the police can’t hurt you more than an eentsy weentsy bit.
That’s because harassment causes street vendors to move into a great many other occupations, one of which is car washing. For every displaced street vendor we get, say, 1/2000 of an extra car washer — bringing wages ever so slightly down in the car washing industry and therefore making both car washers and street vendors ever so slightly worse off.
I do not consider this a significant correction.
With regard to the second point, it would have been more accurate to say this:
- The greater the harassment, the more of its burden falls on consumers in the harassed industry.
More precisely, if we consider the harassment equivalent to a tax of T, then the burden on producers tends to grow linearly in T while the burden on consumers in the harassed industry tends to grow quadratically in T.
However, here are two points I now realize I’d overlooked:
- The linear/quadratic thing is at least partially misleading, because there is a limit on how big T can be — if T grows beyond a certain point, then the first industry disappears entirely. So we’re not looking at arbitrarily large T’s here, making “growth rates for large T” less relevant. Thus workers collectively can in fact — and in contrast to what I said yesterday — bear a substantial burden of the cost.
- While consumer surplus in the first industry shrinks quadratically in T, consumer surplus in the other industries grows quadratically in T, and in fact, the total consumer surplus across all industries can increase as a result of the street harassment. Thus it’s possible for workers to bear more than the entire burden of the harassment!
Here’s an explicit model:
There are N workers spread over K industries. A worker in any industry produces one unit of output. Demand in any industry is given by Q=A-P. Workers in the first industry are subject to a tax of T.
Then it’s not hard to check that we get (N+T-K)/K workers in the first industry and (N+T)/K in each of the others. The tax reduces each worker’s wages by T/K (there’s that eentsy weentsy damage), and therefore the wages of all workers by NT/K (linear in T).
Consumer surplus in the first industry is (N-(K-1)T)2/2K2, which falls off quadratically in T. But consumer surplus in all industries taken together is (and this surprised me, though now that I see it, it’s obvious) (N2+(K-1)T2)/2K, which grows quadratically in T. Because we know there is a net social loss, it follows that the net loss of NT/K to the workers has to be large enough to cancel this gain to the consumers plus the implicit tax revenue.
Bottom line: I stand by the claim that individual workers are hurt insignificantly by harassment. But I’m backing off the claim that the total harm to all workers is insignificant. Consumers in the harassed industry do bear a large burden — and perhaps even more than the entire social burden, because their losses are partly offset by gains to consumers elsewhere. But my simple accounting of yesterday (“the burden on the workers is effectively zero, so the social burden falls entirely on consumers”) was far too simple.
I still don’t get your points. How are the consumers harmed? The harassment is not harassment, but just enforcement of a tax. The taxes seem high to me, but the legislature seems to think that they are fair and appropriate. If consumers have to pay a fair tax, that is not harming them. That is just consumers fairly paying for what they get. Maybe those taxes are being spent in a way that benefits the smokers.
1) There’s been a varyingly explicit “in the long run” throughout this entire discussion.
If the police start indiscriminately murdering members of a particular profession, what does the theory of “they’ll change jobs” have to say about how many people will die before equilibrium is reached?
2) What’s your intent in stating “If you’re a street vendor, the police can’t hurt you more than an eentsy weentsy bit.”? The most charitable I can come up with is “Police brutality hurts us all, even if it doesn’t appear to hurt us directly”, but that’s extremely roundabout.
3) All of this breaks down when the police “tax” a “profession” that can’t be displaced, like “black people”. Telling someone who’s being targeted because of their race that they should just change jobs is unhelpful at best.
David Sloan:
Regarding your first point, yes, absolutely.
Regarding your second, my intent is to demonstrate that economic reasoning can sometimes surprise us.
Regarding your last, one corollary is that, holding fixed the number of people who are harassed, it’s better for them to be chosen on the basis of race (or height, or something else unchangeable) than on the basis of occupation.
Steve, am I right in drawing the following conclusions from your post?
1. Nonsmoking consumers have an incentive to lobby for increase cigarette taxes, even when the “tax” doesn’t actually collect revenue for the government (like chokeholds).
2. It’s possible that even smokers could have an incentive to lobby for higher cigarette taxes.
Both conclusions (if true) are absolutely startling to a non-economist like me.
