Class Outline: Economic Analysis of Law 1997

School of Law

Santa Clara University

D. Friedman

This outline contains what I said, what I think I said, and what I ought to have said, so discrepancies
between what it contains and what you remember are not necessarily due to your poor memory.

I. Mechanics of Course:
A. Written material will consist of:
1. Posner: A big book, by a leading figure in the field, covering much more material than
we will have time for.
2. Lots of articles which will be handed out in class; note that in some cases only part of
the article is assigned.
3. An outline of the lecture, which will be handed out each week for the preceding week,
deus volenti. You're reading it.
B. There will be two midterms.
D. The final will cover everything, and be mostly or entirely essay.
E. Previous year's lecture notes and exams are or will be available on the web page
and in the library.
II. Sketch of course:
A. This week-introduction
B. Part I: Economic Analysis of Law, organized by ideas.
C. Part II: Organized by legal field.
III. What is economics?
A. An approach to understanding behavior, based on the assumption that Individuals
have objectives and tend to choose the correct way of achieving them. "Rationality."
B. Some Examples:
1. Why have locks on doors?
a. A sufficiently determined burglar can get into anything short of a bank vault,
with a sledgehammer and crowbar or through the window.
b. A lock makes it harder, thus more work and a greater chance of being caught,
thus a less profitable project.
c. So we put locks on our doors because we believe that burglars are rational.
2. A policeman has evidence that will convict a criminal; what does he do with it?
a. Convictions costs the criminal $50,000, benefits the policeman's career by
$10,000
b. Dragnet answer: arrest the criminal.
c. Economics answer: Sell the evidence to the criminal for some price between
$10,000 and $50,000
d. To keep this from happening, some of the policemen have to be watching the
other policemen instead of the criminals.
e. The solution proposed by Becker and Stigler was a bounty system, where the
policeman receives as salary the fine paid by the convicted criminal. If he takes a
bribe, he is merely cutting out the middleman.
f. Which is how our civil system works-we call the "bribe" an out of court
settlement.
3. Civil forfeiture
a. Under Federal law and the law of many states, property used in certain crimes
can be seized by the government, without proving that the owner was guilty of
anything.
b. In at least some states, the value of the seized property goes to the police
department that seized it.
c. Which gives police departments an incentive to look for rich people they can
find (or create) evidence against, rather than criminals in general.
d. A similar problem exists with punitive damages.
4. If we impose the maximum possible penalty for armed robbery, the additional penalty
for murder if you are already committing armed robbery is zero, so we have made it in
the interest of armed robbers to kill their victims.
IV. What is the Economic Analysis of Law?
A. First Project: The consequences of legal rules.
1. If we impose the maximum possible penalty for armed robbery, ...
a. The implication is not necessarily that we shouldn't do it-it will deter some
armed robberies
b. But increase the number of armed robberies with murder
c. And we should take that into account in deciding whether to execute armed
robbers.
2. Plea bargaining:
a. It seems as though it should weaken punishments, since a defendant will only
accept a plea bargain if he thinks it is a better deal than the risk of being tried and
convicted.
b. But it may strengthen them. Prosecutors have limited resources; if 90% of
defendants plea bargain, resources can be concentrated on the remaining 10%,
making conviction more likely-and making defendants willing to accept harsher
bargains.
3. Requiring CD players in cars. Seems to benefit consumers-until we take account of
the effect on the price of cars. The same argument applies to requiring hot water in
apartments, safe jobs, ... . Fixing one term of a contract in favor of one party is not in
general a transfer, since prices, or other terms in the contract, shift to compensate.
4. Requiring seat belts. Accidents are the result of rational decisions-how fast to drive
etc. Making cars safer makes risky driving cheaper, which increases the amount of it.
Sam Peltzman actually observed this result by statistical analysis of accident rates before
and after a sharp increase in required safety equipment.
B. Second Project: Predicting legal rules.
1. The Posner Conjecture: Common Law tends to be economically efficient.
2. One basis for this conjecture might be a plausible mechanism by which efficient law
would be generated. In my opinion, Posner has never provided one.
3. An alternative basis might be evidence that the observed rules of the common law
tend to be the same rules that economic theory tells us are efficient. Posner offers a lot of
such evidence. One problem is that the conclusions of economic theory are not always
clear cut, so a clever economist who knows what the rule is might be able to construct
arguments for why it is efficient-and might be able to do so for whichever alternative the
law chose.
4. An alternative approach to predicting legal rules-legislated ones-is public choice
theory, which tries to apply economics to the political marketplace. We will probably not
be covering that in this course.
5. This project is also a positive one-explaining what the law is, not what it should be.
C. Third Project: Evaluating Legal Rules-Efficiency as a norm.
1. Controversial, but ...
2. Economic efficiency is a value, and maybe the one law can best achieve
3. Allocation vs distribution
V. What is economic analysis of law good for? Why should people study it?
A. One attraction of the economic analysis of law, from the standpoint of legal
scholars (or potential legal scholars) is that it ties together diverse areas of law,
providing a common theoretical structure.
B. It is also useful for understanding economics, because you learn about ideas by
trying to concretize them.
1. "Property" sounds like a simple idea-and isn't.
a. Does ownership of land include the air above-to what height? The earth
underneath? Oil in the earth underneath?
b. What can you do to protect your ownership? Physically keep people out (build a
high wall)? Shoot trespassers? Sue trespassers-but only for damage actually done?
2. "Cause an externality" seems like a simple concept-and isn't. When I play loud music
while my roommate wants to sleep, there is a resulting cost-but it would be eliminated
either by my turning off my music or by my roommate deciding to get up and party.
3. Coase revolutionized our understanding of externalities, market failure, etc., in part by
looking at how the common law of nuisance treated such problems.
4. "Cause" is also a difficult concept more generally. Consider the case of two hunters
simultaneously shooting a third-one in the head, one in the heart. Who caused the third
hunter's death? Neither-and both.
C. The law is also useful to economists as a vast pool of interesting puzzles.
D. And as data.
1. How people do things is a fact-evidence.
2. When you prove they would be better off doing things differently, you have probably
demonstrated a flaw in your theory-since people generally know what actions are in their
interest.
VI. I have been offering a broad definition of economics, one that goes far beyond
prices, unemployment, and the like. What is the relation between economics defined in
that way and economics more narrowly defined-what you find in the typical micro text?
A. The latter has been worked out in more detail.
B. and it helps us understand economics in the broader sense.
1. To understand marriage, use ideas developed in explaining long term contracts
between firms. Talking about a marriage market is not merely a metaphor-it is a real
market, even if the prices are not explicit or in money.
2. To understand your three year old's apparently irrational behavior ("I want ice cream
and will throw a tantrum if I don't get it, even though you will then send me up to my
room without any ice cream") think about the logic of commitment strategies in ordinary
economics. If he persuades you that he has committed himself strongly enough, next
time you may give him the ice cream to avoid the tantrum.
C. The broad approach helps us intuit the narrow. We do a lot of rational choosing in
our daily lives, and can apply the intuitions developed there to understanding the
behavior of firms, workers, etc.
VII. What is economic efficiency? (Posner calls it "Wealth Maximization")
A. The problem-evaluating different systems of rules, their outcomes, etc.
1. By whose values?
2. How aggregated?
B. The economic solution.
1. Revealed preference: Heroin or Insulin, the fact that someone is willing to pay lots of
money to get it is evidence that it is valuable to him. How valuable? It is worth to him
the maximum amount he would be willing, if necessary, to pay for it.
2. Aggregate effects across people by using willingness to pay as a common unit. We are
using value (utility measured by dollar willingness to pay) rather than utility (units of
pleasure, or happiness, or something similar). A dollar might represent more utility to
one person (poor-or materialistic) than to another (rich-or ascetic) but it represents, by
definition, the same value. So we count a change that benefits Bill Gates by $10 and
hurts a homeless person by $9 as an improvement, which seems wrong.
3. Note that value measured in dollars is not the same thing as pieces of green paper. If I
am willing to pay $400 for a new printer and buy it for $300, I end up with $300 less
money-but I am $100 better off than before the deal.
C. Defense of the economic solution.
1. If we wish to describe behavior, it is revealed preference, not "value as known in the
mind of God," that is the relevant tool.
2. If we wish to predict outcomes, willingness to pay is more relevant than utility.
Whether lobbying, litigating, or simply buying goods on the marketplace, one man's
dollar will have about the same effect as another's, independent of how much happiness
a dollar represents to each.
3. What about using efficiency as a normative criterion-to decide how society should be
organized?
a. Both elements are then problems, but ...
b. It is not clear one can do better. God is not available to run the society. Other
people are available to make choices for us, but even if they (sometimes) know
what is good for me better than I do, they are less certain to seek my good than I
am. They may take the opportunity to seek their own good at my expense.
c. Many economic choices involve gains and losses to large and diverse groups, in
which case differences in utility/dollar may average out.
d. Even if efficiency is a good proxy for utility, however, is utility all that matters?
If there are utility monsters (people who feel much more intently than the rest of
us do) should we devote all of our resources to their pleasure? If you can save the
lives of three people about to be lynched by framing one innocent person for the
crime, should you do it?
e. Utility is probably not all that matters, but it is one big thing that matters, and if
we know how to maximize it and don't know how to maximize the rest of what
matters (or cannot agree about what it does), then rules that maximize utility (or
efficiency as a proxy for utility) may be the best we can do.
D. Efficient vs Inefficient outcome: "Efficient" is a sort of outer bound benchmark,
the best result we could hope to achieve by any economic system.
1. I define a "bureaucrat god" as someone who knows everything anyone in the society
knows, and has unlimited ability to calculate how to coordinate people's actions and to
make them do what he tells them to.
2. An outcome is efficient if a bureaucrat god could not improve it ("improve" in the
sense we have been discussing-make a change that produces total benefits larger than
total costs).
3. If an outcome is efficient, no change in the economic/legal system can improve it,
since we have no tools available that the bureaucrat god does not also have.
4. If an outcome is inefficient, we might be able to improve it. But we also might not,
since we have no bureaucrat gods available to run our economy.
E. Other definitions of efficiency. [I didn't get to this part in class]
1. Pareto:
a. An improvement is a change that benefits someone, hurts nobody. This
definition avoids interpersonal comparison, but ...
b. In practice, few changes in legal rules and the like are pareto improvements.
2. Hicks/Kaldor:
a. An improvement is a change that would be a pareto improvement if combined
with suitable transfers among those affected.
b. This almost always gives the same conclusion as Marshall's definition (what I
started out explaining).
c. Because if there are net benefits, then the gainers gain enough to compensate the
losers, and if the gainers gain enough to compensate the losers, then there are net
benefits.
d. Actually, there is an exception to this apparently obvious conclusion, explained
in my article "Does Altruism Produce Efficient Outcomes? Marshall vs Kaldor" on
my academic web page.
3. Both the Kaldor and Pareto approach make it look as though you are not trading off
gains to one person against losses to another, when in fact you are. That is why I prefer
Marshall's approach.
E. What matters is that you understand these ideas, not that you agree with them.
1. Accept rationality as a working hypothesis for this course.
2. Take Posner conjecture as a conjecture, look at evidence for and against.
3. Feel however you like on the third project.

