1. Econ 522 Economics of Law
Dan Quint
Fall 2016
Lecture 21

2. Logistics Midterm 2 HW4 Online course evaluations are open
It’s long – start it early!
Online course evaluations are open 1 1

3. Last week’s experiment2 2

4. Last Week’s Experiment
You have been asked to serve on a jury on a lawsuit dealing with personal injury. In the case before you, a 50-year-old construction worker was injured on the job due to the negligence of his employer. As a result, this man had his right leg amputated at the knee. Due to this disability, he cannot return to the construction trade and has few other skills with which he could pursue alternative employment.
The negligence of the employer has been firmly established, and health insurance covered all of the related medical expenses. Therefore, your job is to determine how to compensate this worker for the loss of his livelihood and the reduction in his quality of life. 3 3

5. What were we trying to test?
Half of you were asked…
The other half were asked…
Should the plaintiff in this case be awarded more or less than $10,000?
How much should the plaintiff receive? (Please give a number.)
Are you male or female?
Should the plaintiff in this case be awarded more or less than $10,000,000?
How much should the plaintiff receive? (Please give a number.)
Are you male or female?
Testing how much did the “suggestion” in question (a) affect the answers to question (b)? 4 4

6. So, did the first question make a difference?5 5

7. Distribution of answers to question 26 6

8. Distribution of answers to question 2
50k, 100k, 100k, 150k
4 x 250k, 300k
3x 500k, 600, 700
5 x 1MM, 3 more around 1.5 MM, 3 x 2 MM, 2.5 MM 7 7

9. Distribution of answers to question 2
100k
2 x 500k, 700, 900
3 x 1MM, 1.75MM, 3 x 2MM
9 x 5MM, 6MM, 7.5 MM
10, 12, 20, 20 8 8

10. Distribution of answers to question 2
The only statistical test one ever needs is the IOTT, or “inter-ocular trauma test.” The result just hits one between the eyes David H. Krantz
Asked $10,000
Asked $10,000,000
Minimum
50,000
100,000
Maximum
2,500,000
20,000,000
The average was almost 6 times as high, and the median over 7 times as high
Average
872,500
4,961,111
Median
650,000
5,000,000
% below $500,000
31% 4%
% above $3,000,000 0%
56% 9 9

11. But just going back to that graph…10 10

12. What about gender? Asked $10k
In both groups – those given the low prompt, and those given the high – there were fewer women than men, but 3 of the highest 4 answers, and 3 of the lowest 4 answers, were from women
Women overall had slightly higher average suggestions, but much higher variance
(same pattern held last time I ran this: women gave the lowest answer, and four of the five highest, including one woman who suggested $50 MM and another who suggested $100 MM!)
Asked $10MM 11 11

13. What do we learn? Nobody knows exactly what a leg is worth
Even among people asked exactly the same questions…
…answers varied from $50,000 to $2,500,000, or from $100,000 to $20,000,000
So there’s an incentive to fight!
Since no one knows the right answer, “suggestions” matter
Higher suggestion raised average response from under $900k to almost $5MM – a 470% increase!
“Anchoring effect”
Framing effects
Anchoring – we tend to rely too much on the first piece of information we get, even if it’s irrelevant
In this case, the high or low suggestion influenced our estimate of the “right” answer
Other examples (from Wikipedia page on “Anchoring”):
Kahneman and Tversky: people see a roulette wheel stop on either 10 or 65, then are asked what fraction of UN member nations are in Africa; those who saw 65 guessed almost twice as many, on average
Ariely: people write down the last two digits of their social security number, are asked whether they’d pay that much for a random object, then bid for an object; those with high numbers bid more
Framing effect – more general observation that exactly how a question is presented can have a big impact on how people decide on it
People playing “Prisoner’s Dilemma” versus “Cooperation Game” 12 12

14. Criminal Law

15. So far this semester, we’ve focused on civil law
Not Civil (Napoleonic) as opposed to Common Law (British)…
…but Civil as opposed to Criminal
Cases where one private individual is suing another…
…generally looking for monetary compensation…
…for some sort of wrong that was done
Some cases are handled differently
Friedman quote we saw earlier: “When someone shoots you, you call a cop. When he runs his car into yours, you call a lawyer.”
All semester, we’ve been dealing with the second case
Time to deal with the first