Never mind, my second conclusion is false, but is my first conclusion true?
Keshav: I agree that your first is true — at least if you insert the word “might” — and your second is false.
My model assumes no cross elasticities across products. A tax on cigarettes pushes suppliers out of the cigarette industry and into the broccoli industry, reducing the price of broccoli, which is good for broccoli-eating nonsmokers. On the other hand, in a different model, the same tax might lead cigarette smokers to switch from tobacco to broccoli, bidding up the price of broccoli, which is bad for the nonsmokers. The net could go either way.
If you are introducing demand substitution towards other industries and counting the tax income (both adjustments make sense), I’d say this calls for doubling down and introducing income effects, as sellers are also consumers and their income responds to price changes. Even imposing a linear demand, the intercept could depend on the tax rate.
Of course, tax incidence in non-quasilinear worlds is a very tricky business, so there’s something to be said for just the level of complexity that allows reasonably easy algebraic solutions.
Steve Landsburg (#3): “holding fixed the number of people who are harassed, it’s better for them to be chosen on the basis of race (or height, or something else unchangeable) than on the basis of occupation.”
If I recall correctly from my university econ classes, this claim is made on the basis that penalizing something changeable causes people to change in costly ways to avoid the penalty, while penalizing something immutable does not, since they can’t change.
A couple related objections:
1) I think it’s unreasonable to hold harassment constant, as if it were some sort of innately useful thing that we need to make quota on.
2) The only possible benefit of harassment is to discourage people from doing the thing you’re harassing them for doing. If you’re targeting people who cannot change what you’re harassing them for, this benefit is entirely lost.
3) Predictably, if you penalize people for immutable characteristics, they’ll get pissed off and lose any faith they had in the system. How expensive are revolutions?
David Sloan:
2) The only possible benefit of harassment is to discourage people from doing the thing you’re harassing them for doing. If you’re targeting people who cannot change what you’re harassing them for, this benefit is entirely lost.
But we (or at least I) can easily imagine a situation where harassment is adopted because it serves special interests, but discourages people from being socially useful. In that case, I’d rather see the harassment directed against random people chosen for characteristics they can’t change. Of course, this is strictly a “which is worse?” thought experiment, in the sense that, like you, I can’t imagine a situation where the amount of harassment is fixed and the policy problem is how to allocate it.
Steve Landsburg (#9): “But we (or at least I) can easily imagine a situation where harassment is adopted because it serves special interests, but discourages people from being socially useful.”
I don’t think I understand you here; can you provide an example?
David Sloan:
I don’t think I understand you here; can you provide an example?
Conceivably, the harassment of Eric Garner is an example — it exists because it serves the interests of the “legitimate” cigarette sellers in the neighborhood (and of course the tax collectors) but could well be welfare-disimproving.
The model also assumes perfect information.
It is questionable whether Eric Garner assigned even a remote likelihood to the possibility of being choked to death by the police. You are also assuming that Eric Garner had the easy option of switching to another low-skilled occupation. I’m not sure that is the case either, unless you are thinking of self-employed squeegee men.
This is very interesting, and will take a bit more looking into for me.
I don’t think your answer in 11 is correct. Your comment goes on to say ” I’d rather see the harassment directed against random people chosen for characteristics they can’t change.” Garner was harassed not for these reasons, but due to his chosen activity that he could have changed. Can you give an example where the harassment is directed against random people for unchangeable characteristics?
Keshav’s points out that “Nonsmoking consumers have an incentive to lobby for increase cigarette taxes, even when the “tax” doesn’t actually collect revenue for the government (like chokeholds).” You say this is correct.
Combine this with “I’d rather see the harassment directed against random people chosen for characteristics they can’t change.” So the benefits for the un-taxed are even greater if the taxed cannot change- even if the tax is harassment rather than revenue raising. It seems that majorities have an incentive to persecute minorities. I had understood that such persecution tended to make everyone worse off. What is going on?
An aside – can addictive substances be subjected usual supply / demand treatment?
harold:
Your comment goes on to say ” I’d rather see the harassment directed against random people chosen for characteristics they can’t change.” Garner was harassed not for these reasons, but due to his chosen activity that he could have changed.
I agree. This puts Garner’s harassment in the less desirable category.
Can you give an example where the harassment is directed against random people for unchangeable characteristics?