1/21/97

I. Externalities: The conventional analysis

A. The simple argument for efficiency:
1. If people who take actions receive the benefits and pay the costs of those actions
2. Then they will take those actions and only those actions for which net benefits>net
costs
3. Which is the efficient rule
4. And this condition is met in the ideal competitive market (for explanation, see a Price
Theory text-preferably mine)
B. The problem:
1. If I can take an action some of whose costs are born by other people
2. For example, produce both steel and pollution, sell the steel, pay wages and the cost of
raw materials, but not pay the cost of the pollution
3. Then I will produce steel even if total costs, including pollution, are > total benefits
for at least some of my output-i.e. produce more than the efficient quantity
4. And I will produce steel in the way that minimizes my total cost, not all total cost-for
example, I will not do costly things to reduce pollution, even if they are, on net, worth
doing.
C. Simple solution-direct regulation.
1. The EPA or equivalent decides who can pollute how much where, what methods of
pollution reduction have to be used, etc.
2. The problem is that to do that right they would have to know both the damage done by
pollution and the cost of all the alternatives-which they probably don't
3. And it is not in the interest of firms to tell them, since a firm that can persuade the EPA
that its output is important and there is no way of reducing the associated pollution at
any reasonable cost gets to keep on polluting.
D. Fancier solution-effluent tax.
1. Charge each firm for the damage done by its pollution.
2. It can decide whether to pollute and pay, control pollution, or stop producing.
3. And it is in its interest to choose the most efficient alternative.
E. There is still an informational problem, since the taxing authority has to estimate
the damage done by the firm's pollution in order to set the tax-but at least it does not
have to estimate the cost of not polluting. It can trust the firms to solve that problem
in their (and our) interest.
F. Both solutions face a further problem-how to make it in the interest of the
regulator/taxing authority to make the rules that produce an efficient outcome.
1. It might be more politically attractive to sell out to the polluters in exchange for
campaign contributions.
2. Or impose very stringent rules in order to be bribed not to enforce them.
3. We will be mostly ignoring this class of problems at this point.
4. But it is a good example of the general point that you can keep expanding the
economic approach to incorporate what you had been interpreting as the framework:
rational cops, judges, legislators, voters, ... .
II. Externalities in the law
A. Direct regulation of pollution-EPA
B. Laws as effluent fees: Speeding tickets.
1. If we wanted to eliminate all speeding, we would hang speeders-or confiscate their
cars.
2. If we believe that speeding imposes some, but not infinite, cost on others, we try to
impose a penalty sufficient to deter speeding-except when speeding is worth enough so
that the driver is willing to pay the price.
C. Damage payments as effluent fees: Tort law.
1. One way of inducing drivers to take precautions is to make them liable for the cost of
the damage they do to others if they don't.
2. By allowing their victims to sue them for damages.
D. Criminal punishments as effluent fees.
1. Crimes impose costs on their victims.
2. By making the criminal pay a punishment equal (on average) to those costs, we deter
crimes, except
3. When the benefit to the criminal is greater than the cost to the victim, so the criminal
is willing to pay the price. Example: Hunter lost and starving in the woods who breaks
into a cabin, steals food, phones for help.
4. In other words, we deter inefficient and only inefficient crimes.
5. An alternative approach, if we think the court can tell when offenses are efficient, is to
special case them-to punish only the inefficient crimes. The lost hunter is excused from
punishment under the "doctrine of necessity."
6. Note that criminal punishments are typically costly. When I pay a damage payment,
someone receives it. When I am executed, nobody gets the life I lose. When I am
imprisoned, nobody gets my liberty and the taxpayers have to pay for the prison.
7. We should modify our rules to take account of this. There may be some offenses that
are inefficient-we would like to deter them if we could do so costlessly-but cost more to
deter than it is worth.
E. Some of these examples involve a payment to the "victim" and others involve
payment to the state. We will see shortly that this difference is sometimes important.
III. Coasian critique of externality theory: An argument in three acts.
A. Nothing works: Two sided causation.
1. A doctor builds a new consulting room onto his house, adjacent to where a
neighboring candy maker is operating machinery. The machinery was no problem
previously, but its vibration now prevents the use of the consulting room. The doctor
sues.
2. As this example shows, "external costs" are typically due to joint actions.
a. Polluting and breathing
b. Using the same lake for a summer resort and a place where a factory dumps
chemicals
c. Operating a recording studio next to a fraternity house that has parties
3. Making either one party liable to the other eliminates the other party's incentive to
solve the problem.
4. A railroad throws sparks that start fires in the adjacent grain fields. One solution is for
the railroad to put on a spark arrester that will prevent the sparks. An alternative is for
the farmers to plant some non-inflammable crop (hereafter "clover") along the rail line.
B. Everything works: The Coase Theorem
1.The doctor wins his his case, but the efficient outcome is to move the consulting room,
not the factory. So the candy maker pays the Doctor to move the consulting room and let
him continue to operate his factory.
2. As long as there are net gains from changing the outcome, there is some way of
allocating the gains that makes everyone better off, so the change should occur. More
precisely:
3. The Coase Theorem: If transaction costs are zero, any initial allocation of property
rights will lead to an efficient outcome.
4. Here "transaction costs are zero" means "if there exists a transaction that benefits all
parties, it will happen."
C. It all depends-on transaction costs.
1. 1 Railroad and 1 farmer-fine. Whatever the legal rule is, they bargain to the efficient
outcome.
2. Railroad, 100 farmers, railroad has the right to throw sparks.
a. And should... the damage done by the fires is less than the cost of putting on a
spark arrester. No problem-sparks
b. And shouldn't-the cost of the spark arrester is less than the damage done by the
fires or the cost to the farmers of switching to clover. We may still get sparks
because of the public good problem faced by the farmers.
c. A public good problem occurs when a good, if produced, will be available to all
the members of a pre-existing group. In this case, if someone pays the railroad to
install the spark arrester, all of the farmers will benefit-whether or not they helped
pay. So each farmer may try to be a free rider-get the benefit without paying for it.
3. Farmers have the right to enjoin.
a. And should...the spark arrester is the lowest cost solution. No problem. No
sparks, and that is the efficient outcome.
b. And shouldn't...holdout problem. Suppose that the cost of a spark arrester is
$200,000, the cost of switching to clover is $100,000 ($1000/per farmer). RR
offers each farmer $1100 to switch. One farmer notices that if everyone else
agrees, his agreement will let the RR save a lot of money-but if he blocks the
settlement (i.e. insists on enjoining the RR from throwing sparks) they will have to
pay for a $200,000 spark arrester. So he tries to hold out for much more than
$1100. So do others. No agreement. No sparks. No clover.
4. Farmers have the right to sue and can collect $3000 damages each, equal to the
damage done by the fires:
a. If no sparks is the efficient solution, you get it immediately-the RR installs the
spark arrester instead of paying damages.
b. If sparks +clover is the right solution, get it after bargaining. A farmer can try to
hold out-but the RR can settle with the other farmers and pay the holdout $3000
for the fires in his fields. The holdout loses out on whatever premium the RR was
paying above the cost of switching to clover.
c. If sparks without clover is the efficient solution, that happens immediately too.
The RR is only willing to pay the farmers up to $3000 each to switch to clover-but
switching costs them more than that (that is why clover is not the efficient solution
in this case).
d. So the damage rule appears to be the best solution-if damages are easily and
cheaply measurable by the court. We have so far ignored both the cost of litigation
and the risk of court error.
e. Note that the first two approaches used values determined by market
transactions; this one introduces an external measure of value--which should be
done with caution. If external parties are entirely competent to measure value, and
can be trusted, we are back with some version of the pure regulatory solution--that
court simply tells everyone what to do.
IV. Some further implications:
A. We should include other alternatives:
1.the fourth option-RR has the right, but protected only by a liability rule. Any farmer
can enjoin sparks-but is then liable to the railroad for the cost of putting on the spark
arrester.
2. Which gives a public good problem if no sparks is the right solution. No individual
farmer wants to enjoin, because he will then pay for the spark arrester while everyone
else gets the benefit of no sparks.
3. And we could consider still other possible legal rules, including one where the RR can
force the farmers to switch to clover-but owes them damages for doing so, and rules
where the damage payments go to the state instead of the other party.
B. Think of the alternative rules as a 2x2 matrix:
1. Who has the right to decide whether sparks are thrown on a farmer's property (RR or
Farmer) and
2. who has to bear the costs of the decision (farmers bear their own costs are RR must
compensate them.
3. Then add a second decision: what does the farmer grow. Who controls, who pays.
C. How we would figure out the optimal property rules, if we knew just barely
enough: The spaghetti diagram.
1. We start with three possible legal rules, four possible outcomes (one, spark arrester +
clover, is uninteresting here, although one could imagine situations where we wanted
both parties to take precautions).
2. We are going to apply our legal rules to a lot of different situations in the future, and
the courts will not know enough to revise the legal rule for each situation. In other
words, we can't say "in any situation where the efficient outcome is sparks, give the
railroad the right to throw sparks, in any situation ... ." We need a single legal rule to
cover many future cases.
3. We have some estimate of how likely different situations (sparks efficient, clover
efficient, etc.) are to arise in the future.
4. So we consider, for each initial rule, how hard it is to get from that rule to each of the
possibly efficient outcomes. We then choose the least costly rule, allowing for how likely
it is that each outcome will be the most efficient (and how much is at stake, and how
costly getting to that outcome from that initial rule will be).
5. Note that the cost of transactional problems that block the move from initial rule to
final outcome is of two sorts.
a. The cost of organizing a transaction-bargaining out a solution to the public good
problem.
b. The cost of failing to organize such a transaction, and thus ending up with an
inefficient outcome.
V. Damages vs fines:
A. The non-strategic case with bargaining:
1. There is a factory polluting a small lake; preventing the pollution will cost the factory
$1,000,000/year.
2. The pollution does $700,000/year worth of damage to the resort at the other end of the
lake-ruins the fishing, people don't want to swim in water that smells bad, etc.
3. Efficient outcome-keep polluting, since prevention costs more than it is worth.
4. The government imposes a Pigouvian tax-every year, the factory must pay $700,000
in fines to the EPA for its pollution. The factory keeps polluting. The resort keeps losing
money.
5. The resort owner offers the factory $400,000/year to stop the pollution-which is a
good deal for the resort, since stopping the pollution is worth $700,000 to it.
6. The factory accepts. It saves $700,000 in fines and gets an additional $400,000 from
the resort, so it is worth paying the $1,000,000 cost.
7. The result is inefficient.
8. Coase + Pigou is too much incentive. "Incentive" is not automatically a good thing;
what we want is the right incentive, neither too much nor too little.
9. Note that the problem will not arise if the $700,000 is a damage payment to the resort
instead of a fine-now the resort doesn't care whether or not the factory pollutes (and pays
for it).
B. The non-strategic case without bargaining-auto accidents.
1. The probability of accidents depends on precautions by both parties.
2. We want each to take any precaution whose payoff in decreased accidents is greater
than its cost.
3. To give them an incentive to do so, we must make each party liable for all of the costs.
He pays his own costs and pays a fine to the state equal to the other driver's costs. The
other driver does the same. Double liability-between them, they are paying twice the
total cost.
4. But note that there may strategic behavior embedded in this situation--dented car.
C. The strategic case: suppose the RR owes a fine to the government for each fire;
further suppose that the spark arrester is the efficient solution, but clover is better
than fires.
1. The RR keeps throwing sparks, pays fines for a while, knowing that ...
2. The farmers, who are suffering from the fires and not collecting the fines, will
eventually switch to clover.
3. Note that this depends on a particular asymmetry-one RR, many farmers.
a. With one of each, the RR has no bargaining advantage. The farmer might refuse
to switch to clover in order to get the RR to put on the spark arrester.
b. With many of each, no strategic behavior. Consider the case of auto accidents. It
doesn't make sense for me to drive recklessly in the belief that everyone else will
be very careful to compensate-because I am only one one-millionth of their
driving environment.
c. But it might make sense to leave the dents in my car, so as to influence the
incentives of drivers close enough to see me.
VI. Fines vs Damage Payments in summary:
A. Strategic argument for damage payment-where parties are identifiable and one
side has a big commitment advantage.
B. "Coase+Pigou is too much" argument for liability.
C. Joint cause argument for fine-liability removes the other party's incentive to take
precautions
D. Liability has the advantage (and disadvantage) of private enforcement.
1. An incentive to report the accident
2. But also an incentive to sue when you have not been injured-in order to collect.
E. Think about the difference between bargaining, which requires only low
transaction costs, and strategic behavior--which depends, in part, on a breakdown in
bargaining.
1. Indeed, it can be viewed as a transaction cost--RR case above.
2. Note that strategic behavior involves my acting so as to influence your incentives to
act--and exists even in the simple case of driving a dented car.
3. It makes analysis much more complicated; we will discuss it a little later.
VII. Other Coasian stories:
A. Suppose rights with regard to polluting the lake are well defined but for some
reason the factory and the resort find it hard to bargain to the efficient solution.
Perhaps there are many resorts. Perhaps it is hard for the resort to tell how much
pollution the factory is putting out.
B. One solution is for the factory to buy the resort, or vice versa. Once they belong
to the same firm, there is no longer an externality-the cost of pollution and the cost
of pollution control are being paid by the same people.
C. Generalizing the argument provides an explanation of why we have firms-which
was the subject of Coase's other famous article:
1. If you have taken an intermediate micro course, you know that trade on the market
provides the perfect system for decentralized coordination. So why have firms? Why not
do everything by contract?
2. Suppose we do it that way (the "Inside Contracting System," which was widely used
in the 19th century). Each stage of production is being done by a separate firm or
individual, with contractual agreements among them. "I will deliver ten assembled rifle
actions per day for you to put into the stocks you are making, and I will get $7 per action
from what he eventually sells them for."
3. Something goes wrong. One party fails to fulfill his contract, imposing costs on
everyone else up and down the line, which he may not have the resources to reimburse.
4. More generally, the market solution to coordination is limited by transaction cost
problems of various sorts, while the hierarchical solution (the ordinary firm) is limited
by the problem of controlling people by command (the worker's objective is to maximize
his welfare, not the firm's profits-only if the worker owns the firm, as he does in a system
where every worker is his own firm contracting with other such firms, are the two
objectives the same), the informational problems of centralized authority, etc.
5. Think of the firm as a balance between the problem of transaction costs and the
problem of hierarchical control. The bigger the former problems are the larger firms will
be, the bigger the latter problem, the smaller firms will be.
VIII. Coase vs Pigou-a summary:
A. In some contexts, such as traffic accidents and much pollution, we have large
numbers on both sides, so no Coasian negotiation is practical.
1. So property rights will end up where they start,
2. So we want the "right" Pigouvian solution-which requires an opinion on who can best
adjust.
3. And might mean fining both parties to make both adjust-the auto accident case.
B. In some contexts, it is obvious who is the lowest cost avoider-so the Pigouvian
solution is correct. But watch out for the Coase+Pigou problem if transaction costs
are low.
C. In contexts where transaction costs are low and it is not obvious who is the
lowest cost avoider, the Coasian solution is best-define rights clearly and let the
parties bargain to the efficient solution.
D. The Coasian approach gives the same double incentive on the margin as does the
double liability solution I described for auto accidents. My incentive to take a
precaution that saves both me and you costs is the sum of my savings plus what you
will, if necessary, pay me to take the precaution-and similarly for your taking
precautions. So on the margin, each of us takes account of all costs and so takes all
precautions that are worth taking.
IX. Another take: It's all a matter of information
A. Consider the RR/Farmer situation with a fully informed court.
1. If you scrape my car and instead of having it repainted I wait a year until it has rusted
through then sue you for the value of the car, I will lose; the victim of a tort is obliged to
take reasonable care to minimize the damages.
2. So if the court knows that clover is cheaper than fires, farmers who have fires due to
sparks will only be able to sue for what it would have cost them to switch to clover-that
being the reasonable way of minimizing the damages.
3. The reason this usually does not work in practice is that the court can observe the fires
and the damage done, but is unlikely to have enough information to figure out the cost of
all the alternative precautions the farmer could have taken.
B. More generally, if a court knew everything relevant to figuring out how people
ought to act (i.e. what action-spark arrester, clover, checking your car brakes,
driving at exactly 43 miles an hour, operating a candy factory only for seven hours a
day, having a party at the frat house this Saturday but not next Saturday because
next saturday is the best time for the recording studio next door to tape a new album,
...) it could solve all legal problems by telling everyone to do the right thing and
hanging those who didn't. Everyone will obey, nobody will be hanged, so we get the
perfect outcome with no litigation and no punishment.
C. This is analogous to the central planning solution to running a society-or the
direct regulation solution to pollution problems. Figure out what everyone should do
and make them all do it.
D. But since in practice the various parties to the dispute have private information
that the court does not have, we try to use an incentive system instead-Pigouvian
taxes, Coasian bargaining, or whatever-to make it in their interest to use their private
information to choose the correct action.
E. Why don't we solve the whole problem this way-set up legal rules under which
the court does not have to know anything, because it is in the interest of all parties to
use their information to produce the right outcome?
F. Because we don't know how to do it for the hard cases.
G. For ordinary markets that is just what we do. All measures of value are
determined by market transactions reflecting the values of the participants. But when
we have to deal with externalities, public goods, transaction costs, etc. it becomes
hard, and perhaps impossible, to set up institutions that work this way.

1/28/97: Insurance

I. Insurance: why it is interesting:

A. Because one recurring legal issue is who will bear the costs if something goes
wrong--i.e. who is insuring whom. For example:
1. Contracts. Something changes, so either I don't want to produce what I agreed to or
you don't want to buy it. Who bears the costs resulting from such problems?
2. Auto accidents: Who pays for them?
3. Product defects and the associated costs. When a coke bottle explodes, injuring
someone, is Coca Cola liable for the resulting medical costs?
4. Wider political issues of welfare, unemployment insurance, health insurance, etc.
B. All of these involve the same set of issues, as we will see.
C. Starting with straightforward private insurance, which provides the simplest
introduction to the analysis.
II. Why insure?
A. In expected value terms, insurance is a losing deal for the customers--on average,
they pay out more in premiums than they collect in claims. If that were not true,
insurance companies would go broke, since they not only have to pay claims but
also other expenses (salaries, rent, advertising, ...).
B. But a dollar as not a dollar is not a dollar. Insurance may be a bad deal if you are
calculating in dollars paid out and collected on average, but a good deal in terms of
what those dollars buy.
C. Consider two futures:
1. 50/50 chance of $10,000/year or $90,000/year
2. or a Certainty of $50,000/year
3. Which do you choose?
4. Why?
D. Consider the goods you would buy in each case. The goods you would give up if
your income dropped from $50,000 to $10,000 are probably a lot more important to
you than the extra goods you would buy if it rose from $50,000 to $90,000.
III. The economics of risk aversion:
A. On average, an additional dollar is worth less to you, in terms of the importance
to you of what you use it for, the more dollars you have. This is the same reason we
were not entirely happy with the solution to interpersonal comparisons that we used
in defining efficiency--compare value measured in dollars--but this time in an
intrapersonal context.
B. Where revealed preference solves the problem of how to measure the difference
C. And people buy insurance--
1. precisely because a dollar in the future where their house burns down is worth more to
them
2. than a dollar in the alternative future where it doesn't.
D. This argument implies that you will insure against large risks but not against
small risks, since the difference in value to you between the goods you would buy
with an extra dollar if your income was $49,990 and the goods you would buy with
an extra dollar if your income was $50,010 is small.
IV. Von Neumann and Morgenstern formalized this argument in the form of Von
Neumann utility--which is defined in such a way that individuals, in choosing among
alternative gambles, maximize expected utility, rather than expected income. In other
words, "utility" already has built into it the fact that dollars are worth more to you in some
situations (when you don't have very many of them) than in others.
A. The varying value of a dollar shows up as declining marginal utility of income.
B. Also known as "risk aversion."
C. Which does not mean "fear of uncertainty" but simply "declining marginal utility
of income."
D. Someone could perfectly well be risk averse in money (he prefers a certainty of
income X to a lottery whose expected value is X--where X was $50,000 in our
earlier example) yet risk preferring in years of life (he prefers a 50/50 gamble of
zero or 20 years to a certainty of 10 years). He has declining marginal utility of
income, increasing marginal utility of years of life.
E. the figure below shows that declining marginal utility of income (the slope of the
utility function becomes flatter as income gets higher) implies risk aversion (the
utility of the average payoff from a gamble is higher than the average of the utility
of the possible outcomes). In the example shown, the gamble is a 50/50 chance of
$10,000 or $90,000. The utility of the average outcome ($50,000) is greater than the
average of the utilities of the outcomes .

V. One problem with insurance: Moral Hazard. [Price Theory Link]

A. I insure my factory for 100% of its value, then stop taking (costly) precautions to
make sure it won't burn down. If it does burn, that is the insurance company's
problem, not mine.
B. Solution--only insure for 80%.
1. The result is still inefficient.
2. If I can reduce the expected loss from fires by $100 at a cost of $10, I will--since I am
saving myself $20.
3. But if it costs $30 I won't--and should.
C. Or have the insurance company mandate precautions.
D. We have been here before:
1. Moral hazard is another name for an externality--voluntarily adopted.
2. It results in the usual inefficiency
3. Mandated precautions are the regulatory solution--but by someone with a private
interest in mandating the right precautions, which might not be true for the EPA.
E. Presumably risk aversion gains are worth inefficiency cost, or the customer
wouldn't buy the insurance policy, but ...
1. Not if others bear some of the moral hazard's cost.
a. Consider auto accidents. If I am carrying liability insurance, the risk of having
to pay for the damage done by my risky driving will not deter me, so I will drive
more carelessly--harming others.
b. That does not matter if the legal rule is such that I end up paying the full cost of
my dangerous driving, but ...
c. Maybe we should ban insurance if we don't have adequate internalization via
damages, fines, etc.
d. And, in fact, we do ban insurance against criminal punishments.
2. And not if insurance is not the result of a voluntary transaction:
a. Either mandatory--as in assigned risk pools. If you are such a bad driver that no
insurance company wants to insure you at any price you are willing to pay, in a
state where insurance is required to drive, you get put in the assigned risk pool,
which is divided up among the insurance companies. They have to insure you, at a
price at which they would rather not. A system for keeping bad drivers on the
road.
b. Or implicit in the legal system, as in legal rules that you may not contract out of
(non-waivable contract rules) or cannot (tort rules for auto accidents).
F. In some cases, "moral hazard" is a feature, not a bug, either because
1. We want people to engage in risky behavior that produces positive externalities. Give
cops life and medical insurance. Or ...
2. Because we are transferring the risk to someone more competent to deal with it:
a. Insurance company vs employee of large firm.
i. Company has more expertise
ii. and better incentives, due to inadequate bonding of the employee.
iii. The employee in charge of keeping the building from burning down
skimps on precautions. Ninety-nine times out of a hunded nothing happens,
and he gets a raise for keeping costs down. One time out of a hundred the
company loses a ten million dollar factory--and fires the employee. A better
gamble for the employee than for the firm.
b. I read an article arguing that the failure of this mechanism was the reason that
nuclear reactors have poor safety design:
i. In other risky fields, safety engineering research is produced by the
insurance companies that insure chemical factories etc. and want to know
how to make sure they don't end up paying out a lot of money due to major
accidents.
ii. Nuclear reactors are largely insured by the Federal government.
iii. Their concern is not convincing experts--safety engineers working for
insurance companies--that they are safe, but convincing amateurs--mayors
and city councils and citizens of towns where a reactor is to be located--that
the reactor is safe.
iv. Which is done by having big switches, big dials, the sort of engineering
that is well suited for science fiction movies but makes it hard for the guy on
duty to actually keep track of what ishappening.
c. Sharecropping--in the case where the landlord is also providing inputs of
capital, expertise, etc. The landlord receives a fixed fraction of the tenants' crop as
rent. [Price Theory Link]
d. Note that this is the double incentive problem again--ideally we would like each
of the two (landlord and sharecropper) to receive 100% of any increase or
decrease in output.
e. We could do it via a third party risk bearer--but landlord and sharecropper
would then conspire to produce an inefficiently high output.
3. A different way of putting this is that we have been implicitly assuming that the
person who bears the cost if the legal system does nothing (lets costs lie where they fall)
is also the person best able to control the cost, so transferring it to someone else
increases the inefficiency. Sometimes it might be the other way around.
a. Exploding coke bottles.
b. Suppose the only way of preventing the problem is better quality control by the
manufacturer.
c. If so, transferring the cost to the manufacturer solves a moral hazard problem
instead of creating one.

VI. The other problem with insurance: Adverse Selection.

A. The market for lemons. [Price Theory Link]
1. There are two kinds of used cars--creampuffs and lemons. The potential seller knows
which he has; the potential buyer does not.
2. Both would rather own a creampuff, so the price a buyer will pay for a creampuff (if
he knows that is what he is getting) is higher than for a lemon, as is the price an owner
will require before he is willing to sell.
3. Buyer actually offers a price representing a weighted average of the value of the two.
4. Suppose cream puff is worth $4000, lemon is worth $2000, half of the potential sellers
have creampuffs.
5. Buyer offers up to $3000, but ...
6. That is a good price for a lemon, so all owners sell, a bad price for a creampuff, so
few owner sell.
7. So most of the cars actually purchased are lemons.
8. So buyers, anticipating that problem, change their weights and offer only $2500, and
...
9. Even fewer creampuffs sell.
10. We could end up with only lemons selling--at lemon prices.
11. This is inefficient-because cream puffs (as well as lemons) are worth more to
potential buyers than potential sellers.
12. But the fact that the seller accepts your offer is evidence that he is selling a lemon.
So you only offer lemon prices. So cream puffs don't sell.
B. Possible solutions:
1. If everybody knows, no problem.
a Require repair records?
b. Permiting repair records will do, if they can be authenticated. It is in the interest
of someone selling a creampuff to produce the records that prove it is a creampuff.
So if you don't produce the records, the buyer knows it is probably a lemon.
c. The analogous approach to the AIDS problem is to have someone produce a
reliable ID card, certifying that the bearer was HIV negative as of three months
ago. People who are negative have an incentive to buy and show the card--people
who are positive can't.
2. If nobody knows, no problem. Not an option in the auto case, but might be in other
cases.
3. So the problem is assymetrical information.
4. Guarantee the car.
a. Which brings us back to insurance problems
b. And double causation. Now that breakdowns are insured against, the owner has
less incentive to take precautions against them.
c. Ami Glazer's solution to the lemons problem: Find a car you like; ask the dealer
if, for an additional sum, he will guarantee it. Repeat until someone says yes. Then
buy the car--without the guarantee. The fact that he was willing to guarantee the
car is evidence that he doesn't expect it to break down--unless he has correctly
 guessed your strategy.
C. Selling life insurance.[Price Theory Link]
1. If someone comes into your office and demands to buy a large life insurance policy
right now, you don't sell it to him
a. because you suspect he knows something about his odds of collecting that you
don't.
b. Someone has just taken out a contract on him
c. Or he tested HIV positive
d. Or his girlfriend jilted him, and he is planningto fall off a cliff this afternoon.
2. More generally, the customer has some private information relevant to how likely he
is to collect.
3. His decision to buy signals you that the private information is in the direction of
making him a bad risk, so you adjust the price accordingly.
4. Note that we apply exactly the same logic when we ask somebody to signal-- "If
you're so sure, what will you bet?" Putting your money where your mouth is is a form of
signalling.
5. And betting that you are going to die early (i.e. buying life insurance) signals that that
is what you believe.
C. Or health insurance
1. You know whether your hobby is reading science fiction or kickboxing or jumping out
of airplanes
2. Which is relevant to how good a risk you are
3. And your buying signals that you are a poor risk.
D. Or anything else with assymetric information.
E. Again, the problem disappears if everyone knows or if nobody knows
1. Suppose we can identify people prone to various medical problems by genetic testing
(as, increasingly, we can).
a. If nobody knows, there is no adverse selection problem--both buyer and seller
of insurance treat the buyer as a random individual.
b. If everybody knows, there is no adverse selection problem--and people with
unhealthy genes pay more for their insurance.
c. In the latter case, having unhealthy genes is an uninsurable risk--because you
know how the gamble turned out before you buy the insurance.
d. We could prohibit genetic testing, whether by customers or insurance
companies. But how do we make the prohibition stick outside the U.S.? You can
always take a trip to Mexico, get tested, then buy or not buy insurance.
2. This is analogous to the economic argument for welfare. Risk aversion is normally
dealt with by market transactions--buying insurance. But you can't insure against being
born poor--because by the time you are there to buy the insurance, you already know
how the gamble turned out.
3. Suppose we forbid insurance company from discriminating by gender in the price of
life insurance--they must sell at the same price to men and women, even though women
on average live longer.
a. Men and women can still use their information about life expectancy to decide
whether to buy
b. So we get adverse selection. Men buy too much life insurance, since they are
getting it at less than its real cost (given their greater chance of collecting early)
and women buy too little.
c. In this case, we have the effect of assymetric information because one side of
the transaction is forbidden from using information that both sides have.
F. Note also that this is an efficiency problem, not (only) a fairness problem.
1. If you have a cream puff that is worth $3000 to you and $4000 to me, but it does not
get sold because of adverse selection (I only offer $2000 because I believe that it is
probably a lemon) we are on net $1000 worse off.
2. If I get insured at too high a price (because the insurance company thinks I am a bad
risk) that is only a transfer, but ...
3. If I don't get insured because the price the insurance company charges (knowing that
people who buy insurance are likely to be bad risks) is more than the price I will pay
(knowing that I am a good risk) that is a net loss. Ditto if women don't get insured--
because they have to pay a price appropriate to a market where 70% of the customers are
men (because men are getting a better deal).