16. Criminal law differs from civil law in several ways
Criminal intended to do wrong
Case brought by government, not individual plaintiff
Harm done tends to be public as well as private
Standard of proof is higher at trial
If found guilty, defendant will be punished

17. What is the goal of criminal law?
Just like with civil law…
To achieve efficiency, minimize total social cost
Social costs consist of…
Social cost of crimes that are committed
Cost of detecting (catching) criminals
And cost of punishing offenders
error costs
Civil law: overall goal was to minimize total social costs, which consisted of administrative costs and error costs
Here, we have basically the same idea
Cost of crimes is analogous to error crimes, everything else is analogous to administrative costs
administrative costs

18. Intent Unlike a tort, a crime generally requires intent
Mens rea – a “guilty mind”
(Literal intent occasionally not required)
You’ve been hired as a lifeguard or a nurse
You show up to work drunk, and as a result someone dies
Criminally negligent homicide
(Sometimes intent is enough even without harm)
Attempted murder
The first distinction is that a criminal conviction requires intent
The legal term is mens rea, “a guilty mind”
Tort law considered harms that happened accidentally
An injurer might cause an accident while taking precautions, and still been liable under a strict liability rule
An injurer might cause an accident while not taking sufficient precautions, and been liable under a negligence rule
An injurer might cause an accident while being reckless, and been liable for punitive damages on top of compensatory damages
But in any of these cases, the injurer did not intend for the harm to occur.
There are circumstances where the line of intent gets blurred
If someone is in a position of responsibility for someone else’s safety and, through a pattern of negligence, causes that person’s death, they can be charged with “criminally negligent homicide” even though they did not intend for the person to die
(If you’re hired as a lifeguard and are drunk while on duty and someone drowns, you’re in trouble)
The flip side of this: sometimes intent alone is enough, as when people are charged with attempted murder even though no actual harm occurred.

19. Criminal cases are brought by the state
Recall wrongful death tort cases
Victim is dead, can’t receive compensation
Family/friends can sue for lost wages, lost companionship, etc.
Criminal cases are brought by the government
Don’t require living victim
This allows prosecution of “victimless crimes”
Theory is that all crimes harm the public – are “public bads”
That is, breakdown of law and order in society harms everyone
So public (represented by state) brings criminal actions
The next difference is who initiates the case
In civil law, a case is brought by the victim
In criminal law, a case is brought by the state
Recall that in tort cases involving accidental death, the victim could never receive damages; his family and friends could only sue for lost companionship or lost wages, not the lost life
On the other hand, criminal cases do not require a living victim.
This also allows prosecution of “victimless crimes” – that is, crimes which do not cause any clear harm to any individual, such as gambling, prostitution, and drug sales
The theory here is that all crimes harm the public
That is, the breakdown of law and order in society is a public bad
So the public (represented by the state) brings criminal actions, not the victim.

20. Distinction between civil remedies and punishment
Nuisance law, contract law, tort law: damages serve two purposes
Compensate the victim
Cause injurer to internalize cost of harm done
When injurer internalizes harm, we get pollution, or breach, or accidents, only when they are efficient
Criminal law: intention is to deter crimes – that is, prevent them entirely, not just the “inefficient ones”
Punishment need not be limited to magnitude of harm done
Criminal punishments – imprisonment, execution – destroy resources
Make criminal worse off, may not make anyone better off
“Ex post, punishment is inefficient”
A key difference between civil and criminal law is the distinction between civil remedies and “punishment”
In nuisance law, contract law, and tort law, damages serve two purposes
They compensate the victim
And they cause the injurer to internalize the cost of the harm
When the injurer internalizes the harm, we get pollution, or breach of contract, or accidents, to happen only when they are efficient, that is, when they are less costly than preventing them
In criminal law, punishment is not meant to compensate the victim directly
And in criminal law, punishment is meant to deter crimes – that is, to prevent them altogether, not just to prevent the inefficient ones
When we talked about civil law, the goal wasn’t to eliminate accidents, or eliminate breaches of contract – it was to get only the efficient amount of them
The goal of criminal law is different. We don’t want to get “only the efficient number of murders” – we would like to eliminate murder entirely, to the extent that it’s not too costly
And so punishment need not be limited to the magnitude of the harm done
Civil remedies generally transfer resources from one party to another, without destroying anything
Criminal punishments, such as prison time and execution, often destroy resources
That is, they are designed to make the criminal worse off without necessarily making anyone else better off.
So civil penalties are designed to make injurers internalize the costs of their harms, so that these harms only occur when they are efficient; but criminal punishments are designed to deter crimes, to discourage their commission in all situations.