I don’t think it’s hard historically to find examples of harassment directed against peopld solely because of their race. I was going to suggest the pogroms against the Jews. The Jews, of course, could have converted, but didn’t, so they might as well have not been able to.
Keshav’s points out that “Nonsmoking consumers have an incentive to lobby for increase cigarette taxes, even when the “tax” doesn’t actually collect revenue for the government (like chokeholds).” You say this is correct.
Combine this with “I’d rather see the harassment directed against random people chosen for characteristics they can’t change.” So the benefits for the un-taxed are even greater if the taxed cannot change- even if the tax is harassment rather than revenue raising. It seems that majorities have an incentive to persecute minorities. I had understood that such persecution tended to make everyone worse off. What is going on?
I’m not sure what contradiction you’re seeing. Harassment based on race costs the victims (say) $100. Harassment against cigarette vendors costs cigarette vendors $100, costs car washers $20, and benefits the consumers of car washes $10, for a net cost of $110. I (who care about social gain) prefer the harassment based on race, whereas the consumer of the car wash (who cares only about his own gains) prefers the harassment based on vending.
“The key observation here is that it’s very easy to move back and forth between street vending and other occupations that require little in the way of special training or special skills”
If one assume that is not true does your model break down ?
I become a street vendor because I’m good at bargaining and make $1000 a week. My second best profession would be a car-washer where I have less good skills and could only earn $800 a week. I get harassed, and if that harassment is worth more than negative $200 a week I will switch professions.
I am worse of by $200 a week. Society is worse off roughly to the extent of the difference between how they value $1000 of street vending compared to how they value $800 of car-washing. Its impossible to see how much subjective utility they lose but using reasonable assumptions the ex-vendor has paid a large part of the price of the harassment.
If the harassment extends to the whole vending industry and not just me then to the extent that other vendors also lose the ability to sell their vending skills at market price they will all lose as a result of the harassment.
@15
I don’t think a skill of bargaining in this case provides any significant benefit to society, in fact it may impose a negative cost.
That is actually an interesting question in and of itself.
Rob Rawlings: Street vendors with a special talent for street vending might or might not be made worse off (or better off) by harassment. Getting harassed is bad; having your competition harassed is good. Not all of them have special talents, so some will leave and your rewards will grow.
Under my “no special talents” assumption the good and the bad must exactly cancel. Under more general assumptions, either could be bigger.
Steve,
One could indeed argue that some vendors are better off as a result of harassment.
Start with “no special talents” assumption. When street vending starts getting harassed then a “special talent” of successfully dealing with harassment is introduced and the market price of this skill increases.
People with that talent who choose to be in the street vending business (and those who who exploit it elsewhere) after the harassment starts may actually be better off than they were before.
Steve Landsburg (#11): “it serves the interests of the “legitimate” cigarette sellers in the neighborhood”
The mechanism by which it does that seems to be my “The only possible benefit of harassment is to discourage people from doing the thing you’re harassing them for doing.” Specifically, it’s discouraging other street vendors from street vending.
“(and of course the tax collectors)”
How does police harassment help the tax collectors? I know you’ve been phrasing harassment in terms of a tax on street vendors, but that doesn’t mean anyone is actually collecting money from it.
“I (who care about social gain) prefer the harassment based on race”
It seems disingenuous to keep making this point in a world where you’ve already agreed “I can’t imagine a situation where the amount of harassment is fixed and the policy problem is how to allocate it.”
I (who also care about social gain) prefer *no* harassment. Taking your example numbers from #14 as given, the social benefits of eliminating harassment entirely are an order of magnitude higher than shifting it onto a group that is literally assumed not to be able to fight back.
Conflating taxation and harassment has done your point a disservice, I think. Taxes hurt for all the reasons you’ve described, but at least there’s the plausible social benefit of spending the collected tax on something. Harassment is pure pain inflicted on the world; all the benefits are secondary effects. The best argument you’ve presented in favor of harassment (back in your previous post) is that it’s an unavoidable byproduct of attempting to legally enforce anything, and I think a strong counterargument could be made that the pain of harassment is most efficiently avoided by a change in behavior of the harassers.