VII. Application to product liability. Who is liable for damage done by exploding Coke
bottles?
A. Consider two alternative legal rules:
1. Caveat Emptor (latin for "let the buyer beware"): You take your good as it is, defects
and all. If the coke bottle explodes, the injured consumer pays his own bills.
2. Caveat Venditor ("let the seller beware"): If something goes wrong with the product,
the seller is liable for the damage.
3. I am treating the distinction as sharper than it really is. If your wife shoots you with
your gun, something has gone wrong, but the fact that a gun shoots the person it is
pointed at does not generally count as a product defect, even if it happens to have been
pointed at the wrong person. So implicitly we are talking only about some subset of
"things going wrong" that are plausibly associated with something wrong with the
product.
4. But that can still be a large and hazy category.
B. Insurance, risk spreading argument: The Coke company can self-insure, the
customer cannot.
1. If Coca-cola is selling ten billion bottles a year, of which one in a million will blow
up, they can count on almost exactly 10,000 exploding coke bottles per year. It is still a
cost, but not much of a risk, since the size of the cost can be predicted with accuracy.
2. So transferring the cost to Coca-cola reduces the risk aversion part of the cost. Before
the change, the customer faced a very uncertain cost. After the change, neither party
faces a very uncertain cost.
3. This only works if the events in question are independent. One bottle blowing up does
not affect the probability that another one will.
4. On the other hand, Coke cannot self-insure against the risk that legal rules will shift
against them. If courts decide to award much larger damages to consumer injured by
exploding Coke bottles, that will apply to all of the bottles, so the effect does not average
out.
C. Moral hazard argument:
1. If the relevant margin for control is care in manufacturing, then liability (caveat
venditor) reduces the moral hazard problem.
2. If the relevant margin is careful use by the buyer, then caveat venditor increases the
problem.
D. Adverse selection argument:
1. What do we assume about consumer information?
2. If consumers cannot observe quality or judge by reputation, then we have a lemons
problem with caveat emptor--everybody produces a below optimal quality, because a
higher quality, being unobservable, cannot command a higher price.
3. A different way of saying this is to consider a situation where the seller knows how
safe the coke bottle is, the buyer does not.
a. If he sells me a coke bottle, I don't know what I am buying (since I don't know
the risk of explosion and do care), so don't know whether it is worth the price
b. I buy too much if it is riskier than I think (since I am underestimating the true
price--or, if you prefer, overestimating the true value), too little if it is safer than I
think.
c. If he sells me acoke bottle with a guarantee, he knows what he is selling, since
he knows the chance that he will have to pay off on the guarantee.
d. And I know the value to me of what I am buying. I don't know the chance of an
explosion--but I don't care, since he has insured me against that risk.
4. If consumers are well informed about average quality by firm, then caveat emptor is
fine from both an adverse selection and a moral hazard standpoint. No moral hazard,
because Coke has an incentive to provide good quality control so that consumers will
want to buy their product.
5. Adverse selection can also apply on the customer's side. Under caveat venditor, I don't
want to sell a car to someone who will use it for drag racing, then sue me when there is
an accident.
E. So far it looks as though we can simply balance risk aversion, moral hazard, and
adverse selection to choose the optimal rule.
1. Either the optimal broad rule-- "producers are always liable," "producers are never
liable"
2. Or have the court look at the product and decide which rule is preferable.
3. Or have the court look at the particular facts, and decide which rule is preferable
under those circumstances. Coke is liable unless you were shaking the bottle. This is the
approach embodied in negligence rules.
4. As we go from more general to more specific, the amount of detailed knowledge the
court needs increases, and the predictability of the result decreases.
F. The conclusion seems to be:
1. that if consumers are well informed about average product quality by firm, or can
judge individual products, the optimal rule is caveat emptor.
2. Ditto if the main risks comes from decisions by consumers (shaking a warm coke
bottle up and down while making a point in an argument--you can't idiot proof
everything, because idiots are too damn ingenioius).
3. Caveat venditor is the optimal rule if consumer is poorly informed both about
individual product and firm average, and the important decisions relative to accidents are
decisions by the firm.
4. And the advantage tilts towards caveat venditor if the loss is large and the producer is
much better able to self insure than the consumer. Note that not all imaginable cases
involve a small buyer and a big seller--consider GM as a buyer of labor.
5. But ...
F. Litigation cost and insurance:
1. So far, we have not been talking about insurance in the literal sense--we have a coke
company, a customer, but no insurance companies.
2. Why not? Instead of asking whether people want protection against risk, why not ask
whether tort liability is the cheapest way to produce it? What about ordinary insurance?
3. It has two big advantages, one disadvantage.
a. An insurance company wants a reputation for paying out readily; a firm may
well want the opposite reputation. A reputation for being tough might make
consumers less willing to buy the product--but in order to justify caveat venditor,
we must assume that consumers are poorly informed. A reputation for being tough
makes injured consumers (and their attornies) less willing to sue--and tort attornies
may be well informed about the litigation reputation of the firms they consider
suing.
b. Risk aversion doesn't depend on whose fault it was, so product liability is lousy
insurance. What if you get hurt and it wasn't due to a faulty product? You are
better off buying insurance that covers injuries from a wide range of causes.
c. Disadvantage--product liability may put the incentive somewhere useful (on the
firm that could take greater precautionsnot to make defective products), insurance
doesn't (except to the extent that your health insurer mails you medical tips, etc.)
4. One advantage of letting the loss lie where it falls is that nobody has to sue anybody,
so you don't have any litigation cost.

VIII. We have been running through arguments about how to decide whether or not
the law should force Coke to "insure" its customers against exploding Coke bottles.
Why doesn't the law make these calculations for ordinary insurance, deciding for
you whether you are insured or not?
A. Because under ordinary circumstances, you have the right incentive to make
them for yourself. You get risk protection, pay moral hazard plus administrative
costs.
B. Where that is not true, there is an argument for government insurance:
1. Mandatory auto insurance--to make sure that if you cause an accident, you can pay for
it.
2. National health insurance--if we believe that the rest of us won't let you die in the
street, so your decision not to get insured may impose costs on us.
3. Or if we think that the adverse selection problem, which gives an inefficient outcome
on the private market, is sufficiently serious.
4. This is a public approach analogous to a group policy, which is a private way of
controlling adverse selection.

IX. Why not apply the same analysis to tort liability?

A. Instead of asking "when should the firm be liable"
B. Ask "when will the firm choose not to be liable although it should be, or vice
versa?"
C. In other words, the alternative to both caveat emptor and caveat venditor is
freedom of contract.
D. The court sets a default rule, which applies if the parties do not specify some
other rule of liability.
1. If the default is caveat emptor, the seller can offer a guarantee, thus shifting the rule
(for whatever sorts of defects the guarantee covers) to caveat venditor.
2. If the default is caveat venditor (malpractice liability), the buyer can agree to sign a
waiver of liability committing him not to sue, thus converting the rule to caveat emptor.
3. Under current U.S. law, 1 is normally enforceable, but 2 often is not. You can agree
not to sue your physician for malpractice, perhaps in exchange for a lower price--but if
something goes wrong you can change your mind and sue anyway.
E. This is an issue that will appear again and again.
1. One reason for legal rules is to fill in gaps that parties cannot, or do not bother, to fill
in themselves.
2. A different reason is to compel rules that the parties would not choose.
3. Different sets of arguments apply to the two cases:
a. "What is the rule that maximizes their joint benefit?" Make that the default--
because if you don't, you are merely putting them to the trouble of contracting
around the default rule.
b. "Why is it not in their interest to choose the optimal rule?" That tells you what
direction you want the mandatory rule to alter their behavior in, and what attempts
by the parties to avoid the mandatory rule you have to guard against.
4.When you are writing waivable rules, "fixing" inefficiencies due to externalities etc.
isn't really an option, because if you try the parties will contract around your rule.
F. Why might you want a non-waivable rule?
1. Because you believe the parties are incompetent
a. Children are not free to sign binding contracts.
b. Nor people who are senile, or feeble minded.
c. Perhaps you believe that waiving a particular rule is so obviously against one
party's interest that the fact that he wants to waive it proves he isn't competent to
make decisions in his own interest.
d. This belief is likely to be more common in judges who don't understand the
relevant economics--the fact that accepting unfavorable (but efficient) terms in one
term of a contract may produce net gains for you via the compensating changes
you can then get in other terms.
2. Because there is some form of market failure--signalling problems, for example.
a. Suppose some doctors know they are incompetent, and thus very much want
their patients to sign malpractice waivers.
b. Patients (correctly) interpret a request for such a waiver as a signal of
incompetence, and go to another doctor.
c. So you might end up getting malpractice liability via freedom of contract, even
though it was inefficient. In which case a non-waivable rule limiting the doctor's
liability might be efficient!
3. Because something other than efficiency matters--child sales. Organ sales. Blasphemy.
4. Because other parties' interests are involved.

2/4/97: Strategic Behavior

I. Strategic behavior: The problem (see my webbed chapter)

A. Bilateral monopoly.
1. Economic example: I have the only apple, worth $1 to me. You are the only customer,
and value it at $2. If we can agree on a price, there is a net gain of $1 to be divided
between us, with the division implied by the price.
2. My six-year-old threatens to throw a tantrum if she doesn't get her way. Doing so
imposes net costs, so there is a gain to finding some mutually acceptable outcome. Don't
assume you can simply be firm and always win. You may have thought out the logic of
bilateral monopoly bargaining better than she has, but she has a hundred million years of
evolution on her side--during which offspring who succeeded in getting a larger share of
parental resources were more likely to survive to reproduce.
3. Doomsday machine.
a. The idea. Lots of very dirty bombs buried under the Rockies. If the Russians
attack, the bombs go off and the fallout kills everyone on earth. So the Russians
won't attack, and we don't need bombers and missiles.
b. The reality: Our (and their) nuclear systems were doomsday machines, with
human triggers. They worked, and therefore were not used--fortunately.
4. Bully: If you train yourself to punch out anyone who gets in your way, and people
know it, they will stay out of your way, and you don't have to fulfill the commitment--
until you run into someone else following the same strategy, and one of you ends up
dead. A doomsday machine on the individual level.
5. In general, the outcome of such games depends largely on issues of commitment and
reputation, which are hard to include in our analysis.
B. Prisoner's Dilemma
1. Simple: Both prisoners are better off if both keep their mouths shut, but given what
one does, the other is always better off confessing. So they both confess.
2. Iterated. If we play the game multiple times, you might think that a prisoner would
keep his mouth shut one time, for fear his partner would betray him next time in
revenge.
3. Why it doesn't work:
a. On the last play, no further threats of retaliation remain, so we are back with the
original game--and both players betray.
b. But since I know you are going to betray me on the last play, there is no cost to
me to betraying you on the play before last.
c. Repeat. The whole series of plays unravels, and we both betray every time.
d. It seems intuitively wrong, but logically right.
4. Why it does work:
a. In the real world, we are playing such games an indefinite number of times
b. And want a reputation for being people who don't betray their partners, so that
people will be willing to work with us, and ...
c. Not betray us, since they want a reputation for not betraying honest partners.
d. All of which is hard to model.
5. Plea bargaining and armies running away are examples of multiplayer prisoner's
dilemmas.
a. The defendants would all be better off if none of them copped a plea, since the
prosecutor would have to divide his limited resources among many cases, and
would get few convictions.
b. But, given what everyone else is doing, it is in my interest to accept a plea
bargain if it is a little better than my expected result at trial.
c. Almost everyone accepts, leaving the prosecutor enough resources to make the
expected result at trial pretty bad for the defendant.
d. Armies run away because each soldier is better off running, although all of them
are better off if all stand than if all run.
6. Tit for tat and experimental game theory.[See Price Theory Link]
a. Rappaport programmed a computer to run a prisoner's dilemma tournament.
b. People submitted strategies in the form of computer programs. Each strategy
played each other strategy a hundred round game of iterated prisoner's dilemma
(i.e. 100 consecutive games--which can be considered a single game of 100 round
prisoner's dilemma).
c. The winning program was a very simple one, called "tit for tat." Always
cooperate (don't confess), unless the other party betrayed you (confessed) on the
previous play.
C. In my view economics should, and generally does, avoid including strategic
behavior in its analysis wherever possible.
1. On the other hand, a lot of journal articles involve game theory nowadays.
2. I think too many.
3. But some should, because ...
D. You can't always avoid strategic behavior:
1. RR example: throw sparks, despite fines, to get the farmers to switch to clover.
2. Punitive damages example: Beat people up, despite fines, to get them to do what you
say next time (there are complications which will get discussed later).
3. Why do some cases get tried instead of settling out of court, and thus holding down
litigation costs?
a. one reason is because the two sides disagree about the likely outcome.
b. But another is because each side is trying to get a settlement on its side of the
bargaining range--and the result is no settlement.
4. Nuisance suits: Sue in order to settle.
5. Strategic imposition of costs: Threaten that if they don't settle, you will make the trial
costly for them--by requiring the company president to spend three days answering your
questions.
E. "Solutions." These are discussed at much greater length in Chapter 11 of my Price
Theory book, on the Web. The basic problem is to come up with a clear definition of
what a "solution" to a game is.
1. Von Neumann solution to two person, fixed sum game. A pair of strategies and a
value V for the game, such that if player A follows his strategy he will get at least V
whatever B does (and maybe more), if player B follows his strategy he will lose at most
V whatever A does (and maybe less).
2. Von Neumann solution & the Core are solution concepts for multi-person games:
Each is a solution concept where one solution can include many outcomes--perhaps an
infinite number. There may be multiple Von Neumann solutions. There is at most one
core--but may not be any.
3. Nash equilibrium: Given what everyone else is doing, I am doing the right thing--and
the same is true for every other player.
a. Unstable against cooperation. Driving on the left side of the road was a Nash
equilibrium for Sweden-but one day everyone switched. Consider a mob of
convicts escaping from death row, facing a guard with one bullet in his gun. Stable
against any single individual, unstable against any pair who decide to rush the
guard.
b. What the equilibrium is depends on how you define a strategy.
c. In the case of oligopoly, does "I adjust, keeping what everyone else is doing the
same" mean "when I change my output, everyone else keeps his output the same
and sells it for what he can get" or "when I change my price, everyone else keeps
his price the same and sells what he can at that price?"
d. The two different definitions of strategy--quantity and price--give very different
predictions for the workings of an oligopoly.
4. Subgame perfect equilibrium. Assume away commitment. If at any point in the
decision tree, someone at that point would never find it in his interest to make a
particular choice, eliminate that choice from the tree. This is how we proved that
repeated prisoner's dilemma gave the same result as a single play of prisoner's dilemma.
5. None of these solutions is very satisfactory, with the possible exception of the first--
which only applies to a situation where there really is no strategic behavior.

II. Examples:
A. Scissors, Paper, Stone: a 2 player zero-sum game.

The Von Neuman solution: Each player rolls a die each time. 1-2 Scissors, 3-4
Paper, 5-6 stone. If I do that, then whatever you do I will, on average, win 1/3rd of
the time, lose 1/3rd, tie 1/3rd, for an expected payoff of 0. I cannot do better than
that because you can hold my payoff to zero by playing that strategy, and you
cannot do better because I can hold yours to zero by playing that strategy. So we
have a mixed strategy that is the VN solution

B. Majority Vote: A, B, and C vote on how to divide a dollar among them:

1. A and B agree to divide it (.5,.5,0). Then C proposes to B that (0, .6,.4) is better--and
B agrees. Then A proposes (.5, 0, .5) and C agrees. Then ...
2. We are stuck in an infinite cycle--any proposal is dominated by another, meaning that
two of the three prefer the other, and have the votes to get it.
3. VN solution is defined as a set of divisions such that none of them dominates another,
and every division not in the solution set is dominated by some solution in the set.
4. One such solution, but not the only one, is (.5,.5,0),(.5,0,.5),(0,.5,.5)
5. Not a very useful solution concept, since not only does it include multiple outcomes,
but there turn out to be, for this game, an infinite number of solutions, most of which
consist of an infinite number of outcomes. An example is the set of imputations (1-x,x,0)
for all values of x between 0 and 1.
C. Two successive plays of prisoner's dilemma: Consider it as a decision tree.

First Step

Second Step

At the bottom of the tree, player 2 on the second round knows that he is better off
confessing, so we can eliminate the "don't confess" option (grey on First Step
figure), and similarly for player 1. After eliminating those branches, we are left
with the Second Step figure, and each party, knowing what will happen at the
second play, can calculate he is better off confessing. An example of subgame
perfect equilibrium.