21. Criminal cases have higher standard of proof
Most civil cases: preponderance of the evidence
Interpreted as: 51% certainty plaintiff is correct
For punitive damages, clear and convincing evidence
Higher degree of certainty
In criminal cases, prosecution must prove guilt beyond a reasonable doubt
Why so much higher?

22. Why should standard of proof be so high in criminal cases?
Think about error costs in either tort or criminal case:
If “false positives” are more costly in criminal law, suggests conviction should require more certainty
Probability of wrongful exoneration
Social cost of wrongful exoneration +
Probability of wrongful conviction
Social cost of wrongful conviction X X
Torts: reduced precaution
Criminal: reduced deterrence
Torts: excessive precaution
Criminal: excessive deterrence (?)
and social cost of excess punishment

23. Are crimes ever efficient?
Most crimes are clearly inefficient
To steal my laptop, you might break my car window
And, my laptop is worth more to me than to other people
Stolen cars are worth much less than legally-owned ones
And if you value my car more than me, there’s a legal alternative to you stealing it
We said: civil law aims to get the efficient amount of accidents, or breach, or nuisances, while criminal law aims to deter all crimes
This does beg an interesting question: are crimes ever efficient?
Clearly, most are not
In order to steal my laptop out of my car, someone might have to break my window
My laptop has lots of personal stuff on it – papers I’m working on, pictures of my friends – it’s probably worth more to me than to him
Plus, my window got broken
So the cost to me is likely larger than the benefit to the thief.
Similarly, stolen cars are worth much less than legally-owned cars
If someone steals my car, he’ll probably sell it for much less than it will cost me to replace it
Plus, he may have to reduce the value of the car to steal it – by breaking a window, jamming a screwdriver in the ignition, etc.
Again, the crime is inefficient.
And if a crime were efficient – say, someone stealing my car because they value it much more than me – there is often a legal alternative – offering to buy the car from me.

24. Are crimes ever efficient?
Most crimes are clearly inefficient
To steal my laptop, you might break my car window
And, my laptop is worth more to me than to other people
Stolen cars are worth much less than legally-owned ones
And if you value my car more than me, there’s a legal alternative to you stealing it
But Friedman offers examples of efficient crimes
Starving hiker lost in the woods finds cabin with nobody home, breaks in and steals food
Efficient murder
Rich guy decides he’d derive immense pleasure from hunting a human
Offers 10 people $1,000,000 each to draw straws, he gets to hunt and kill the loser
If they all agree, is this transaction efficient?
However, Friedman gives a couple of examples of efficient crimes
One is a situation we saw earlier: a hiker is lost in the woods, on the brink of starvation, and stumbles on a cabin with nobody home. He breaks into the cabin and steals some food.
(In this case, we have a legal doctrine to deal with the problem: private necessity. Rather than paying the penalty for trespass, breaking and entering, and theft, the “criminal” would likely just be liable for the damage done.)
Friedman follows this with a much creepier example: efficient murder
A very rich man decides he would derive a huge amount of pleasure from hunting and killing a human being
He finds ten people who all agree that he can pay them each $1,000,000, they’ll draw straws, and he can hunt and kill whoever draws the shortest straw
Since the transaction is voluntary, it seems like it must be efficient.