David Sloan:
“I (who care about social gain) prefer the harassment based on race”
It seems disingenuous to keep making this point in a world where you’ve already agreed “I can’t imagine a situation where the amount of harassment is fixed and the policy problem is how to allocate it.”
But the thought experiment is still relevant to policy.
There are many types of harassment I dislike. I have limited resources to devote to trying to get some of them eliminated. So I have to decide both which ones I can be most effective against and which ones I dislike the most. For the latter, it helps to know that “all other things being equal, I prefer type A to type B”. This is useful even though all other things are never equal.
For example, if I’m choosing a new car, it helps to keep in mind that, all other things being equal, I prefer extra-powerful air conditioning. This doesn’t mean I expect to ever find all other things equal, and it doesn’t mean I will always buy the car with the extra-powerful air conditioning. But when I’m making my list of pros and cons for the Honda versus the Toyota, I don’t want to ignore this factor completely.
Conflating taxation and harassment has done your point a disservice, I think. Taxes hurt for all the reasons you’ve described, but at least there’s the plausible social benefit of spending the collected tax on something. Harassment is pure pain inflicted on the world; all the benefits are secondary effects.
This seems off point to me. The essay you’re referring to concerned the effect of harassment on sellers, and to sellers, harassment is equivalent to a tax. I agree that if we’re talking about social gain, the two are not equivalent, and I agree that it’s important to talk about social gain sometimes. But that doesn’t mean we can never talk about anything else.
“Conflating taxation and harassment has done your point a disservice, I think. Taxes hurt for all the reasons you’ve described, but at least there’s the plausible social benefit of spending the collected tax on something.”
Perhaps pleasure to harassers from harassment is, however unsavory to most of us, a benefit to the harassers and thus part of social benefit and thus parallel to benefits from tax revenue.
#14. Thanks, I think I get it now.
David Cushman (#21): “Perhaps pleasure to harassers from harassment is, however unsavory to most of us, a benefit to the harassers and thus part of social benefit and thus parallel to benefits from tax revenue.”
If you’re going to honor that sort of preference, I prefer to spite harassers by inflicting roughly twice as much pain as they get pleasure out of harassing others.
I am not sure how one gets to “the police can’t hurt you more than an eentsy weentsy bit” from this.
The argument seems to be that street vendors, collectively, are hurt only so much as to create the current equilibrium between the supply of such people and the supply of workers in other fields with comparably low barriers to entry.
Since this does not tell us where that equilibrium lies, it may be that police activity drives the supply of street vendors far below the level it would be in the absence of attacks. (It also assumes that police attack street vendors, but not car washers. For the argument to make sense, one must compare street vending, with its risk of death, to some available alternative that lacks this risk). It may be that this is the overwhelmingly most important factor determining the supply of street vendors. Absent the risk of death at the hands of the police, perhaps there would be many more street vendors. The fact that the supply is in equilibrium hardly implies a negligible effect.
Then there is the question of whether “if you are a street vendor” then police “can hurt you more than an eentsy weentsy bit.” Of course they can. They can kill you. The argument also assumes that, although the police could kill you, they rarely actually do this. But this was not a condition in the original statement.
You could get closer to the claim about the effect on street vendors collectively by assuming that police almost never kill these workers. But even then, it hardly means that this low risk has little effect on the supply. You would need to know the value people place on the, presumed, greater safety of car washing vs street vending to have any idea the economic effect of the risk of death.
It would be accurate to say that the danger of both occupations influences the equilibrium, as do whatever other considerations would favor one or the other. It hardly follows that, because an equilibrium exists, every one of the factors must be have a minor effect.
Then step back a bit and ask what the reality might be here? Is this really the case that, among a huge number of low paying jobs with low barriers to entry, the police select one and create a tiny risk of death for those engaged in it? Or do we have a situation where the risk of death attends to people of the wrong race, in the wrong neighborhood, regardless of their occupation? In that case, which seems more plausible, the choice of occupation may have little, if any, impact on this risk. If people believed this choice had major implications for survival, they might adjust their jobs to reflect this. But the estimates of risk could be quite unstable, changing dramatically each time another car washer, newspaper vendor, or whatever, was executed.
Then the police could hurt anyone of the wrong group in the wrong area, with their job having no effect on risk. The economic effect would be random shifts in supply in various occupations, with obvious economic inefficiencies.