III. Ex Ante
Ante, ex Post
Post:
A. Consider two different approaches to preventing auto accidents:
1. Ex Ante--laws against speeding, required brake inspections, laws against DUI, etc.
2. Ex Post--tort liability, criminal penalties for drunk drivers in accidents, etc.
B. Consider the treatment of attempted murder:
1. If we are punishing ex post, according to damage actually done, attempted murder is
not a crime since nobody was killed. The bullet went into a nearby tree--which is doing
fine.
2. If we are punishing ex ante, according to the effects we predict behavior will have on
average, then attempted murder is a crime--shooting at people results, on average, in
killing some of them.
3. For a further puzzle, consider the (moral?) question of why we punish murder more
severely than attempted murder. If punishment reflects desert--what the crime tells us
about the wickedness of the perpetrator--shouldn't the punishment be the same? Being a
bad shot is not a moral virtue.
4. What about the case of impossible attempts--trying to kill someone by sticking pins in
a voodoo doll? Should that be a crime?
C. What are the advantages of ex post?
1. Ex post has some of the advantage of effluent fees over regulation. The court must
measure damage done, but doesn't have to know the relationship between precautions
and accidents--the driver provides that.
2. Nor does the court have to be able to monitor behavior in order to punish its
consequences, and thus deter it.
3. The driver has a lot of private information about costly activities that are hard to
monitor. Driving when sleepy, angry, paying attention to the conversation instead of the
road, ...
4. On the other hand, both methods use the driver's private information about how costly
the precaution is (in that respect, both are analogous to effluent fees), since with speed
limits, a driver can still violate them-and pay the ticket-if he thinks doing so is important.
5. Ex post has the further advantage that measuring the costs easier after they have
happened.
D. Advantages of ex ante?
1. Popular wrong answer: "ex post doesn't prevent accidents." Sure it does--by deterring
them.
2. Risk aversion--do you want to lose your house, all your property, and be indentured
for five years if you make a mistake driving?
3. Other punishment costs--can't collect ten million dollar fines.
4. What can we do that is equivalent in deterrent effect to one chance in a million of a
ten million dollar fine?
5. Maybe one chance in a million of execution, or life imprisonment, or ...
6. None of which pay us anything.
7. So ex post forces us to use punishments which are large, and as a result inefficient.
E. Implication:
1. People are risk averse only for large losses. So small ex post punishments are
unambiguously superior to ex ante.
2 So an entirely ex ante system, with no ex post costs, cannot be the right answer
3. We want ex post at least equal to the largest fine you can pay without serious risk
aversion costs.
4. Plus possibly some ex ante if that isn't large enough.
F. Freedom of contract solution?
1. Require insurance for appropriate amount.
2. Permit insurance company to impose ex ante rules as a condition of insuring you.
3. When the cops stop you for speeding, they check your insurance to see whether you
have agreed not to speed.
4. Higher tech version--they scan the barcode on your bumper to find out what traffic
laws they are enforcing on you on behalf of your insurance company.
5. This sounds odd, but the equivalent is common for fire insurance etc.--without help
from the cops. You agree to certain precautions as a condition of insurance.
G. Note that ex ante/ex post can work at several levels. My speeding might or might
not cause an accident, which might cause a variable level of damage.
1. Act---->(maybe accident)---->varying damage if accident
2. Pure ex ante is on act. Pure ex post depends only on damage.
3. Intermediate case: punishment is triggered by injury, but does not depend on damage
done.
4. To some extent this is implicit in drunk driving laws, for example, since you are much
more likely to be tested if you run into something. And the crime of murder does not
depend on the age of the victim, although we also have an element of pure ex ante in the
crime of attempted murder. Are there any pure cases of a punishment that depends on the
accident happening but not on the damage done?

IV. Bright lines: Rules v Standards.

A. We make a lot of use of bright line rules.
1. 21 to drink--not "mature enough to be save with alcohol."
2. 35 to be president--not "as mature as the average 35 year old was when the
Constitution was written."
3. Specific speed limit. Level of blood alcohol.
4. Someone is an adult human with the same rights as all others, or a minor, or a lunatic,
or an animal--even though the factors that separate those categories are mostly things,
like age, intelligence, rationality, that vary continuously.
5. Grading "objective" exams--meaning exams where every grader, following the key,
will give the same grade.
B. We also use fuzzy line rules--standards. "Reasonable care." Reckless driving.
Negligence. Sexual harassment. Grading essay exams.
C. The advantages of bright lines are
1. they are easy to judge--you either are or are not 21.
2. they reduce litigation costs. It might make sense to spend a lot of money convincing a
court that I was behaving reasonably--but it is probably a waste of money to hire an
expensive lawyer to prove I am 21 when I'm not.
3. Since they involve less judgement by the court, they give a more predictable result.
4. So it is less likely that someone will violate them unintentonally, or that the court will
incorrectly convict someone of violating a bright line rule when he didn't.
5. So the punishment for violating a bright line rule is less likely to need to be applied.
6. Which may make bright line rules especially suitable in criminal law, which tends to
use expensive punishments.
D. The disadvantage is that they sometimes give the wrong answer. A lot of the
behavior we want to regulate involves complicated issues, where any simple rule is
likely to be wrong much of the time.

V. Consider the difference among three alternative approaches to controlling behavior that does an
uncertain amount of damage:
A. Base punishment upon what the court believes the risks were.

B. Base punishment upon what the court believes the criminal thought the risks were.

C. Base punishment upon what happens.

A. A works if the court has superior knowledge and can define the rules determining
punishment in a way that makes it in the interest of the actor to act as the court wishes.
1. "It is illegal to drive more than 55 mph" as opposed to "it is illegal to drive in a way likely to
cause accidents."
2. The penalty can be adjusted so that it reflects the court's view of damage done.
3. One common special case is where the court doesn't think it can distinguish risks among
similar cases, so imposes a fixed punishment based on the average damage over all cases.
Loses on "superior knowledge" but gains on litigation cost, since it is a bright line rule.
B. B uses information that is inferior to the criminal's information (since it is information
about the criminal's information).
1. If punishment is proportional to damage done and criminals are risk neutral, C gives the
same result as B would if it were done perfectly, (see my Payne v Tennessee article for the
demonstration).
2. B might be superior either because it produces lower punishment cost (criminal can pay a
fine equal to the expected damage but not one equal to the maximum possible damage) or
because the optimal punishment is not the expected value of the punishments appropriate for
the offenses done with certainty. This gets into details of optimal punishment theory that are
discussed in the article, and will be covered later in the course.
3. B requires more work by the court than C, since under C the court simply observes what
happens and lets the criminal to the expected punishment calculation for himself when he is
deciding whether to commit the crime.
C. Under C, the criminal (or tortfeasor) uses his estimate of the probability of each of the
possible consequences of his act (killing a high value or low value victim, running his car
into a tree or a person), plus his knowledge of the punishment for each consequence, to
calculate the expected punishment.
D. B vs C is really ex ante vs ex post, without the court trying to take advantage of any
superior information it might have and the actor lack.
E. Examples:
A. You may not discharge a gun within city limits (but more severe punishment, in practice, if
you hit someone). Drug laws (if based on risk that user will commit crimes under the
influence). These also fit the pattern of combining some ex ante with some ex post.
B: Any law that sets a fixed punishment for a crime, independent of actual damage done.
a. Fairly common in criminal law, which uses broad categories, such as petty larceny and
grand larceny, rather than making punishment depend directly on damage done.
b. Payne v Tennessee was about whether the opposite policy was unconstitutional in the
capital punishment case.
C. The normal rule in tort law. This fits the pattern of tort law using money payments, which
are low cost punishments--but ... .
a. If tortfeasors are judgement proof against high judgements, it might be better to make
each tortfeasor liable for the average damage done by such torts rather than the damage
done by his tort.
b. One argument for the eggshell skull rule is that if no damage were done, no damages
would have been paid.
c. But one can imagine torts where it is easier to measure the average damage than the
actual damage. Suppose, for example, a reactor releases radiation which raises the cancer
rate by 10%. Noboby knows which cancer patients would not have gotten it save for the
leak.

VI. "Impossible attempts": Should there be a penalty for attempting to kill someone
by a method that cannot work?
A. Argument against--we don't mind if people stick pins in voodoo dolls, so why
impose costs on us and the attempted murderers to prevent it?
B. Argument for: The legal rule isn't about voodoo, it is about methods that don't
work. There must be some uncertainty about what methods work, or nobody would
use the ones that don't.
C. Consider a rational murderer choosing between voodoo and poison:
1. Rational believer in voodoo?
2. Why not--we all believe in, and under some circumstances trust our lives to, things we
don't really understand, and trust because people we trust tell us they work.
3. If he knows it doesn't work, he won't use it.
4. If he is certain it works, our legal rule (punishment for impossible attempts) has no
effect, since he is certain it does not apply to him--and the rule is costly, so why do it?
5. But if he thinks either method might be the one that works, penalizing him either way
helps to deter him. When he decides whether to try to commit murder, he must take into
account the risk that he will choose an impossible method and not only will fail, but
might be caught and punished.
D. Of course, another way of providing the same additional deterrence might be by
raising the punishment for success. But what if raising the penalty on methods that
do work is not an option--we can't hang him twice. We can hang him once if he
succeeds, and lock him up if he fails--even by an impossible method.
E. Consider that there is a sense in which any method that fails is "impossible."
1. You can't kill someone by shooting a bullet that misses him (or at least are unlikely to-
-he might get a heart attack).
2. If the world is deterministic, then every attempt had, if only we knew enough, either a
certainty of success or no possibility of success. We are back with the question of
whether to punish attempts.
F. My article on Payne v Tennessee discusses the ex post, ex ante issue in a
somewhat different context.
1. Should murderers be punished more severely (for example, execution instead of life
imprisonment) for killing particularly valuable victims--for example, a mother with
young children?
2. If the answer is "yes," that is a form of ex post punishment--punishment by the
damage actually done.
3. The alternative is to punish according to what the court thinks the murderer knew
when he committed the murder--a sort of ex ante punishment.
4. In other words, punishing you particularly severely if the court thinks you knew the
victim was likely to be a particularly valuable one, but independent of whether the
victim actually was particularly valuable (you get the death penalty for killing someone
in the process of committing suicide, because the court believes that you didn't notice the
empty bottle of sleeping pills next to her).
5. If criminals were risk neutral, and optimal punishment proportional to damage done,
then punishing by damage done works perfectly, since it gives the right <P>.
6. As in ex post/ex ante and impossible crimes context, this assumes that you either do
not have superior knowledge or cannot convery the fact of your superior knowledge to
the criminal. He can only use his knowledge, and you want to give him an incentive to
use it correctly.
7. The article (not the part assigned) also discusses some questions about the relation
between punishment and moral desert which are outside the subject matter of this course
but which some of you may find interesting.

VII. Property rules vs liability rules:

A. The difference:
1. Property rule: If I want to use your property I have to get your permission, and if I use
it without your permission I receive a punishment which may be much larger than the
damage I actually did. In effect, the rule pushes people into coordinating by mutual
agreement, rather than by unilateral action followed by litigation.
2. Liability rule: I can use your property without your permission and then pay (in the
form of a damage judgement when you sue me) for the damage done.
B. A liability rule requires the court to measure the value of what has been taken,
damaged, etc., so the harder it is for courts to measure value, the greater the
advantage of a property rule.
C. Under a property rule, the only way someone other than the owner of the
property gets to use it is via a transaction with the owner, to get his permission. So
liability is more attractive the harder it is for parties to transact.
D. Consider ordinary private property. A property rule is clearly superior to a
liability rule. If I want your property, labor, etc. I buy it. The alternative would be a
system where I get groceries by going into the store, taking them, and having the
owner sue me for their value.
E. Consider pollution, or risk of airplane crashes-liability is clearly a superior rule. I
can't contract, before taking off, with all of the people through whose roofs my
airplane might conceivably crash if enough things go wrong.

Summary before First Midterm

I. What is economics, what does it have to do with law?

II. What is economic efficiency (aka "wealth maximization")?

A. Important to L&E for two reasons:
1. The conjecture that legal rules tend to be efficient
2. The argument that legal rules ought to be efficient.
B. Judging rules, outcomes by the summed effect on everyone, where
1. The effect on each person is judged by how much he would, if necessary, pay to get or
avoid the outcome.
2. Effects are summed across people in dollars.
C. Not a perfect measure of the good, but ...
1. One which is part of the good for most people (people getting what they want)
2. And may be the part that legal rules are most capable of providing.

III. Externalities: Why inefficiency happens and what to do about it.

A. If everyone bears the costs and receives the benefits of his action, then rationality
for the individual implies the efficient set of actions.
B. If not, it doesn't; individual rationality does not imply group rationality.
C. Simplest solution: Direct regulation. Tell people to do the efficient things.
D. More sophisticated: Pigouvian tax aka effluent fee.
1. Impose the external cost on the actor, and then
2. Leave him free to choose his actions, since his net benefit now equals total net benefit.
E. Coaseian Critique I: Causation is double sided, Pigouvian taxation assumes that
you know which party ought to adjust.
F. Coasein Critique II: If rights are defined and the parties can easily bargain, they
will move themselves to the efficient outcome without Pigouvian taxes.
G. Coaseian Conclusion: It all depends on transaction costs. Which takes us to a
new paradigm for understanding inefficiency which includes dual causation,
bargaining, and transaction costs.

IV. Insurance: A way of thinking through issues of risk and cost allocation.
A. Risk aversion implies gains from transferring, spreading risks.
B. Moral hazard seems to imply a loss from transferring risks.
1. More precisely, it implies a loss from transferring risks away from the person in the
best position to use the incentive from bearing costs to prevent them from happening.
2. But that is not necessarily the person the risk would fall on if we did nothing (didn't
buy insurance, didn't have a legal rule making someone else liable, etc.)
3. So "moral hazard" is sometimes a feature, not a bug--when we shift the risk from the
person it directly falls on (the consumer injured by an exploding coke bottle, the
corporation whose factory burns down) to someone better able to control it (the
producer, the insurance company that specializes in keeping factories from burning
down).
C. Adverse selection:
1. when either buyer or seller has information that the other party lacks, and that cannot
be (convincingly) shared, some worthwhile transactions may fail to take place.
2. One solution is to put the risk on the person who knows how great it is, since that way
the knowledge no longer matters to the person who doesn't have it.
D. Litigation cost is an argument for letting costs lie where they fall.
E. Freedom of contract may provide a way of getting the parties to estimate the net
effect of all these considerations, and draw up the rules accordingly--as we usually
do in the case of insurance.

V. Strategic behavior: See my Price Theory Chapter for more details.

VI. Ex Post/ Ex ante, bright line rules vs fuzzy standards.

VII. Property rules (allocation by consent of the owner) vs liability rules (allocation by
someone else taking, owner suing for damages). Why each makes sense for some
problems.

I: What is property?
A. we take the institution for granted--for familiar things
B. But it is a specific kind of rule and concept, as you can see by applying it in less
familiar contexts. My obligations to you are good only against me, but your
ownership of property is good against the world--everybody is legally obliged to
respect it.
C. You ask a friend to borrow his bicycle. Implicit is:
1. he can have it back when he wants
2. If you give it to someone else, the owner can still claim it.
D. You ask a friend for information on a good dentist.
1. He has no right to have it back--to have you stop using the dentist unless you pay him
for the information. Or even to demand that you give him the dentist's name if he
forgets.
2. He has no right against third parties.
E. A gives you a business idea; you sign a non-disclosure agreement, etc. You tell it
to B who tells it to C, who uses it. A can sue you, but he has no claim against C to
"have his idea back," assuming C did not induce the breach.
F. There is a very partial exception to this in trade secret law, discussed below.
G. And a partial exception to the idea that you can always reclaim your property in
the legal rule of "adverse possession," which implies that if someone else treats your
land, or some right with regard to your land (such as the right to walk across it) as
his for long enough (about seven years, varying by state) and you do nothing to
indicate that it is yours, it becomes his.
H. So a recording system becomes important for maintaining property rights,
especially for valuable items such as land.
I. And how fancy the property rights are--to what degree you are free to unbundle
them and sell only part of the bundle--may depend on how easy it is for innocent
third parties to figure out what they are getting. If you unbundle the tires from a car,
the purchaser can learn it by inspection. But what if you unbundle mineral rights
from the land? The right to decide what color your front door is from a house?
J. So for most forms of property other than land, the bundle of rights is fixed.
1. You can sell me your car, along with an agreement that I will not drive it on Sunday or
sell it to anyone who will.
2. But if I resell without requiring the buyer to sign such a contract, you have no claim
 against the new owner--he got the full bundle of rights to the car
3. only a claim against me for violating the contract.

II: Determining property rules involves a set of questions:

A. How rights should be bundled (Coase article, our earlier discussion)
B. How should you be allowed to defend your property rights (property vs liability)
C. What should be property?
D. We are concentrating on the last at the moment.
E. Note that intellectual property provides a good case to think about, since it is on
the border between things that we assume are property (land, cars) and things that
we assume are not property (words, names).