25. Are crimes ever efficient?
In 2001, Armin Miewes posted an ad online, “looking for a well-built 18-to-30-year-old to be slaughtered and then consumed.”
And someone answered.
They met, discussed it, and agreed Miewes would kill and eat the guy.
So he did. And videotaped it.
At the time, cannibalism was not illegal in Germany
Is it a crime to kill someone who has consented to be murdered?
In 2004, Miewes was convicted of manslaughter
In 2006, he was retried, convicted of murder, sentenced to life in prison
But also… if Miewes and his victim agreed he should be killed and eaten, and no one else was harmed, was this crime efficient?
In fact, there’s a recent case that sounds more like this than you’d expect
In 2001, a German man, Armin Meiwes, posted an ad online “looking for a well-built 18-to-30-year-old to be slaughtered and consumed.”
Seriously. And someone answered.
The two met, discussed it, and agreed that Meiwes would kill and eat the guy. (This would fulfill various sexual fantasies that both men had.)
So he did. And videotaped it.
Some of the details of the story are gruesome, I won’t share them.
In 2002, Meiwes was arrested, and admitted to what he had done
At the time, cannibalism was not a crime in Germany
So the question was primarily whether someone could be guilty of murdering someone who had consented to be murdered.
In 2004, Meiwes was convicted of manslaughter and sentenced to 8 and a half years in prison
In 2005, he was ordered retried, and in 2006, convicted of murder and sentenced to life in prison.
So again, if the rich guy and his ten greedy friends all consent to the hunting contract, or Meiwes and his victim agree that he should be killed and eaten, are these crimes efficient?
The answer depends on how much you buy into the argument of public bads
One view is that certain actions are so repugnant they harm others just by their stench
That is, everyone is somehow made worse off just by “living in a world where this type of thing is allowed to happen”
We’ll come back to the question of efficient crimes in a bit
For the most part, though, it’s probably safe to assume most crimes are inefficient

26. Why not use tort law to cover crimes too?
Tort law creates an incentive to avoid harms
If it worked perfectly, might be no need for criminal law
Reasons tort law may not work for certain offenses
Relies on perfect compensation, which may be impossible
Loss of life, crippling injury
Even if possible in theory, might be impossible in practice
If probability of being caught/convicted is less than one, deterrence requires punishment more severe than benefit received
And if we made civil penalties severe enough, criminals might be judgment-proof
Next question: why not just use tort law to cover crimes too?
We saw that tort law can create an incentive to avoid certain harms
If tort law worked perfectly, there might be no need for separate criminal law.
The book offers three reasons why tort law is inadequate to deal with certain types of offenses.
First, tort law depends on perfect compensation; but perfect compensation may be impossible, as in cases of the loss of a life or a crippling injury.
Second, even if perfect compensation is possible in principle, it may be impossible in practice
Court has no good way to figure out how much money would make me indifferent about losing an arm
There’s no clear market substitute; and if they ask me, I have an incentive to lie.
Third, if the probability of getting caught and convicted is less than one, then to deter the act, the punishment must be more severe than the benefit received
We made this argument when we talked about punitive damages
But the problem is even greater with deliberate acts.
(If anyone caught stealing money was simply required to return it, we would expect this to lead to an awful lot of theft.)
Plus, even if we could make the penalty more severe, criminals might be judgment-proof – unable to pay large enough damages – and the system would break down
So in cases where perfect compensation is impossible, where being caught is unlikely, and where defendants might be judgment-proof, tort law might not work too well, and there is room for criminal law.

27. Theory of criminal law A theory of criminal law must answer…
Which acts should be punished as crimes?
How should they be punished?
Cooter and Ulen:
Acts should be punished when aim is deterrence
Acts should be priced when aim is internalization
Aim should be deterrence when…
perfect compensation is impossible
people want law to protect rights instead of interests
or enforcement errors undermine liability
Much of today is from ch. 15 of Friedman book
(Recall a long time back, when we distinguished between injunction/property protection and damages/liability protection
We can recast this as the distinction between protecting a right and protecting an interest
If someone is free to cut off my arm, provided they pay me for it, then my interest in that arm is protected; but my right to have that arm is not
The book tries to make the case that in certain instances, it is better to protect rights than interests.)

28. General model of crime and punishment

29. Economic model of crime and punishment
Key assumption: rational criminals
Potential criminals weigh private cost – chance of getting caught, times severity of punishment – against benefit
If enforcement were free, we could eliminate crime
Hire enough police to detect nearly all crimes
Punish them very severely
Nobody rational would commit a crime
But enforcement isn’t free, making things interesting
The basic economic model of crime and punishment makes the following key assumption: rational criminals.
That is, we assume that potential criminals consider the costs of a crime – the chance of getting caught, and the severity of punishment – and weigh it against the benefits they get
If enforcement were free, we could hire so many police that nearly all crimes were detected, and punish them severely; this would lead to nobody rational ever committing any crime
The reason we don’t do this is enforcement costs.