III. Property rules--including Intellectual Property (I.P.), primitive societies, et multae

caetera:
A. Property vs commons.
1. Argument for property is obvious. Why plant if you can't harvest? Tragedy of the
commons.
2. Argument for commons? Cost of enforcing and/or transacting over property.
3. English language as a commons. We could give ownership of every new word to its
inventor, and there would be some advantages to doing so--but less than the costs.
4. From primitives to cyberspace.
a. When does it pay to define and enforce property rights in land? Not if you are
hunting large animals across it--the rights you want are in the animals. Yes if you
are planting on it.
b. Giving away information on the internet vs charging for it. Porn does the latter,
almost everything else does the former.
c. A lot EMail is transmitted on an "I'll forward your packets, you'll forward mine"
basis.
d. All you can eat restaurant as a commons.
i. There is some cost to monitoring how much you eat, whether by the
waiter bringing it to you are the cashier weighing and/or counting. All you
can eat avoids that cost.
ii. The cost of giving the food away is the resulting ineffiently large
consumption.
iii. If I eat an extra 4 oz of steak that costs the restaurant $1 and is worth
only $.40 to me, that is a net loss of $.60 (the restaurant can collect the $.40
in their fixed price).
iv. Much less of a cost for salad or spaghetti--and the same savings to
balance it.
e. An Internet Service Provider with a fixed monthly charge is a commons too--no
charge for additional use. These are voluntary commons in the middle of private
property, so it can't be just a case of "it didn't occur to them to treat it as property."
B. Boundaries. Floating islands story: Stack island, which belonged to someone, floated down
the Mississippi (erosion at the upstream end, deposition at the downstream end), into a strip of the
river where all islands belonged to the owner of the land on the west bank of the river. Who did it
belong to?
1. Property rights would be hard to work with in a world where that was the norm instead of
the exception--where the boundaries that defined land ownership kept moving in unexpected
and inconsistent ways. Arguably, in intellectual property it is the norm.
2. In patent law--what is the boundary of an idea?
a. This shows up as the question of what does or does not infringe a patent.
b. And in how much you can claim--Morse's 6th claim was for all ways of using the
electromagnetic force to transmit signs or images to a distance. That includes the fax
machine, the internet, the telephone, .... . The court denied it.
3. In copyright? Not much of a problem defining boundaries if we if only prevent literal
copying, since it is easy to see if that has occurred. But as you get farther from that ...
a. Advertisements using the framework of a popular cartoon series may infringe its
copyright.
b. The movie Dr. Strangelove seems to have been based on the novel Red Alert, although
with large changes.
4. Wild animals are generally not private property, because the cost of keeping track of which
animal belongs to whom is too difficult.
5. Underground petroleum moves around as you pump some of it out.
a.If I pump long enough, and you don't have not drilled your own oil well and started
pumping, I may end up pu mp out all of your oil as well as mine.
b. So each of us has an incentive to drill too early andpump too fast, in order to (at some
cost) get the oil before the other does.
c. The usual solution is "unitization," a legal rule under which some fraction, often 2/3,
of the owners of land above a common oil pool can vote to have the oil treated as a
common resource, belonging to all the land owners.
C. Defending. Dogs and property rights. The bionic burglar alarm.
1. Changes in technology change what rights can be enforced.
2. Consider a primitive people trying to decide whether land should be private property or a
commons. The cost of private property includes the cost of guarding it--there is no point in my
planting wheat if my neighbor is going to come by at night and harvest it for himself. That cost
was substantially lowered by the domestication of dogs, since dogs, being territorial can learn
to identify their master's property and bark at trespassers. Perhaps that was what tilted the
balance in favor of private property. If so, we may owe most of human civilization to the dogs.
3.The widespread availability of copiers means that copyright can no longer be enforced
against people who make small numbers of copies for their own use. Similar arguments apply
to enforcing copyright in computer programs.
4. One approach to dealing with enforcement issues in I.P. is to enforce your rights not against
those violating them but against those making the violation possible, under the doctrine of
contributory infringement. In one case, a firm had a process patent on a way of using a
chemical in growing rice. Other firms sold the (unpatented) chemical, along with instructions
on how to use it--the purchasers then violated the patent by using the chemical in the patent
way. The patent holder sued the firms selling the chemical for contributory infringement and
won, there being no other substantial uses for the chemical.
5. Sony v Betamax is a famous case that went the other way--making a VCR is not
contributory infringement. The argument was that, while a VCR can be used to tape a movie
off TV, infringing the copyright owner's rights, it can also be used for time shifting programs,
so there are substantial non-infringing uses, so the manufacturer of the VCR is not responsible
for the infringing uses.
D. Transaction costs
1. Chasing large animals across private property. It would be costly to bargain with each
property owner when you reach his boundary for the right to cross his property--and while you
were bargaining the animal would be getting away.
2. Propertising English. Suppose each word belonged to its inventor. Writing a sentence would
require a lot of licensing negotiation. One argument against patents for software is that they
present a similar problem. A programmer must do a patent search on every programming trick
he uses to make sure someone has not patented it.
3. All you can eat restaurants. Cost of monitoring is higher than the efficiency loss--for
spaghettti, lettuce, but not steak. The more expensive something is, the greater the net cost due
to consuming an inefficiently large amount of it.
E. Creation of Property rights: Rent seeking problems.
1. Homesteading. If I get ownership of 160 acres of government land by being the first person
to farm it for X years, there is an incentive to start farming prematurely--losing money on the
farming in exchange for obtaining ownership of valuable land. If I wait until the land is worth
farming, someone else will already be there.
2. Patent races. If I make my invention a day before you do, I get all the rights to it, you get
none. So there is an incentive for an inefficient competition to be first.
3. This should not be a problem for copyright. Nor is it a problem that someone who builds a
house or an automobile then owns it. What is the difference?
4. Crucial difference--homesteading or a patent race takes something from the commons. It
eliminates a valuable opportunity (to homestead that piece of land or make that invention) that
others previously had. The other examples do not.
5. In other words, such an act confers a negative externality on others.
6. But this may be a good thing if there is a positive externality to be balanced.
a. Perhaps the homesteading law was a good idea, because settlement that was
"premature" in terms of the profit of farming to the farmer was not premature in terms of
total costs and benefits. The first settlers might provide a "defense externality" by each
making it easier for the others to defend themselves against Indians (or Canadians in the
Pacific Northwest or Mexicans in the southwest).
b. Similarly, the inventor generates not only knowledge that he owns (how to practice his
patent) but also knowledge (how to make other inventions outside of the patent, by using
unpatented parts of his idea) of value to others. This is an argument for giving him a
benefit in exchange (at other people's expense), but it is not an argument for making the
benefit depend so heavily on when he makes the invention, since we get very little extra
from his making it a week earlier--which may be what gives him the patent.
c. How soon we want something invented depends in part on how rapidly the cost of
inventing it is falling with time. If it costs the same amount to invent it this year as next
year, we might as well invent it this year; if it will cost a tenth as much to invent it a year
later, we might be better off waiting. One advantage of trade secret protection over
patent protection is that it gives qualitatively correct incentives in this regard. The faster
the cost of inventing something is falling the shorter the period for which you can expect
to maintain it as a trade secret, hence the lower the incentive to invent it early.
F. Supply elasticity.
1. What is the cost of leaving a positive externality external? That depends on how much the
reduced incentive reduces output.
2. I get only part of the benefit from not dressing horribly or not having a deliberately ugly
house. But since it costs me very little (perhaps less than nothing) to do those things, even part
of the benefit is enough--so few people have deliberately ugly clothing or houses.
a. In such situations, a problem arises only when tastes differ--so that your "deliberately
beautiful" (or at least "desirable in some way") is my "deliberately ugly" (punk haircuts,
for example).
b. Or when one party is malevolent and wants to hurt the other.
c. Or when one party is engaging in extortion--imposing costs on others in order to be
paid to stop doing so. Most people are strongly committed to a strategy of not giving in
and paying off in such situations.
3. Would there be "almost enough" intellectual property even without protection?
a. Reputation and other tie-ins provide some incentive. People wrote books, poems, etc.
long before the invention of copyright (Homer, Dante, ...). The first firm to come up with
a product has a first mover advantage--consumers identify the product with that firm.
The first firm may be able to establish standards, and consumers are uncertain whether
other firms are successfully imitating them.
b. c. 1900 the U.S. did not recognize British copyrights, yet British authors got sizable
royalties from their American sales. With large fixed costs and time delays due to
typesetting, the first publisher (who got the manuscript from the author) had a sizable
advantage over pirates (who had to wait for the first publisher to print his edition before
they could start pirating it). Lotus has no patent on the idea of a spreadsheet (which
would have gone to Visicalc anyway), yet has a large first mover advantage.
4. A related issue is the supply curve of the copier. A good which is very expensive to copy
(paintings, at least until recently, live lectures--although Hal Holbrook's Mark Twain
performances may be a counter-example) may not need intellectual property protection.
5. This is the same issue as steak vs salad at all you can eat restaurants--except that there it was
the consumer not the producer who controlled quantity.
6. A related issue raised by Kitch is whether it is important to coordinate the production of
intellectual property in some area, in order to make sure six people are not working on one part
of the problem and none on another. Ownership of the area via a patent is one way of doing
this. The owner can hire everyone else, or subcontract, and coordinate their efforts.

IV. Intellectual Property Law: What the law is.

A. Copyright Law
1. Applies to expression, not idea.
2. Applies mostly against literal copying
a. A bit vague at the edges--Superman (or at least his copyright owner) won an
infringement suit against another superhero that was too similar to him.
b. Is Superman an expression of the idea of a superhero, or is a particular
Superman comic book an expression of the idea of Superman? On the former
reading, Superman is copyrightable, on the latter he is not.
c. A movie derived from a book would violate the book's copyright even if there
was very little verbatim copying.
d. In computer law, the issue of non-literal copying has become an important one.
3. Copyright protection has very weak requirements of originality. The information in a
phone book is not copyrightable, on the grounds that producing it was a mechanical
process requiring essentially no human creativity at all.
4. Copyright provides very long protection.
B. Patent Law
1. Applies to the idea, not just the expression.
a. Inventor must describe the best implementation he knows,
b. But better implementations are still covered by his patent, and
c. Constructive reduction to practice is now an option, which means that he does
not have to actually produce a working model--just a description from which a
patent examiner believes that one skilled in the art could produce a working
model.
2. Novel and non-obvious.
a. The inventor hasn't been doing it for more than a year
b. Nobody else has been doing it in public for more than a year
c. One skilled in the art could not readily figure out how to do it.
3. Useful.
a. In the 19th c. this was interpreted as "not positively undesirable," on the theory
that giving a patent on a useless invention did no harm. A famous case stated that
an invention to assist debauchery or private assassination would probably not be
found patentable.
b. One infringer actually won by using this standard to challenge the validity of
the patent he infringed. He argued that the invention was positively undesirable, so
the patent was invalid, so he should be allowed to practice the undesirable
invention. It was a process to mottle tobacco, thus making it look like tobacco that
came from places whose product was of high quality--a legal fraud on the
consumer.
c. More recent courts have given more substance to the requirement of utility,
holding, for example, that a process to synthesize a chemical with no known uses
is not useful, thus not patentable.
4. Bounds of the invention are set by the claim.
a. If the claim is too broad, you don't get your patent, either because someone else
has already done something that fits your broad claim or because the court thinks
you are claiming more than you invented. Morse's sixth claim (disallowed) was
for all ways of using electromagnetism to transmit letters and figures to a distance-
-and thus covered the fax, EMail, TV, ... . That was his idea, the telegraph he had
invented was simply one implementation of that idea. The court didn't buy it.
b. If the claims are too narrow, the patent is not worth much.
5. First to invent gets the patent in the U.S., first to file elsewhere.
6. Patents provide relatively short protection.
C. Trade Secret law:
1. Almost entirely deals with the consequences of acts that would be banned anyway,
such as:
a. An employee violating his obligation of loyalty to his employer by selling a
competitor secret information.
b. Breaking into a lab at night and photographing the working model.
c. etc.
2. Trade secret law allows the victim to sue for sums based on the value of the secret.
3. And gives some rights against an innocent third party who has received the secret.
D. Trademark/trade dress law: In effect, this body of law propertizes a limited part
of the language. If a particular restaurant design has come to mean your chain in the
minds of consumers, you may be able to prevent another restaurant from copying
the design. Similarly for a trademark--a word that has become closely associated
with your product.

V. The Economics
A. Why have protection?
1. Traditional answer: Incentive to produce writings and inventions.
a. But is it the right incentive?
b. The more people will pay to license your patent or buy your books, the greater
the value you have produced, and the greater the reward you get, hence the
incentive.
c. The longer it will be before someone else would have produced the same
intellectual property independently, the greater the value you have produced. And
Copyright accordingly gives longer protection than patent.
d. But patent is still a "one size fits all" system--it is not likely that all patented
inventions would have been independently reinvented in exactly 17 years.
2. To discourage secrecy
a. A patent make secrecy unnecessary, since if someone copies your idea you can
sue him.
b. Giving up secrecy is the price of getting a patent, since you must reveal the best
method of practicing the invention that you know in sufficient detail to allow
others to use it.
3. Kitch homesteading alternative--the objective is to propertize ideas.
a. Once a range of ideas has been converted into private property by a patent, the
owner can develop it, coordinating future research
b. just as a prospector who has filed a claim can develop the mine without
worrying about other people digging holes next to his and taking "his" ore out.
B. Why not use contract instead of intellectual property?
1. The originator of a work or invention could transfer it to others on condition that they
agree not to make copies or use the idea, and only to transfer it to other people who
accept the same agreement.
a. Instead of suing someone for violating your patent or copyright you sue him for
violating his contract with you.
b. But when you find copies being sold, how do you know which of the people
who bought it from you andsigned the contract is the source of the original that is
being copied?
2. Copyright: The problem is how to enforce it against innocent 3rd parties.
a. It could be done by a legal system in which the buyer has only bought part of
the bundle of rights corresponding to the work. He does not have the right to make
copies, so cannot sell it to others. In effect, copyright law creates such a system,
with respect to removing that one item from the bundle.
b. This is the way our legal system handles restrictions on real property, such as
easements and restrictive covenants that "run with the land."
c. But extending it to other forms of property, without the elaborate registration
systems we have for land, could be a problem, and is something our legal system
is generally reluctant to do.
3. Patent.
a. Using contract instead of patent is done to a considerable extent to protect trade
secrets. The rule on innocent 3rd parties who get a trade secret through someone
else's violation of his obligation, without knowing they are doing so, is that they
can practice the secret only if not doing so would impose serious costs on them
(i.e. if they have already built the factory using the secret process).
b. But there is a serious problem with this approach in situations where use and
sale of the patented good, or the good produced by the patented method, reveal the
invention. It is hard to enforce a contract binding every buyer to hide the good he
bought from everyone who will not agree to respect the inventor's sole right in the
ideas it embodies.
3. The issue of contract as a substitute for intellectual property may become an issue
again in the context of computer networks, where new technology may make the
enforcement of intellectual property law very difficult.
4. One solution for copyright would be to somehow tag every copy of the computer
program you sell, so that if a pirate copy appears you can prove which buyer is
responsible--and sue him for violating his agreement not to allow others to copy the
program.
C. Why do we have different patent and copyright rules, and why are they applied to
the things they are applied to?
1. The Constitution refers to "writings and discoveries" but does not specify the rules for
each.
2. Commons problem:
a. Small for copyright--if I didn't write and copyright the book, it is very unlikely
that someone else would have written it.
b. Big for ideas--quite often, several people are trying to make the same invention.
c. So we have much tighter rules and shorter protection for patents than for
copyrights.
3. Costs of fighting over fuzzy boundaries are an argument against protection.
a. Literal copying is easy to spot and prove.
b. Copying of ideas is much harder.
c. So we have easier and longer protection for copyright.
D. Copyright Protection for software.
1. A program is not a writing, since it is intended to control a machine, not be read by a
person (more like a cam--some courts got that right).
2. But it has the characteristics that make copyright suitable to writings.
3. As long as protection is limited to literal copying.
4. A lot of problems appear when you push it to cover non-literal copying.
E. Patent protection for software.
1. The Supreme Court opposed it under the "mental steps doctrine," (you could not
patent something that was simply a series of mental steps, such as the rule for doing long
division) but ...
2. Gradually backed down, under pressure from the CAFC, which was an inferior court
but knew more about the subject and had stronger opinions on it.
3. Many believe creating a problem
4. Not because software is not an invention, but because its characteristics may make
protection cost more than it is worth.
a. A new field, so courts don't understand it very well and give patents for things
already in the art.
b. Very fast changing.
c. Arguably the cost of writing programs is low compared to the cost of making
machines, the sort of thing patent law was designed for, so the incentive of patent
may be unnecessary. Lots of programs were written and lots of progress made
before the Supreme Court started recognizing software patents.
F. Why doesn't patent law pre-empt trade secret law?
1. Sears vs Steiffen says it sometimes does--a state cannot give the equivalent of patent
protection to things that do not qualify for a federal patent.
2. But in general it does not. Kewanee v Bicron says that states are allowed to have trade
secret law, even though it deals with some of the same things as federal patent law, and
does so in a different way.
3. Trade secret helps fill in some of the gaps left by patent law, since ...
a. Patent is one-size-fits-all; either you get it or you don't.
b. You can use trade secret law to protect secrets that are not sufficiently important or
non-obvious or whatever to qualify for patent protection but are still worth something.
c. You can use trade secret to protect secrets too important for patent--ones that you
believe you can keep secret much longer than 17 years (the formula for Coca-Cola, for
example).
d. You can use trade secret to protect ideas that you think are non-obvious but the patent
examiner thinks are obvious. If he is right, someone else will invent the idea, making
your secret worthless. If you are right, you get some of the protection you deserve.
e. And trade secret law has the right incentive pattern with regard to timing--it is not
worth spending a fortune to invent something this year that will be easy to invent next
year, since your trade secret will only last a year.

VI. Coase Article: Classic article that laymen, or at least econ majors, can read.
A. The problem associated with an externality is jointly caused--the result of actions
by both parties.
B. Farmer and cattleman, outcome does not depend on who is liable
1. But bargaining, attempted extortion, etc., is possible
2. So Coase's result holds when everything has been bargained through
3. i.e. in a zero transaction cost world.
C. A bunch more cases with the same logic:
1. Confectioner/doctor
2. Bleaching coconut matting
3. Blocking the draft of a chimney. Who "caused" the nuisance? Both. so
4. Each should bear the full cost--and will under either rule with bargaining.
5. Jolly Anglers brewing. Confined channel of air is presumptively property, by rule of
lost grant belonged to the brewer. So the verdict is the opposite from that in the chimney
case--which also involved blocking an (unconfined) channel of air.
D. Judge's grounds for deciding the cases seem irrelevant to economist, but ...
1. Perhaps all that matters is a predictable rule, (to reduce uncertainty and rent seeking)
and the judges have one, or ...
2. Perhaps the rule is a proxy for relevant considerations, such as
3. Coming to the nuisance.
a. I have a pig farm; the city expands towards it.
b. When you build a housing development next to me, can you enjoin my pig farm
as a nuisance, or ...
c. Can I win by arguing that you "came to the nuisance?"
d. The economic argument is the assymetry of sunk costs: It is cheaper for you to
"move" your housing development before you build it by building it somewhere
else than for me to move my pig farm, which is already there.
e. So the doctrine of coming to the nuisance, which sounds like a moral argument,
may make sense as an economic one.
f. In fact, in modern American law, courts mostly reject the doctrine. Should they?
Not clear.
E. A firm is one solution to the transaction costs of the market
1. Consider a shopping mall.
a. The owner provides free parking, and makes his money back in store rentals.
b. He figures out what mix of stores, restaurants, etc. will make people want to
come, and so maximize the total return
c. He keeps the public areas clean
d. In fact, he provides a centralized alternative to nuisance law, government, etc.
2. Whether it makes more sense to solve problems by putting both actors into one firm
depends on the tradeoff between administrative costs of the firm and the alternative
market costs.
F. Government regulation is another solution
1. Which has costs, and errors, and so may give worse results than the other solutions
2. Or better.
G. And a final solution is to do nothing. Some problems cost more to cure than the
cure is worth.
1. That is how we deal with lots of externalities
2. Positive ones like beautiful buildings, and
3. Negative ones like people wearing ugly clothes.
H. Where transaction costs are high, court decisions matter, and cases suggest at
least some general recognition of reciprocal problem and cost/benefit issues.
I. Common law of nuisance v statute.
1. Statute may extend or reduce the coverage of the law of nuisance.
2. What we observe is frequently government authorised, not the result of lack of
regulation
3. And perhaps should be.
J. Pigou
1. State action is needed; what additional state action?
2. But Pigou's railroad example is due to state action, and ...
3. Not necessarily undesirable. Double causation with use of land.
K. What is owned is a right, not a thing. For instance, the right to produce pollution.

O. Digression on simplified pictures.

A. In thinking through the logic of a problem, whether in physics, mathematics, or
economics, we frequently use simplified pictures, designed to bring out the
particular issue we are interested in.
B. For example, we worth out the logic of Newtonian motion by analyzing the
trajectory of canonballs in a vacuum--without worrying about the fact that in a
vacuum, the canoneers would die before they could fire their cannons.
C. Similarly, much of what we are doing in this course involves deliberately
simplified pictures, intended to let us think through the logic of one problem or
another.
D. After doing so, one is better equipped to deal with the much messier conditions
of real life problems.