30. Economic model of crime and punishment
To deter crime, we need to do two things:
Catch offenders…
…and punish them
Catching a higher fraction of offenders is more costly
Requires more police, more detectives, etc.
More severe punishment also tends to be more expensive
Most common punishments are fines and imprisonment
Fines cost nothing – state even makes money
But fines don’t always work, because not everyone can pay them
Besides fines, most punishments are inefficient – make offender worse off, and are costly to state
In order to deter crime, we have to do two things: catch offenders, and punish them.
Catching a higher fraction of criminals is more costly: it takes more police officers, more detectives, and so on.
More severe punishment also tends to be more expensive.
The most common punishments are fines and imprisonment
Fines aren’t costly at all – the offender pays $1000, the state receives $1000, the total social cost of the punishment is 0
However, fines don’t always work, because not everyone can pay them
Suppose there was a crime we wanted to deter by imposing a punishment equal to a $100,000 fine
A lot of the people who commit the crime may not have $100,000, so they would not face a sufficient deterrent.
(this is the problem of “judgment-proofness” we discussed in tort law)
So when fines are not sufficient, we have to turn to less efficient means of punishment, such as imprisonment
This clearly has a social cost: going to jail makes the offender worse off (as planned), but it also makes the state worse off, since it has to build the jail, staff it, and so on
So imprisonment is inefficient
Most punishments (aside from fines) are inefficient.

31. Gary Becker, “Crime and Punishment: An Economic Approach”
Suppose some crime is punished by 20% chance of being caught and convicted, and a punishment equal to $20,000
We could save money by…
Fire half the police and judges, so probability of being caught and convicted dropped to 10%
Raise the punishment to $40,000
Punishing someone $40,000 may cost more than punishing them $20,000, but not more than twice as much…
…and half as many people to punish
So cost of punishing people would be same or lower
But we’d save money on detection/apprehension
Repeating the argument, the “optimal” system is an infinitely low probability of an infinitely severe punishment!
There’s a famous paper by Gary Becker, on the syllabus, that launched the modern economic literature on criminal law
Suppose we had a crime that was punished by a 20% chance of being caught and convicted and a punishment equivalent to $20,000
What would happen if we fired half the police and judges, so the chance of being caught and convicted dropped to 10%; but we raised the punishment to $40,000
Punishing someone to the extent of $40,000 might be more costly than punishing them to the extent of $20,000. But it wouldn’t be more than twice as costly
And there’d be half as many people we needed to punish
So the costs of punishment would be the same or lower
On the other hand, we’d save money on detection/apprehension
So we’d get the same expected punishment – and therefore the same deterrence effect – at a lower cost
Repeating the same argument over and over, Becker concluded that the optimal criminal system was an infinitely low probability of an infinitely severe punishment.
(Becker wasn’t really suggesting we should do this.
For one thing, empirically, criminals respond more to probability of being caught than to severity of punishment
And this is actually what Becker predicted – this has to be true “in equilibrium”, or else we could save money by doing what he said.
But another problem: would juries really be willing to convict you of littering if the penalty were death?)

32. Marginal cost of deterrence
With rational criminals, raising the expected punishment should lead to fewer crimes being committed
On a per-crime basis, raising either the probability of being caught or the severity of the punishment is costly
But as we increase expected punishment…
we get fewer crimes committed,
and maybe fewer offenders we need to detect and punish
So the cost of punishing those criminals we do catch could go up or down
Which means the marginal cost of deterring another crime could be positive or negative!
We argued before that increasing the expected punishment of a crime – by increasing the chance of being caught, or the severity of punishment – is more costly on a per-crime basis
However, as we increase the expected punishment, this creates a stronger deterrence effect, which decreases the number of offenders we have to detect and punish.
Which means that the total cost of administering this punishment might go down. (Or might go up.)
So the marginal cost of deterring another crime could be either positive or negative!
(For an example of how this works, suppose that at some cost, we could come up with a punishment so severe that nobody would ever commit the crime. Then we would never have to impose the punishment, and never have to solve any crimes, so the cost of deterrence would be very low.)