I. Contract law: Why have it?

A. Why do we want contracts to be enforceable?
1. Because performance often occurs over time, with transaction specific investments.
2. House building problem, for example.
B. What about private enforcement?
1. Via reputation: Orthodox jews in the diamond industry example.
a. The industry depends heavily on trust, keeping your contract, etc.
b. But orthodox Jews are not permitted to sue each other.
c. If you violate the agreement, there are mechanisms (take it to the Rabbi) for
proving the fact, and then publicizing it.
d. And presumably, if you don't have enough of a stake in your reputation yet for
that to matter, you won't be trusted the deals where you could make a lot of money
by breaking your contract.
e. Also, there are social costs as well as economic ones with such a tight-knit
community
f. Perhaps patterns like this--of small, close knit ethnic minorities dominating
some industry, are evidence that this is a better method of enforcing contracts.
g. But people who are not members of such groups, not engaged in repeat dealings
in markets where reputation is important, still want to sign contracts--building a
house.
2. Via bonding with a reputable enforcer
a. This requires you to have enough assets to post bond
b. But legal enforcement requires you to have enough assets to pay damages
c. The difference is that you don't have to actually transfer them to a bond agency.
3. So there may be a legitimate role for government enforcement
a. And whether legitimate or not it is there and worth studying
b. Many of the economic issues are the same whether we are talking about public
courts or private arbitrators.
c. It is worth remembering the latter case, since it represents a large fraction of all
commercial dispute settlement, but insn't usually included in what law students
study.
C. Why do you need contract law--why not just enforce the contract, letting the
parties define all the rules?
1. Because you don't think they should be permitted to define the rules--paternalism or
something similar.
2. Because you still have to decide when they have signed a contract, what rules they
agreed to, etc.--issues of consideration, mistake, etc.
3. Because there isn't enough fine print in the world to settle all possible issues that
might come up, and you need a court to fill in the gaps when they arise.

II. Why should we or shouldn't we enforce contracts if we can?

A. Argument for--efficient rules because:
1. Any change that produces net benefits can be combined with a price change to make it
an improvement for both parties, so ...
2. A sensible lawyer or other contract draughtsman will look for the efficient terms in
order to maximize the gain, then try to get as much of the gain as possible for his side,
rather than picking inefficient terms biased towards his side.
3. What about a monopoly?
a. Can he charge any price he like, and combine it with most favorable possible
terms to him?
b. No--monopolies don't sell for an infinite price.
c. They are constrained by what customers will pay (more precisely, by how much
they can sell at what price), so moving to a more efficient contract and
compensating themselves in the price makes them better off, except ...
4. That contract terms might be a device for discriminatory pricing.
a. I have two customers, one of whom will pay $10 and one $15; currently I
charge $10 and sell two units--at $10 each, since I don't want to lose the first
customer. Assume I can't charge them different prices, perhaps because I don't
know which is which.
b. A change in contract terms will make the first customer worse off by $1, the
second by $5, and benefit me by $4--so it is inefficient.
c. I make it, cut my price to $9, losing $2 in sales revenue but gaining $4 in more
favorable contract terms.
d. So in such situations, where inefficient contract terms are a device for price
discrimination, we might get them.
e. For example tie-in sales.
f. But note that price discrimination has some benefits--it may make it possible to
produce a product that could not otherwise cover its costs, or
g. Make the product available to people who would not buy it at the profit
maximizing single price.
B. Obvious exception--third party effects.
1. Assassination contracts are not enforceable
2. Nor contracts agreeing to keep silent about illegal acts.
3. Except for the marriage contract--which includes an enforceable obligation not to
testify against your spouse.
C. Second exception--parties we believe incompetent. Children and the insane.
D. Shakier exception--parties we believe incompetent because we are sure those
terms cannot be in their interest.
1. Consumer repossession agreements--might be in the consumer's interest.
2. Penalty clause in a contract: An agreement that the breaching party must pay a large
penalty to the other party, representing more than the actual cost of the breach.
a. Maybe it is the best solution to adverse selection--I know I won't default and
you don't, so I "tell" you convincingly by agreeing to a penalty clause.
b. Maybe it compensates for the risk of judgement-proof breach--someone who
breaches and then goes bankrupt.
c. Maybe it reflects distrust of the court's ability to correctly measure damages... in
fact
d. It is simply the private version of a property rule (you can only breach with the
other party's consent--or bad things happen to you) instead of a liability rule
e. Which we think reasonable in settings where transaction costs are low and/or
courts are expensive or incompetent.
E. Making sense of the idea of duress:
1. Real duress--when you demand my money or my life and I offer a check instead,
should I be free to stop payment?
a. If I am, either you don't take the check or you take it and then kill me so I can't
stop payment.
b. But if I am not, that increases your incentive to find me alone, point a gun at
me, and say "your money or your life."
c. So whether such agreements should be enforced depends on the elasticity of
supply of such situations (if such contracts are enforced, how much does that
increase their frequency) and how much they ameliorate situations that would
have occurred anyway.
d. For example, we may want to enforce parole/ransom rules during wartime--it
isn't why you were captured, but it may be a good reason not to kill you.
2. Semi-real duress: The sinking ship or the starving traveller in a snowstorm.
a. It isn't duress in the previous sense, because the tug isn't the reason your ship is
sinking.
b. Do we want to let the tug bargain for the full value of the sinking ship?
c. Yes if we want the right incentive for tugs to steam around looking for ships in
trouble.
d. No if we want the right incentive for ships to stay out of trouble.
e. This is our old problem of dual causation in a new form.
f. There is much to be said for some prearranged rule, whatever it is, since
bargaining costs may be excessive if the water is rising while you bargain.
g. And admiralty law provides for a "reasonable" salvage fee.
h. Analogous case in the snowstorm.
i. Posner gets this wrong--worries about "excessive" efforts when the efforts to
rescue due to getting the full value are efficient.
j. Not like the patent race or the sunk treasure case, because if another rescuer or
salver is expected in ten minutes, the first one isn't going to be able to get a very
high price.
3. Bogus duress: "Contracts of adhesion" (take it or leave it form contracts); the
argument for efficiency does not depend on bargaining.
a. Consider the sort of contract that a consumer faces when he rents a car or an
apartment, buys a computer program, ... .
i. It is drawn up by the other party in a standard form, applied to many
transactions, and ...
ii. Although he may have a choice as to some optional terms, the consumer's basic
choice is to sign or not to sign.
iii. Does this count as a sort of "duress," since one party cannot bargain about the
terms?
iv. And does that imply that the arguments for freedom of contract do not apply to
such contracts?
2. Innocent explanation: Form contracts reduce drafting costs, eliminate the problem of
the firm having to control the employee who negotiates the individual contracts.
3. The argument for freedom of contract still applies, since the firm, in drafting the
contract, will take account of benefits and losses to its customers. Anything that makes
the terms of the deal more attractive to the customer will also increase the amount he is
willing to pay.
4. The freedom of contract argument even applies to a monopoly. The more unfavorable
the contract is to the consumer, the lower the price the monopolist will be able to charge
and still sell his goods. Although a monopolist has no competitor, his customers still
have the alternative of not buying the good at all--which is why monopolists to not
charge an infinite price.
5. Although the argument does not always hold in the case of a monopolist, since
(inefficient) contract terms might be a device for profitable discriminatory pricing--
charging a higher (pecuniary plus nonpecuniary) price to those consumers who will still
buy at a higher price.
6. Courts sometimes refuse to enforce the terms of form contracts on the grounds that
they represent a sort of duress. These arguments suggest the courts are wrong.
7. Although we have not considered the separate issue of contracts sufficiently
complicated so that the consumer does not really know what he is signing.

III. Is there a contract there to enforce?

A. Fine print problem--did the parties know what they were signing?
B. One sided contracts--detrimental reliance.
1. In an expansive mood, I offer to put my nephew through college
2. He gives up his job, relying on that offer.
3. I change my mind.
4. He sues--and wins--even though there was no consideration--he didn't pay me
anything for my promise.
5. But he relied on it, and I should have been more careful not to promise what I was
unwilling to perform.
C. One sided contracts--should the offer of a reward be enforceable
1. By someone who knew about it.
2. By someone who didn't.
a. Making it enforceable will make it more likely that lost things will get returned,
since although the finder does not know I have offered a reward, he knows I might
have, and if I have it will be enforceable.
b. On the other hand, it also makes it more expensive to offer such rewards.
c. Think of the reward that only goes to someone who knew about it before he
returned my property as a sort of price discrimination--a higher price to people
who search for things in order to get rewards than to people who just find things.
D. Zero-sided contracts--emergency medical treatment.
1. A doctor finds someone injured and unconscious, treats him, bills him.
2. And collects.
3. What if the injured person doesn't want the treatment--he is a Christian Scientist or an
attempted suicide?
4. From the supply side, it is irrelvant--it costs just as much to treat him--but
5. It isn't worth as much.
6. The right rule would seem to be no payment if the treatment is unwanted, and a
payment if the treatment is wanted scaled up a little to compensate the doctor for the risk
that his patient will turn out to be a Christian Scientist, but ...
7. That rule would be an invitation to fraud--by people who converted to Christian
Scientists after receiving the Doctor's bill.

IV. If we do want to enforce the contract, how do you fill in the details?
A. By what would be efficient, either
1. Because we want efficient law, or
2. Because that predicts what they would have agreed to, or
3. Because if you don't they have to waste resources specifying their contract in more
detail so as to avoid your imposition of the terms you want and they don't.
B. Who should bear risks? We've been here before
1. The party who can best risk spread.
2. The party who can best control the risk: moral hazard.
a. Himalayan photographer. If he doesn't tell the photo labs that his six rolls of
film cost thirty thousand dollars to get, they don't owe him thirty thousand when
they lose the film.
b. Risk of strike, factory burning down, is best controlled by the producer--who
owns the factory and negotiates with the workers, but ...
c. Risk of buyer deciding he doesn't need the product is best controlled by the
buyer.
3. The party who best knows the risk--adverse selection.
4. Note that this is relevant both to negotiating the contract and to filling in the details.
C. What happens if someone breaches the contract?
1. Objective. Efficient breach--breach if and only if it makes the parties on net better off.
+ efficiency on other margins which will show up shortly.
2. Nothing: no enforceable contract. Inefficient breach? Not if Coase Theorem applies.
3. Breach forbidden--specific performance. Inefficient performance. Not if ... .
4. Expectation damages:
a. Give the right incentive to breach.
b. The wrong incentive to rely.
c. the wrong incentive to sign if there is asymmetric information
5. Reliance damages:
a. Wrong incentive to breach.
b. Wrong incentive to rely.
c. Right incentive to sign if breaching party has the asymmetric information.
6. Liquidated damages--agree in advance on what the damages will be if a breach
occurs.
a. Right incentive to rely--because damages don't depend on reliance expenditures.
b. Right incentive to breach if and only if the amount agreed on is what
expectation damages would be.
c. Right incentive to sign if and only if the amount agreed on is what reliance
damages would be.
d. Penalty clause--liquidated damage equivalent of specific performance.
7. Property/liability issue:
a. Specific performance or penalty clause is like a property rule--you need the
other party's permission to breach the contract.
b. Expectation or reliance damages are like a liability rule
D. Fraud: Laidlaw v Organ
1. What happened:
a. Purchaser of tobacco knew the war (of 1812)was over; nobody else did.
b. Before to tobacco was delivered the news broke, seller tried to reneg on the deal
c. The court didn't let him.
2. The benefits of the court's position: The ability of people who have advance
information to make money by it, for instance by buying tobacco at wartime prices when
the war is (just barely) over, gives them an incentive to generate such information, and
their use of it puts that information into market prices--the contract bids up the price of
tobacco, as would the news.
3. Why isn't it prevented here by contract? Why doesn't the seller specify that the
contract is void if the buyer has special information?
E. Consider the more general case of speculation.
1. Speculation produces a social benefit by reallocating resources from times when they
are plentiful to times when (an expert can predict that) they will be scarce--the
successful speculator buys low and sells high.
2. By buying grain before other people anticipate the coming shortage he drives up the
price early, giving other people an incentive to use less grain (slaughter hogs early, for
example, to save their feed for human consumption), produce more food of other sorts,
etc.
3. So when the shortage hits and the speculator puts the grain he bought back on the
market, the famine is less severe and the price does not go up as high as if he had not
intervened.
F. This only works if the speculator has secure property rights.
1. If, when the famine happens, a mob seizes his barn full of grain, or the government
confiscates it, speculation won't pay and won't happen.
2. And when there are bad harvests, people will starve
3. Which suggests that the belief that speculators cause famine may be one of the most
lethal errors in human history.
4. It is a result of applying a useful rule of thumb in an inappropriate context.
a. The rule is "cui bono"--to find out who is responsible for something happening,
first figure out who benefits by it.
b. It doesn't work in the case of production, since producers benefit by high prices
but the act of producing increases supply and so makes prices lower than they
would otherwise have been.
G. The successful speculator produces a benefit and gets a reward, but the latter is
not equal, or even closely related to, the former. The relation between the private
value of his activity and the social value of his activity is correct qualitatively (if he
makes money he is also doing good) but not quantitatively (he might do a little good
and make a lot of money, or vice versa).
1. Even in a case where elasticities are low, so that the speculator produces only a small
reallocation and a small benefit, he might still get a large profit--since his speculation
means that resources belong to him instead of to someone else at the instant when their
price goes up.
2. So we might get inefficient speculation--spending $1000 dollars to get information
whose private value is $1100 but whose social value is only $100.
3. aka rent seeking.

V. The Marriage Contract:

A. Why does it exist:
B. What is the optimal form--in particular, rules for breach.
C. Why has the actual contract changed over time, and with what consequences?

VI. Why does it exist: The reason for long term contracting.
A. Sunk costs in the marriage relation.
B. Problems in monitoring and enforcing compliance.
C. No Divorce as one solution
1. Have to live with each other, might as well figure out how.
2. Incentive for more careful search.
3. But note that in real world cases, "no divorce" meant "no remarriage"--separation was
still an option.
D. Easy divorce as another, but ...
1. Cohen's problem, so only if you can get the damage rule right will it lead to only
efficient divorce, but ...
2. That is hard to do, because how do you know who really breached?
3. And what about effects on children?

VII. What happened?

A. The facts:
1. Easy divorce, high divorce rate,
2. Increasing, and historicaly extraordinary, rate of illegitimacy
3. This pattern exists across the industrial world--why?
4. Pattern across income classes--hit first in the lower income groups.
B. Part of the puzzle--less sunk cost in marriage.
C. Part of the puzzle--welfare as alternative source of support.
D. Better job opportunities for women, in part due to B?
E. Breaking the sex/childbearing market link.
1. Consider two markets
a. A market for sex
b. A market for childbearing. Suppose that ...
c. On average, men want more sex than women, but ...
d. If cost of children is evenly divided, fewer children.
2. In a world without contraception, the markets are linked
a. One act of sex produces one sex act worth of conception.
b. If we look only at the sex market, a woman who has sex risks having the whole
cost of bearing children as a result, which makes her much less willing to have
sex, so ...
c. The condition of having sex is a commitment to help support the children, if
any, meaning ...
d. Either only sex within marriage, or a society which provides shotgun marriages
if the woman gets pregnant.
3. In a world with contraception, the link is broken:
a. Women can provide sex to men who don't want children (if accompanied by the
cost of supporting them) at low cost to the women--by using contraception.
b. So women who want children are not in a position to demand guarantees of
support from men who want sex, only from men who want children, and ...
c. On the childbearing market, the equilibrium price favors the man, so ...
d. Some women end up rearing their children themselves.
4. Puzzle: This should have hit first at the high end of the income scale, not the low,
 4. Puzzle: This should have hit first at the high end of the income scale, not the low,
because
a. Contraceptive knowledge and availability was greater among more educated
people, and
b. Safe illegal medical abortions, or foreign abortions, were more available to
higher income people
c. But it hit first at the low income end, which is evidence against this explanation
of what was happening.

III. The economics of wedding rings-argument from an article by Margaret Brinig (not
in the packet):
A. Fact: diamond engagement rings became common only starting in the 30's,
peaked in 50's, declined since.
B. Men preferred wives who had never slept with anyone else.
C. Men and women like sex.
D. This creates a problem for those not yet married:
1. For woman, who knows that sex lowers her value on the marriage market
2. For men, who can't find women willing to sleep with them, because ...
E. Solution?
1. Each generation believes it invented sex, despite the evidence, but previous
generations managed somehow.
2. Part of the solution is sex after engagement but before marriage.
3. Which creates a problem of opportunistic breach. Seduce and abandon.
F. One way of controlling this used to be the legal action for breach of promise of
marriage. The victim could collect damages, in large part for her reduced marital
opportunities.
G. Between 1935-1945 the action was abolished in U.S. states containing half the
U.S. population; it has now entirely vanished.
H.The custom of the male giving a valuable engagement ring, which the woman
could keep if he walked out on her, arose--arguably to solve this problem. Think of
it as a performance bond.
I. More recently, the custom has declined--along with the importance of virginity on
the marriage market.
IV. Adoption market:
A. Currently, it is illegal for the adoptive parents to pay the natural mother for her
consent to the adoption. It is legal to pay the lawyers who arrange the adoption, the
mother's medical costs, etc. One consequence seems to be high transaction costs (as
potential adopters bid via lawyers), and a market that shifts between surplus and
shortage.
B. One argument against payments by adopting parents to the natural mother in
exchange for her giving up her child to them for adoption is that the rich will be
buying and the poor selling.
1. This is surely a serious oversimplification; most buyers will be (most Americans are)
neither rich nor poor.
2. In any case, what is wrong with that? Would they ban the markets for housecleaning
and child care on the same grounds?
3. Is it better for the (richer, on average) adoptive parents to get the child from the
(poorer, on average) biological mother, and not pay her--which is what happens now?
C. A more interesting argument, although not one that convinces me:
Commodification
1. How we act and observe other people acting affects how we think. If we treat
something as a commodity, we and others will think of it as a commodity, which might
have bad consequences.
a. If babies can be openly bought and sold, perhaps we will think of them more as
possessions and less as people.
b. If sex can be openly bought and sold (legalized prostitution) perhaps men will
think of women more as sex objects and less as people.
c. So perhaps this is an argument for banning certain acts, not because the acts are
bad, but because they teach a bad lesson--"commodify" things that ought not to be
thought of as commodities.
2. But this argument has an odd feature, constitutionally speaking.
a. We have an act which is both an act and speech.
b. As an act it ought to be permissable (transaction with prostitute or baby seller)
c. We ban it because we disapprove of it as speech
d. Which seems to violate the first amendment.
3. As one student pointed out, a similar argument would seem to apply to laws that
impose special penalties for "hate crimes." If you punish a crime more because, in
addition to doing injury, it also conveys the message "group X are bad people who
should be hated," isn't the additional punishment a punishment for speech? After all, you
couldn't punish someone (in the U.S.) for a speech or a book arguing that "group X are
bad people who should be hated."