33. “The marginal cost of deterring another crime could be positive or negative”
Sample exam question (Fall 2008 final exam):
Suppose a particular crime is always inefficient: it harms the rest of society $10,000 more than it benefits the criminal.
Every time an offender is caught, he or she is tried, convicted, and imprisoned; the total (social) cost of trials and punishment is $100,000 per criminal caught.
Recall that the aim of criminal law is to minimize the sum of three things:
(1) the social cost of the crimes that are committed,
(2) the cost of detection, and
(3) the cost of trying and punishing the offenders who get caught.
A city is considering hiring additional policemen dedicated to detecting this particular crime. This change would increase the fraction of offenders who get caught from 15% to 20%.

34. “The marginal cost of deterring another crime could be positive or negative”
Social cost of each crime: $10,000
Cost of trial and punishment: $100,000
Increase fraction of crimes detected from 15% to 20%
(a) Suppose this increase in detection would result in a decrease in the number of crimes committed from 1,000 a year to 700 a year.
i. Calculate the effect that hiring the new policemen would have on the social cost of crimes committed.
before: 1,000 X $10,000 = $10,000,000
after: 700 X $10,000 = $7,000,000
effect: $3,000,000 reduction in social cost of crime
ii. Calculate the effect it would have on the cost of trying and punishing offenders.
before: 1,000 X 15% X $100,000 = $15,000,000
after: 700 X 20% $100,000 = $14,000,000
effect: $1,000,000 reduction in cost of trials and punishment
iii. From an efficiency point of view, what is the most that the city should be willing to pay for the new policemen?
$4,000,000, since this is how much social costs are reduced by having higher detection

35. “The marginal cost of deterring another crime could be positive or negative”
Social cost of each crime: $10,000
Cost of trial and punishment: $100,000
Increase fraction of crimes detected from 15% to 20%
(b) Now suppose instead that the increase in detection would decrease the number of crimes committed from 1,000 a year to 900 a year.
i. Calculate the effect that hiring the new policemen would have on the social cost of crimes committed.
before: 1,000 X $10,000 = $10,000,000
after: 900 X $10,000 = $9,000,000
effect: $1,000,000 reduction in social cost of crime
ii. Calculate the effect it would have on the cost of trying and punishing offenders.
before: 1,000 X 15% X $100,000 = $15,000,000
after: 900 X 20% $100,000 = $18,000,000
effect: $3,000,000 increase in cost of trials and punishment
iii. From an efficiency point of view, is there any positive amount that the city should be willing to pay for the new policemen?
No – higher detection increases social costs, so even if the new policemen were free, from an efficiency point of view, we wouldn’t want them!

36. “The marginal cost of deterring another crime could be positive or negative”
Social cost of each crime: $10,000
Cost of trial and punishment: $100,000
Increase fraction of crimes detected from 15% to 20%
(c) Defend the following statement applied to this type of crime:
“Even when detection is cheap, more detection is only efficient if the supply of crimes is elastic.”
When the supply of crimes is inelastic, detecting more of them increases social costs – the number of crimes does not drop much, but more is spent punishing those who are caught.
When the supply of crimes is elastic, detecting more of them reduces social costs – fewer crimes get committed, and fewer criminals need to be punished.

37. Optimal deterrence
So depending on how much the crime rate responds to deterrence, increasing the likelihood of being caught could…
Reduce social costs – by reducing both the number of crimes committed and the number of criminals we have to punish
Increase social costs – by increasing the number of criminals we catch and have to punish (in addition to requiring more spending on detection)
What does this say about the optimal level of deterrence?
Or, if you prefer, the optimal level of crime?
Or the optimal level of punishment?