V. It is often claimed that, when I have a child, I impose net costs on others, so that
leaving people free to decide how many children they have will result in
overpopulation.
A. But it is not clear what the sign of the net externalities from my having another
child is. Positive externalities include:
1. My child may find the cure for cancer and
2. Will reduce the amount of the national debt your child must bear
3. And the amount your child must pay for national defense, or scientific research, or any
other government service whose costs is roughly independent of the population.
B. This argument is one example of an error common in political discussions in
many areas. You calculate net externalities considering only the externalities of one
sign (negative if you want to ban something, positive if you want to subsidize it) and
ignoring those of the other sign.
C. How in principle do we define the optimal level of population?
1. If we measure it by per capita income, utility, or something similar, we commit a
fallacy of composition.
a. Consider one additional person, who will impose neither net costs nor benefits
on the rest of us. How can we say that his life (utility 10) is a good thing if the rest
of us have an average utility of 9, but the same life is a bad thing of the rest of us
have an average utility of 11?
b. Mead's example: Consider a world with two communities, A and B. Both are
attractive places filled with happy people, but A is a little happier than B. Is the
world a better place if B is wiped out by a plague? Average happiness goes up.
2. If we measure it by total utility, we need to define a zero point, so as to know if an
individual contributes positive or negative utility to the total.
a. The obvious zero point is death, but ...
b. That leads to a rule very favorable to large populations.
c. Two unhappy people are better than one happy one unless they would be willing
to flip a coin: heads they get to be the happy one, tails they die. They will have to
be pretty unhappy to agree to that.

VII. Regulation of Sex. Why do we do it? Adultery laws, fornication laws, ...

A. Contract enforcement? Like copyright--it makes it easier to enforce the contract

against infringement involving partners not party to it.
B. Overenforce to protect third parties?
1. Children--who may be injured if adultery leads to marital breakup
2. Keep down VD, AIDS
3. Law against prostitution can be viewed as a way in which wives enforce the marraige
contract by making it harder for husbands to break it.

I. Tort Law:
A. What it is: a private action for a wrong, typically for damages, although
injunctions are also possible.
B. Differs from criminal law in being a private rather than a state action.
C. And from contract law in not being based on any contract between the parties.
D. Four issues arise:
1. What makes it wrongful: Competition is not a tort. Some system of rights is assumed,
and a tort involves a violation.
2. Causation: What does it mean to say A caused the harm?
3. Liability--strict liability, negligence, contributory negligence, etc.
4. Damages--how calculated?
E. One can think of all these issues in terms of economic efficiency, although
whether doing so describes how the law actually works is an open question.

II. "Wrongful"
A. Tort damages are a version of a Pigouvian tax, so ...
1. We want to treat as wrongful only acts that impost a net externality.
2.And are worth the trouble of dealing with through the legal system.
a. We can use injunctions or punitive damages to enforce a property rule, where it
is the injured party who owns the property.
b. Where the injuring party owns the property and is using it within his rights, we
solve the problem by a transaction--the other party buys the property, or the right.
3. And can be dealt with better in this fashion than by a property rule.
B. Should competition be a tort?
1. When I become the 101st physician in San Jose, the wages of the first 100 physicians
fall--should they be able to sue me?
2. No, because their loss is their patients' gain.
3. So I have imposed only a pecuniary externality--a transfer between two other people,
not a net injury to other people.

III. Causation complications:

A. Coincidental causation.
1. I stop my friend for a moment in the street to talk, he then continues on and is killed
by a falling safe.
There is a real case along these lines.
A tree fell on a bus.
The plaintiff demonstrated that in order for the bus to be where it was, the driver
had to have been driving above the speed limit, and ...
Breaking the law is per se negligence, meaning that it automatically classifies as
negligent even if it isn't really.
2. Should I be held liable? If I had not stopped him he would not have died.
3. No. My stopping him does not affect the probability of his dying--it might just as
easily have resulted in his not being under the safe.
4. A different way of putting this is that if I do not get rewarded when I produce the
positive externality (my stopping him happened to save him) I should not get punished
when I produce the negative one.
B. Dual causation: The victim is accidentally shot by two hunters; each bullet would
have been sufficient to cause death by itself. Should they be liable?
1. This puzzle is based on a somewhat less dramatic real case: Summer v Tice.
2. The benefit of my taking a precaution that reduces the chance I will shoot him is not
the reduction in probability of shooting him but the reduction in probability of killing
him (times the value of his life). Shooting a corpse does no damage.
3. When I change my probability of shooting him, I should benefit only by the resulting
change in probability that he will die, so ...
4. I should not be liable in the two bullet case case!
5. This ignores possible problems of proof, etc. It also ignores the possibility of
agreements in advance between two hunters; I am assuming both were independently
responsible for the accident.
6. Finally, it assumes that we can costlessly impose a fine on a hunter that fully
represents the damage done--which is unrealistic inthis case. If we can't, punishing the
two bullet case may be an indirect way of raising the punishment for the action (careless
shooting) that leads to the one bullet case.
C. Probabilistic injury.
1. Suppose a radiation leak increases the number of cancer cases in the surrounding area
over the next 20 years from 100 to 110. Is the reactor liable, for what, to whom?
a. If we must show that it is more likely the death is due to the radiation leakage
than that it is not, the defendant always wins and the reactor pays nothing for the
damage.
b. If we accept a much lower standard of proof, the plaintiff always wins and the
reactor pays much too much--the cost of all the cancers, most of which it did not
cause.
2. One solution: Make the reactor liable for the increased risk to everyone, using a class
action.
a. But diffuse costs are hard to litigate--class actions are driven by the interest of
the lawyers, not the victims.
b. And immediately after the risk has been imposed, evidence on how large it is
will not be very good--nobody has actually died, so we cannot start looking for
evidence of increased cancer rates.
3. What about giving each one who dies 1/11th damages?
a. When they start dying, we have at least some data.
b. Fewer cases to try.
4. Real world equivalent--DES cases.
a. DES was a fertility drug which turned out to have serious side effects.
b. By the time the effects were discovered, it was no longer possible to determine
which pharmaceutical company had manufactured the DES that a particular
patient had used.
c. So if a victim sued a company, the company could (correctly) claim that it had
probably not been responsible for that woman's injury.
d. Solution--divide the liability among the companies in proportion to what
fraction of DES each had produced.

IV. Liability
A. Optimal supply of accidents: The accidents we would get if everyone took all and
only cost-justified precautions.
B. Strict liability gives you the right answer with unicausal accidents.
C. Negligence applied to unicausal accidents.
1. Negligence is defined (by economists, but not necessarily by legal scholars) by the
Hand formula (named after Judge Learned Hand, who applied it in a famous case) (T.J.
Hooper)
a. Hand says that you are negligent if you failed to take a precaution whose cost
was less than the expected benefit in accident reduction, in other words ...
b. If you failed to take all cost justified precautions.
2. The rule "you are liable if negligent" gives the right answer if everything is observed
by the court,since:
a. Either you take all cost justified precautions and are not liable, or ...
b. You don't take all cost justified precautions and are liable, and since you are
liable you bear the full cost of the accident, and since you bear the full cost of the
accident it is in your interest to take all cost justified precautions.
c. So you do take all cost justified precautions, which is the outcome we want, but
...
2. What if the efficiency of some classes of precautions is observable by the actor but not
the court?
a. The usual example is "activity level." The court may be able to observe how
many trips you take, but not how much it is worth to you to take them.
b. Another example would be how much attention I pay to my driving.
c. Under a negligence rule, you take the optimal level of the observable
precautions, you will therefore not be found negligence, you therefore ignore costs
to other parties in choosing the level of the unobservable precautions.
3. Negligence could lead to either more or less litigation cost than strict liability.
a. More because there is one more question to be settled (negligence)
b. More because there are more accidents, due to the problem of activity level and
unobservable precautions
c. Less because accidents where the party responsible was obviously not negligent
do not result in lawsuits--the plaintiff has no claim.

V. Complications in tort damage analysis.

A. Double sided injury in a literal sense--both cars are dented.
1. True of auto torts, but ...
2. Not of most others--pollution, nuisance, ....
3.We could include the problem by treating each such collision as two torts, one each
way.
4. We will avoid this problem for the moment by considering collisions between a tank
and a car--in which only the car is injured.
B. While literal double injury is a special case that applies to auto collisions but not
to most other torts, it is generally true that there are possible harms to both parties in
a very different sense--avoiding injury to you is costly to me.
C. Double sided causation--a much more fundamental problem.
1. We can solve the problem by a legal rule of negligence or strict liability with
contributory negligence--provided the court knows enough to properly implement such a
rule.
2. Or we can avoid it by having each party bear the full cost--i.e. a fine rather than a
damage payment, so the tank pays the cost of the damage done to the car, but the owner
of the car doesn't get the money and must pay for his repairs himself.
3. In which case the tank owner bribes the car owner not to report the accident--and we
have left tort law for criminal law, where enforcement is by the state rather than the
victim.

VI. Liability alternatives

A. No liability--the tank pays nothing, the car owner pays for his own repairs.
B. Strict liability--the tank pays for fixing the car it ran over, whether or not its
driver was negligent.
C. Negligence liability--the tank pays if its driver was negligent (i.e. Did not take all
cost justified precautions), but not otherwise.
D. Strict liability with contributory negligence--the tank pays unless the car driver
was negligent.
E. Note that A and B are really the same rule, with tank and car exchanged.
F. As are C and D.

VII. Summary of implications:

A. Assume one dimension (care) is observable by the court, another (activity level)
is unobservable.
1. Note that "care" and "activity level" are merely convenient examples.
2. The real categories are "activities with regard to which the court can tell whether or
not the actor was negligent" and "activities with regard to which the court cannot ... ."
B. A sketch of the implications of the alternative rules. Notice the symmetry; the
first two columns are the same, with the roles of the actors reversed, and similarly
for the second two.

Is There An No Strict Negligence Strict with

Optimal Liability Liability Contributory
Level of? Negligence

Care by
No Yes Yes Yes
Tank
Tank Act No Yes No Yes
Level
Care by Car Yes No Yes Yes
Car Act Yes No Yes No
Level

C. Why are there accidents? In general, because our assumption that both courts and
individuals have perfect information is not true.
1. The court might be wrong about either what precautions the tortfeasor took or what
precautions he should have taken, and so find him negligent when he was not.
2. The tortfeasor might be wrong about what precautions he should have taken, and so
really be negligent. Or ...
3. The tortfeasor might gamble on the court thinking he wasn't negligent when he really
was--and lose.
D. The "reasonable man" standard.
1. One form that imperfect information by the courts takes is the rule that negligence is
judged according to what precautions would be cost effective for an imaginary
"reasonable man" rather than for the actual tortfeasor, since the court doesn't know
whether the tortfeasor has better or worse reactions, higher or lower alcohol tolerance,
etc. than the average.
2. Suppose you have much better reflexes than the average:
a. One half of the argument for keeping down to what the court considers a safe
speed is that if you don't, you will be liable for damages if you run into someone.
That argument still applies to you.
b. The other half is that if you are liable, it is then in your interest to take all cost-
justified precautions. But keeping down to what the court considers a safe speed is
not cost-justified for you.
c. So you either keep down to the court's speed, in order to make sure that if there
is an accident you won't have to pay for it, or
d. Drive at the efficient speed, which is faster than the court things, knowing that
you will be liable if there is an accident.
3. Prediction: Courts should (if Posner is right about the efficiency of the law) tend to
use strict liability where actors vary a lot, such that the "reasonable man" standard works
badly for many of them.
4. More generally, strict liability not just where activity level is important (Posner's
argument), but more generally where unobserved variables are important.
E. Effects of court error depend on the form of the error:
1. Suppose the court correctly measures your precautions but sometimes overestimates
or underestimates the efficient level of precautions. The higher the level of precautions
you take, the greater your chance of not being found negligent, hence liable. This tends
to push you to a greater than optimal level of precautions.
2. Suppose the court simply makes mistakes at random--a third of the time it decides that
the tortfeasor was not negligent, whether or not he actually was. The result is like a strict
liability rule, with a punishment equal (on average) to 2/3 of the damage done. So
tortfeasors are underdeterred, and take less than the efficient level of precaution.
3. So it is hard to work out a theory that takes account of court error--the effect depends
on the details of what sort of mistakes the court makes.

VIII. Amount of damage payment awarded:

A. "Enough to make good the injury." Or in other words, enough to make the victim
as well off as if the tort had not occurred.
1. This is the right rule if we catch, sue, and win every time.
2. Otherwise it gives too little expected punishment since sometimes tortfeasors don't
end up paying damages, so there is too little deterrence.
3. Why is this the standard rule in tort damages?
B. Punitive damages--more than enough to make good.
1. Very uncommon until recently.
a. N.H. Supreme Court, and others, argued that "punitive damages" were ...
b. A misunderstanding of damages for non-pecuniary injuries.
2. They are supposed to be awarded for a "deliberate or reckless" tort, and the amount is
up to the court.
3. Old examples:
a. King's messenger case:
i. The King's messengers forced their way into someone's house, held him
prisoner while they searched his papers in an (illegal) attempt to find out if
he was the author of articles in a newspaper they considered subversive.
ii. He sued, they argued that they only owed him for actual damages, which
were small.
iii. He got punitive damages.
b. Insult case.
i. One landowner swore he would shoot birds on his neighbor's property
without permission
ii. And did
iii. And the neighbor was awarded punitive damages.
B. Non-economic explanations:
1. Punitive damages are really just compensation for hard to measure injuries--
reputation, etc. Shooting on your enemy's land without his permission shames him,
makes him unhappy, lowers his reputation, ... .
2. To express condemnation
C. Other people's economic explanations.
1. Punitive damages are a probability multiplier. If so, why not base punitive
damages on how large a fraction of those who commit a particular tort get away with it,
rather than on whether the tort was "deliberate or reckless?" A reckless tort should be
very likely to result in a successful suit.
2. Punitive damages are a way of playing safe if damage is hard to measure but
efficient torts are unlikely.
a. A deliberate tort requires a positive cost to commit. So the optimal level is likely
to be zero. But it doesn't have to be--there might be a large benefit.
b. An accidental tort requires a cost to avoid, so the optimal level is likely to be
>0.
c. A deliberate tort, unlike an accidental one, can usually be replaced by a market
transaction--the tortfeasor knows who he is going to injure and should be able to
buy permission if the tortfeasor's gain is more than the victim's loss.
d. Or in other words, punitive damages are a way of enforcing a property rule
within the legal framework normally used to enforce liability rules.
e. But what about a tort that is both deliberate and efficient--polluting, when you
know it is tortious, because you are willing to pay the damage and there are too
many victims to make a voluntary transaction possible. Or throwing sparks in the
cases where doing so--and paying damages--was efficient.

IX. My Explanation 1: Punitive damages are for very deterrable torts.

A. Economic theory of optimal damages with litigation etc. costs
1. Without such costs, expected punishment (<P>) should = damage done. In civil law,
<P> is the damage award times the probability that the victim will sue and win.
2. Deterring the marginal tort at that point does no good, so ...
3. If detering it is costly, don't do it.
4. Cost depends on elasticity, because increased award decreases number of cases but
increases cost per case.
5. If supply is very inelastic, there is a positive cost to deter. So you only want to deter
very inefficient offenses, so you set <P> < damage done
6. If supply is very elastic, you want to deter even offenses that are (slightly) efficient, in
order to avoid the cost of prosecuting them, so you set <P> > damage done
7. So the optimal expected punishment is below or above the damage done according to
how elastic the supply of offenses is.
8. This is worked out in more detail in my punitive damages piece and my "Payne v
Tennessee" piece.
B. A deliberate tort is very deterrable, so the supply of offenses is elastic, so you
should overpunish
C. An accident may not be very deterrable, so the supply of offenses is inelastic, so
you should underpunish.
D. The case where it is efficient to overpunish a lot, deterring almost all offenses and
leaving almost none to be punished, brings us to a property rather than a liability
rule.

X. My explanation 2: Punitive damages are for strategic torts.

A. Why did I beat him up, even though I knew I would have to pay damages?
B. To deter other guys from going out with "my" girl.
C. This only works if ordinary damages undercompensate, since otherwise he
doesn't mind being beaten up--and paid damages.
1. But they do undercompensate.
2. Since they have no probability multiplier and ignore pain and suffering (until recently)
3. And they should undercompensate, since ordinary damages are for offenses with an
inelastic supply curve (or so I have just argued above).
4. This is the same argument used earlier as an explanation of one reason why damages
should go to the victim. Remember the strategic railroad case, back when we were
talking about the difference between damage payments and fines?
D. So we want extra damages in such cases, to deter strategic torts.
E. This makes sense where the tortfeasor is a repeat player.
1. The insult cases--maybe the two landowners were really fighting over local political
control and the like. Perhaps one of them was trying to shame the other so as to prove to
other locals that they had better support his candidate for parliament--or else.
2. The first punitive damages case I know of--king's messenger, North British Enquirer.
The crown wanted to discourage people from writing for subversive newspapers.
3. But it does not work against our equivalent of the king's messengers--law enforcement
agents.
a. Governments are protected from tort liability by the principle of Sovereign
Immunity
b. The federal torts act waives sovereign immunity in some cases, but not to the
extent of permitting punitive damages.

XI. Why pay tort damages to the victim instead of as fines to the state?
A. So he will sue. Private enforcement system
1. But he could threaten to sue in order to be paid by the offender to drop charges.
2. This may be what actually happened in 18th century (and earlier) English criminal
law, where prosecution was private but punishment was by the state.
3. But it works better if the victim owns the case--smaller bargaining range.
4. We could give the right to sue and collect to anyone--but the victim is the one most
likely to know that the tort has happened, and he has the additional incentive of
deterrence.
B. To affect the incentives of the victim?
1. That is a good reason in the strategic case with one tortfeasor who, if damages go to
the state instead of the victims, can get the many victims to take precautions if he refuses
to. This is one way of controlling strategic torts.
2. But just the opposite is true in the non-strategic case (auto accidents). Paying the
damages to the victims reduces their incentive to take precautions to below the optimal
level.
3. Note the difference between an anonymous tort (auto accident, many small players)
and a named tort (deliberate trespass, conversion). where the tortfeasor knows exactly
who the potential victim is, and can bargain with him.
a. Named tort introduces strategic problems, but also ...
b. Possibility of Coase Theorem bargaining
c. Moving us towards a property rule instead of a tort rule.

XII. The problem of measuring and compensating damages for loss of earning
capacity, death, injury.
A. Lump sum vs periodic payments
1. A lump sum reduces disincentive effect, but
2. By the same token lowers the cost of fraud. Once the money is paid you get a ticket to
Lourdes and come back without your crutches.
B. Triple effect of injury
1. On income (readily dealt with)
2. On utility (compensate with cash)
3. On Marginal Utility of Income. This makes compensating for the loss of utility
difficult, expensive, inefficient, perhaps impossible.
4. And explains why the price of a life seems infinite if you try to buy one. It is not that
the value of life is infinite but that the value of money to a corpse is zero.
C. Full compensation vs optimal insurance vs optimal deterrence.
1. Suppose it takes $100 million to make up to you for loss of your sight.
2. You would not insure yourself for that amount.
3. You would not take the precautions, in a case where your actions may risk your sight,
that are implied by that price.
4. So full compensation over insures and over deters.
5. But what if you would not insure at all (death for someone with no dependants)? Does
it follow that the optimal deterrence price is zero? Obviously not.
6. So these are three different values.
D. Correct solution in principle:
1. Set damages at a level that makes people as well off ex ante with the risk as without,
as judged by risk premia for jobs and similar criteria.
2. Allow the sale of inchoate tort claims. Now if my tort claim leaves me "overinsured" I
can sell part of it, so as to consume some of the money now when I am not
dead/blind/whatever.
E. My article which discusses this at much greater length is on the web, linked to the
class page.