38. Optimal punishment – example (Friedman)
Suppose there’s some crime for which expected punishment (probability X severity) equals $900
Suppose raising expected punishment from $900 to $901 would deter exactly one crime. Should we do it?
Depends whether social cost of that one crime is more or less than the social cost of deterring it
Suppose that…
Raising expected punishment from $900 to $901 would cost $50
“Marginal crime” does $1,000 worth of damage
To calculate social cost, also need to consider benefit to criminal
Marginal crime gets committed when expected punishment is $900, but not when expected punishment is $901; so benefit to criminal is $900
So social cost of that crime is $1,000 - $900 = $100
If social cost of raising expected punishment enough to deter that crime is $50, we should do it!
This brings us to a rather elegant theory of optimal punishment.
Friedman motivates it with an example.
Suppose there is some crime for which the expected punishment is set at $900
That is, every time the crime occurs, the offender is subject to a “lottery” – a chance of being caught, and a punishment if convicted – which he values as much as a $900 fine imposed with certainty
Suppose we figured out that raising the expected punishment from $900 to $901 would deter one more crime. Should we do it?
That depends on whether the social cost of raising the expected punishment from $900 to $901 is greater or less than the social cost of that marginal crime.
Suppose that raising the expected punishment from $900 to $901 costs $50 more. (We argued that it could increase or decrease total costs; suppose here that it increases them.)
And suppose that the marginal crime does $1000 worth of damage to people besides the offender.
Since we want to consider total social costs, we also have to consider the benefit the offender got from the crime. (This may sound funny – we’ll come back to it in a bit.)
How much value does the offender get from the marginal crime?
The crime we’re deterring is the one that would have been committed when the expected punishment was $900, but will not be committed when the expected punishment is $901
So the value of this crime to the criminal must be between $900 and $901.
So in this case, the social cost of the marginal crime is $1000 – 900 = 100, and the cost of deterring it is $50 – we can prevent a net loss of $100 at a cost of $50, so we should do it.

39. Optimal punishment – general theory
at efficient level of deterrence:
Cost of deterring one more crime
Social cost of marginal crime =
Cost of deterring one more crime
Harm to Victim
Benefit to Criminal = –
since “marginal criminal” is indifferent…
Cost of deterring one more crime
Harm to Victim
Expected Punishment = –
Now, more generally, if we are deterring the optimal amount of crime, it must be true that the net damage of one more crime is equal to the marginal cost to deter one more crime:
Cost of deterring one more crime = Net damage
By definition, net damage is the damage to the victim, minus the benefit to the offender, so
Cost of deterring one more crime = Damage to Victim – Benefit to Offender
We said that on the margin, the last crime we deter will be the one whose benefit to the offender is exactly equal to the expected punishment, so
Cost of deterring one more crime = Damage to Victim – Expected Punishment
Rearranging,
Expected Punishment = Damage to Victim – Marginal Cost of Deterrence
rearranging,
Expected Punishment
Harm to Victim
Marginal Cost of Deterrence = –

40. At efficient level of deterrence,
Expected Punishment =
Harm to Victim –
Marginal Cost of Deterrence
When deterrence is free, expected punishment = damage to victim
Offender internalizes costs of his actions
Just like with tort law, leads to only efficient crimes
When deterrence is costly, expected punishment < damage to victim
When preventing marginal crime is costly, we allow all efficient crimes…
…and some slightly inefficient ones, because it’s cheaper to allow them then to prevent them
When marginal cost of deterrence is negative, we should set expected punishment > damage to victim
When preventing the marginal crime saves money (because there are fewer criminals to punish)…
We prevent some efficient crimes too, because it’s cheaper to deter them than to allow them and have to punish them!
What do we learn?
When deterrence is free, expected punishment should equal damage to victim
this causes the offender to internalize the costs of his actions
this leads to only efficient crimes, just like in civil cases
When the marginal cost of deterrence is positive – when raising expected punishment to deter one more crime is costly – expected punishment should be less than damage to victim
That is, when preventing the marginal crime is costly, we allow all efficient crimes
And we allow some slightly inefficient crimes, because preventing them would cost too much
When the marginal cost of deterrence is negative – when raising expected punishment lowers costs, since there are fewer crimes to punish – expected punishment should be greater than damage to victim
That is, when preventing the marginal crime saves money, we aim to deter all inefficient crimes
And we deter some efficient crimes too, because deterring them is cheaper than having to punish them

41. Aside: why do we count the criminal’s benefit?
Why count criminal’s payoff when calculating social costs?
We said fines cost nothing – make offender worse off, state better off
Why not just say, screw the offender, fines raise money?
And social cost of crime = damage to victim – benefit to offender
Why not: by committing certain acts, you give up right to be counted?
Friedman argues it this way:
We want an economic theory for why things like murder and embezzlement should be treated differently than nuisances and torts
If we start out by assuming they’re morally different, we’re assuming the answer to our question
If we avoid making assumptions like that, and still come up with reasons they should be treated differently, then we’ve learned something
It might seem odd that we continue to count the criminal’s costs and benefits when calculating social costs
That is, we said that fines cost nothing – they make the offender worse off, and the state the same amount better off – rather than saying, screw the offender, fines generate revenue for the state
Similarly, in calculating the social cost of a crime, we included the benefit to the offender, even allowing for the possibility of some crimes being efficient.
You could argue that by committing certain acts, offenders basically give up their right to be counted.
But Friedman argues it this way:
We’re looking for an economic theory of why things like murder and embezzlement should be treated differently than things like nuisances, breach of contract, and torts
If we start out by assuming they’re morally different – that someone breaching a good-faith contract remains a part of society, but someone embezzling his employer’s money has crossed a line and no longer counts – then we’re basically assuming the answer to our question
If we can avoid making assumptions like that, just start with the idea of economic efficiency and still come up with arguments for why these actions should be treated differently, then we’ve actually proven something.
How comfortable you are counting the offender’s benefit may also depend on the type of crime
You might be very willing to concede that if someone breaks your car window to steal $50, the social cost of the crime is just the cost of the window
You might be less comfortable with the idea that the social cost of a crime like murder or rape should include the utility the offender got from committing the act

42. What would efficient punishments look like?
Friedman:
“[The system] would be designed to squeeze the largest possible fines out of convicted criminals, using the threat of more unpleasant alternatives for those who failed to pay.
If the fines that victims can pay, even under such threats, are inadequate, they are supplemented by penal slavery for criminals who can produce more than it costs to guard and feed them, execution (with the organs forfeiting to the state) for those who cannot.
Any prisons that do exist and do not pay for themselves are as unpleasant as possible, so as to produce as much punishment as possible per dollar of imprisonment cost.
It is a consistent picture, and considerable parts of it can be found in the not very distant past, but not a pretty one.”
We already discussed the fact that using fines as a punishment mechanism has very little social cost, and are likely the most desirable punishment for offenders who can afford to pay the fine.
For those who cannot, imprisonment is costly, but could be made less costly by having prisoners do useful work
This has happened in many places in the past – I remember hearing the irony a long time ago that New Hampshire license plates are stamped by prisoners, but bear the state motto, “Live Free or Die” – and still happens in some places
Finally, given the excess demand for transplantable organs, execution could be made less costly for society (maybe even given a negative cost) by taking executed criminals’ organs for transplant
(quote)
He considers a few counterarguments.
Incapacitation and rehabilitation might argue for prison over fines, but would not necessitate such unpleasant prisons
Execution is very costly in the current system, due to extra legal protections, which may seem necessary because of its irreversibility and the horror of executing an innocent person
And some elements of this may just seem wrong to people – forced labor, or confiscation of body parts.

43. The problem with efficient punishments
Punishments are designed to make someone worse off
So if a punishment has social cost close to 0, it must make someone else better off
With fines: state gets the benefit of the money
But this creates incentive for abuse
State benefits from convicting people!
Drug cases and forfeiture
Traffic cameras and yellow lights
Friedman, however, offers another argument against “efficient” punishments:
As punishments get more efficient, this means the state gets greater and greater benefits from punishing people
(The punishment must hurt the people being punished, but have social cost close to 0, which means the state must be getting a benefit)
But this gives a large incentive for abuse
Already, some people feel far too much police effort goes into drug cases – in which forfeiture of dealers’ money and possessions is common – over other crimes that do not come with this “bonus.”
We’ve been pretending the “state” is some perfect, benevolent institution, but in truth it consists of lots of individuals who respond to incentives; making imprisonment and execution too profitable for the state would lead to very creepy incentives.
(Another example of this: on the syllabus, I put two links to news stories about towns using traffic cameras to catch red light violations. Once the cameras are in place, officials realize that catching people is profitable. This gives an incentive to catch more people. The first article lists six cities caught shortening the length of yellow lights beyond what is allowed by law, in order to “catch” more people running red lights.)
(Friedman gives a fictional example where states can harvest organs when they execute someone; over time, more and more crimes become punishable by death.)