XIII. Criminal Law:

A. Like tort law, someone does a wrong, we impose a cost on him, but ...
1. Enforced by the state, not by the victim. Legally speaking, the state is the victim.
2. State collects fine, if any.
3. Mostly punishment rather than mostly fines.
4. Moral stigma?
5. Higher standard of proof.
B. Note that the Tort/Crime distinction may be less clear than we usually think.
1. Traffic tickets are state imposed, collected, fines w/o moral stigma.
2. Criminal charges are often dependant on the victim complaining, and ...
3. In some cases (Michael Jackson?), de facto out of court settlement is possible.
4. There have been other societies where criminal law was privately prosecuted, or even
...
5. Where all offenses were treated roughly as we treat civil offenses.
C. Simple story: Enforcing property rules.
1. Optimal level of some offenses is close to zero
2. Most obviously because there is a market substitute, so if the result that would be
produced by the offense (your using my land) is efficient it will happen that way even if
the offense is prevented.
a. Theft is the obvious example.
b. Assault not so clear.
c. Would we accept consensual market murder? A rich man offers ten adventurers
$100,000 each, in exchange for their agreement that he may select one of them at
random and try to kill him.
3. So we set very high punishment, > damage done, to deter everything. But ...
D. That doesn't describe the real system, since:
1. Expected penalty for murder is <damage done.
2. Lots of crimes happen--level is not zero. Do we think they are efficient crimes?
3. So what determines the optimal punishment?
4. If it were =damage done we again have a problem--that implies that the reason we
don't raise the punishment for murder is concern that there would be too few murders,
which seems implausible.
E. Taking Enforcement costs seriously:
1. The effective punishment depends on probability and punishment, increases with
both. Think of it as the certainty equivalent of the punishment lottery; I will talk as if
people are risk neutral to make the analysis simpler.
2. We can increase probability by spending more on police, lowering standards of proof,
...
3. We can increase severity of punishment--possibly at a cost:
a. What is punishment cost?
b. Cost to punishee minus gain (plus cost) to punisher.
c. Zero for costlessly collected fine, 100% for hanging, 150% (or whatever) for
imprisoning.
d. Think of punishment inefficiency as the ratio of expected punishment cost to
effective punishment.
4. Becker corner paradox: Why don't we always gain by doubling the punishment and
reducing enforcement expenditures until the probability falls to half its previous level?
5. Solution to paradox: Because punishment inefficiency rises with the amount of
punishment. So apprehension cost will be going down, punishment cost up, and we set
the probability/punishment combination to minimize the sum for a given level of
effective punishment.
6. Litigation cost should also be included, perhaps with enforcement cost.
7. Bottom line: Enforcement cost per offense typically increases with effective
punishment.
F. Analysis of optimal punishment with enforcement costs (repeat of yesterday)
1. MC(deterrence) might be positive or negative.
2. If the effective punishment is P, the result is to deter all offenses for which
value to the criminal V <P, and only those.
3. Suppose we increase to P+ P, deterring one more offense at cost MCd(P)
4. We gain D (damage averted) at a cost of V (to marginal offender) + MCd(P).
5. Keep doing it until further increases just break even--optimal punishment
P*. Note that V=P. So we have: D=P*+MCd(P) => P*=D-MCd(P).
6. This is worked through in more detail in my "Payne v Tennessee" article.
G. Applied to criminal law, it tells us that:
1. Optimal punishment might be more or less than damage done.
a. Less to permit offenses that are not worth the cost of deterring, or ...
b. More to deter efficient offenses to save the cost of punishing them.
2. So the optimal punishment depends on damage done and (through MC(deterrence)) on
supply curve for offenses.
3. If at some point the remaining offenses are very hard to deter, we might stop there--
even though the remaining ones are inefficient.
4. Note two meanings for "inefficient crime."
a. Crime we would rather the offender not commit--what an offender trying to
maximize economic efficiency would care about.
b. Crime that is worth the cost of deterring it.
H. Why have criminal law at all?
1. Hard to get people to prosecute if the defendant is judgement proof, and ...
2. If you put in probability multipliers, offenders who commit crimes that are hard to
prove are particularly likely to be judgement proof.
I. Other possible reasons.
1. Diffuse injury. But that does not apply to most of criminal law, and does apply to
some of tort law--dealt with by class actions, could be dealt with by transferrable tort
claims.
2. To express moral outrage?
3. To get the optimal probability/punishment bundle.
a. The theoretical solution--for each level of deterrence (i.e. of effective
punishment), find the combination of probability and punishment that gives that
amount of deterrence at the lowest (punishment cost + apprehension cost) cost.
The result is the total cost function for deterrence--just like deriving an ordinary
TC function from a production function and input prices.
b. In a private system, the fine is both the punishment and the payment to the
enforcer, so you can't separately set punishment and probability, so ...
c. You get the Landes/Posner argument on the inefficiency of private enforcement.
d. And my solution to it: Let the court set the effective punishment not the actual
punishment. The enforcement firm buys a bunch of offenses from the victims,
pays to catch them, imposes the punishment that combined with the fraction it
catches yields the right effective punishment. Punishment paid by the offenders is
being fixed, the firm gets punishment collected (fines minus cost of jails etc.), pays
for apprehension, so apprehension+punishment cost is a subtraction from its profit,
so it minimizes the sum, giving us the efficient probability/punishment
combination.
e. Which runs into a problem if the market price of an offense is negative, and ...
f. Victims are anonymous.
4. But note that we have no theoretical reason to expect optimal litigation expenditure in
ordinary tort law, which is the analogous problem.
5. And inefficiency is relative to ideal, not to real alternative.
6. Another reason for criminal law is to permit probability multipliers, giventhe fake
offense problem.
7. But criminal law is only a partial solution. Police can frame you, get paid off in
money or information or assistance for not hanging you.
J. Why do we have a higher standard of proof for criminal law?
1. Because false positives are more expensive when you are using more costly
punishments.
2. In fact, in an efficient system,
3. The cost of punishing an innocent defendant is punishment cost.
4. The cost of acquitting a guilty is apprehension+litigation cost.
5. So the number of one error that trade off against another is the ratio of the two
numbers.
K. Consider the special nature of stigma as a punishment.
1. True stigma has negative punishment cost!
2. False stigma has high punishment cost.
3. Stigma is an important punishment--Lott studies actually measured it.

XIV. Why benefits to criminals count:

A. The lost hunter problem.
1. If a hunter is lost and starving and comes to a locked, empty cabin containing food
and a telephone, it is efficient for him to break in.
2. Because the benefit to him is greater than the loss to the owner.
3. Which seems right--but only if we are allowed to count the benefit to him.
B. Even if you want to deter it, how much do you want to deter it?
1. Deterrence is costly.
2. Its benefit is the net gain from deterring an offense.
3. Which is greater the greater the damage done to the victim, but also ...
4. Greater the less the gain to the offender.
5. We are willing to pay more to deter an act of arson committed for the fun of watching
buildings burn than to prevent a hungry man from stealing food--and would be even if
the total damage done to the victims was the same (the hungry man steals lots of food
over a period of years, say).
C. How do we know what is bad before we know what is inefficient?
1. Part of the attraction of the economic analysis of law is that we can start with one very
simple "moral" assumption--the desirability of economic efficiency--
2. And deduce from that that robbery and murder ought to be illegal, contracts ought to
be enforced, ...
3. If we start out already knowing that robbery is bad, hence gains to robbers don't count,
we are rigging the game--guaranteeing that we will get out as efficiency what we put in
as ethics.

XV. Rich vs Poor--should they pay the same fines? In some cases yes, but in others no,
because:
A. For crimes with a payoff in time or utility, the supply curves are different--it
takes a higher expected $ punishment to deter the richer offender.
B. It is cheaper to punish richer offenders, since they can pay fines, but ...
C. On the other hand, they can also hire good lawyers so as to make it costly to
convict them.
D. So the optimal penalty for the rich might be higher or lower than for the poor.

XVI. Marginal deterrence: My article (with William Sjostrom).

A. If armed robbery gets the maximum possible punishment, then the additional
punishment to an armed robber for killing the victim is zero.
B. More generally, in setting optimal punishments we must allow for the possibility
that we may deter someone out of one crime and into another.
C. Which makes the calculation of an optimal set of punishments more complicated.

XVII. The Paradox of Efficient Punishment:

A. Why not hang them all?
1. Argument for efficient punishments.
a. If we shift to a lower cost punishment probability combination giving the same
deterrence ...
b. The expected cost to the criminals is the same
c. The crime rate, hence the cost to the victims is the same
d. And the enforcement system saves money.
2. What would such an efficient system look like?
a. Fine if collectable.
b. Make it collectable by threat of worse punishment if they can't pay.
c. Temporary slave labor if they cannot pay a fine.
d. Execution if they cannot produce more as slave labor than they cost to feed and
guard.
e. And after you execute them, use their organs for transplants.
3. English 18th century example.
a. No imprisonment for serious crime.
b. Hang, transport, or pardon.
c. No imprisonment for serious crime.
d. Costs of slave labor too high to make imprisonment an efficient crime... It looks
as though galley slaves (easy to guard and monitor) more than paid their costs,
other forms of slavery for violent criminals did not. The English didn't use galleys,
because the Atlantic is too rough for them. So their only cheap imprisonment
option was transportation. For more details, see my article on 18th c. English law.
4. If the sort of efficient system of punishments I have described is a bad idea, as most
people believe, why?
B. Why we should not hang them all.
1. The old mistake:
a. Economists in the past have frequently answered the question "what would I
like the government to do?" instead of the question "what should the legal rules
be, given how government will act if those are the legal rules?"
b. In other words, the government was being treated as a philosopher king rather
than as a human institution controlled by people with the same sort of limitations
and incentives as the rest of us.
c. Optimal punishment theory makes the same mistake by ignoring the incentives
of the enforcers: police, courts, legislatures.
2. The costs of efficient punishment--other side of the problem.
a. An efficient punishment, such as a fine, provides a benefit to someone on the
other side, who is collecting it.
b. That is an incentive to convict or punish someone, whether or not he is actually
guilty.
3. Rent seeking: In a world of really efficient punishments, everyone's human, biological
and economic capital is up for grabs all the time.
a. If I can properly manipulate the legal system, I get to seize you and your
property.
b. So we all spend lots of resources defending ourselves and trying to expropriate
others.
4. Civil forfeiture and punitive damages for product liability are modern examples of this
problem.
5. Larry Niven offers a fictional example: a society where the organs of those executed
forfeit to the state to be used for transplants.
a. It looks like an unambiguous improvement--why let organs that could save lives
rot?
b. But it produces an incentive to increase the number of capital offenses so as to
get more organs and save more lives.
c. The protagonist of the story has been sentenced to be executed for his third (I
think?) dangerous driving conviction.
6. The application of the argument to the question of whether to hang them all is more
difficult, since nobody gains--hanging simply eliminates most of the cost of punishment
to those other than the criminal.
a. If government institutions are driven by a weighted average of citizen values,
with the weighting somehow depending on political influence, then a cheap
punishment imposed on uninfluential people might be badly overused.
b. A cheap punishment might be transmuted into a profitable punishment, if the
threat of execution is used to extort a (legal or covert) out of court settlement.

Review before the Second Midterm

I: Determining property rules involves a set of questions:

A. How rights should be bundled (Coase article, our earlier discussion)
B. How should you be allowed to defend your property rights (property vs liability)
C. What should be property?
 1. Property vs commons--things, words, IP in the middle. Primitives. All you can eat.
2. Defining and measuring boundaries. Floating island. Underground oil pools.
3. Defending: Bionic burglar alarm. Pirating software.
4. Transaction costs: Property in large animals, not the land they are hunted over.
English--quiet story. All you can eat restaurants.
5. Creating: Rent seeking. homesteading problem. Inefficient patent races.
6. Supply elasticity--how much do we need the incentives provided by property for
production?
D. IP sketch:
1. Copyright automatic, long term.
2. Patent grudging, short term. Requires "sufficiently big innovation" to count.
3. Trade secret if you guard it. Permanent. But very limited protection.
E. Fits in because:
1. Copyright has no homesteading problem (infinite commons), easy boundaries.
2. Patent has a limited commons, risk of parallel invention, cost of patent races, fuzzy
boundaries.
3. Computer programs fit under copyright not because they are writings but because they
have the characteristics described.
4. Trade secrets not preempted, because they plug some of the holes in the patent system.

II. Coase Article: Explains the bicausal problem, the "all alternatives imperfect"
problem, argues that the common law courts got things about right.

III. Contract Law

A. We want enforceable contracts because performance occurs over time, with
transaction specific investments. House building
B. There are private ways of enforcing, via arbitration, reputation, bonding,
hostages, etc., which are important, but don't cover all of the problem.
C. We can't just let the parties define all of the rules, both because there isn't enough
fine print to cover all alternatives (we need default rules) and because we need to be
able to decide whether there is a contract to enforce. And maybe because we want to
control the terms?
1. Argument for freedom of contracts: It is in the parties' interest to find and include the
efficient terms, and they are more competent to do it than the judge.
2. And the judge can't benefit favored groups even if he wants to, because ex ante the
parties will include his bias in setting the contract price.
3. Some exceptions, such as incompetent parties, third party effects, maybe monopoly
using terms for dp.
D. Duress:
1. Real duress. Arguments for and against enforcing such contracts. Consider the "social
contract." Parole.
2. Sinking ship: Optimal terms somewhere in between, bargaining costs high, so let the
court set them.
3. Contracts of adhesion:
a. Thought of as duress bc not bargained--but argument for efficiency still applies.
b. And lots of good reasons to have form contracts.
E. Is there a contract to enforce?
1. Fine print etc.--problems we haven't much discussed.
2. One sided contracts--detrimental reliance. Reward offers.
3. Constructive contracts--emergency medical treatment.
F. How to fill in the details.
1. The efficient rule, both as predictor, because a good idea, and to avoid incentives to
write long contracts.
2. Risk bearings--as per insurance arguments.Himalayan photographer.
3. Breach rules: Nothing, Specific performance, Expectation, Reliance, liquidated (incl
penalty)
a. Breach/don't breach margin--expectation is right.
b. Rely don't rely margin, nothing or liquidated or (maybe) specific performance
right.
c. Liquidated ideal if you can predict costs of breach in advance.
d. Penalty clause=property rule privately chosen.
G. Fraud and informational issues. Speculation. What must you tell people?

IV. The Marriage Contract:

A. Why does it exist: Relationship specific sunk costs.
B. What is the optimal form--in particular, rules for breach.
1. No penalty risks opportunistic breach.
2. Mutual assent risks opportunistic degradation of services. Analogous to utility
problem.
3. No divorce risks people getting stuck--or not married if there is a risk of problems.
4. Damages set by the court require decision of how is responsible, measurement of
damages--all very hard for a court to do.
5. Freedom of contract solution?
6. And there are third party effects on children.
C. Why has the actual contract changed over time, and with what consequences?
1. Less sunk cost due to reduced infant mortality, increased division of labor.
2. More job opportunities for women (cause or effect?)
3. Welfare--explains the low end but not the high end.
4. Contraception and abortion--wrong prediction of which end starts it.
D. Wedding rings as a performance bond.
E. Regulation of adoption? Commodification.
F. Population externalities?

V. Tort Law: Four issues arise:

A. What makes it wrongful: Competition is not a tort. Some system of rights is
assumed, and a tort involves a violation. Externalities that are best handled via a
privately prosecuted liability rules. As opposed to victim buying right, or property
rule enforced by criminal law.
B. Causation: What does it mean to say A caused the harm? Complications. That
giving him an incentive to avoid the harm will be useful.
1. Falling safe.
2. Two hunters shoot a third.
3. Probabilistic injury. Problem to be solved.
C. Liability--strict liability, negligence, contributory negligence, etc. How to
generate ptimal precautions.
1. For one sided accidents, strict liability works (aside from litigation costs, court error)
2. So does negligence (if courts can measure it) or contributory
3. Negligence or strict with contributory negligence solve the two-sided causation
problem--provided the court can judge negligence. Activity level problem where they
cannot observe or judge. Can solve one side but not both.
4. Reasonable man standard another reflection of court ignorance.
D. Damages--how calculated?
1. To make the victim whole, but ...
2. Not quite the right rule if we take account of litigation costs.
3. Wrong if probability of paying is less than one.
4. Punitive damages might be:
a. Solution to probability problem--but doesn't fit when they are awarded.
b. For acts we are sure are inefficient--but why necessary? Maybe for acts we are
sure we are inefficient and can't measure damage for--back to NH Supreme Court
theory that there are no punitive damages.
c. For especially deterrable acts.
d. To deter strategic torts--bullying.
E. Why pay to the victim?
1. Incentive to prosecute.
2. Victims incentives--good in strategic case, otherwise bad.
F. Damages for injury problem.
1. Lump sum vs periodic.
2. Not infinite value but zero value of money to a corpse.
3. Want optimal deterrence and insurance--get it by combining tort and market.

VI. Criminal Law

A. Simple story--enforcing property rules.
B. Simple story: Enforcing property rules.
1. Optimal level of some offenses is close to zero
2. Most obviously because there is a market substitute, so if the result that would be
produced by the offense (your using my land) is efficient it will happen that way even if
the offense is prevented.
a. Theft is the obvious example.
b. Assault not so clear.
c. Would we accept consensual market murder? A rich man offers ten adventurers
$100,000 each, in exchange for their agreement that he may select one of them at
random and try to kill him.
3. So we set very high punishment, > damage done, to deter everything. But ...
C. That doesn't describe the real system, since:
1. Expected penalty for murder is <damage done.
2. Lots of crimes happen--level is not zero. Do we think they are efficient crimes?
3. So what determines the optimal punishment?
4. If it were =damage done we again have a problem--that implies that the reason we
don't raise the punishment for murder is concern that there would be too few murders,
which seems implausible.
D. Taking Enforcement costs seriously:
1. The effective punishment depends on probability and punishment, increases with
both.
2. We can increase probability by spending more on police, lowering standards of proof,
...
3. We can increase severity of punishment--possibly at a cost:
4. Bottom line: Enforcement cost per offense typically increases with effective
punishment.
E. Analysis of optimal punishment with enforcement costs (repeat of yesterday)
F. Applied to criminal law, it tells us that:
1. Optimal punishment might be more or less than damage done.
a. Less to permit offenses that are not worth the cost of deterring, or ...
b. More to deter efficient offenses to save the cost of punishing them.
2. So the optimal punishment depends on damage done and (through MC(deterrence)) on
supply curve for offenses.
G. Why have criminal law at all?
1. Hard to get people to prosecute if the defendant is judgement proof, and ...
2. If you put in probability multipliers, offenders who commit crimes that are hard to
prove are particularly likely to be judgement proof.
H. Other possible reasons.
1. Diffuse injury. But that does not apply to most of criminal law, and does apply to
some of tort law--dealt with by class actions, could be dealt with by transferrable tort
claims.
2. To express moral outrage?
3. To get the optimal probability/punishment bundle.
I. Why do we have a higher standard of proof for criminal law?. Because false
positives are more expensive when you are using more costly punishments.
J. Consider the special nature of stigma as a punishment. Very efficient punishment--
but only if it is imposed on guilty people.
K. Rich vs Poor--should they pay the same fines? In some cases yes, but in others
no, because both supply curves for offenses and punishment costs vary with income.
L. Marginal deterrence: My article (with William Sjostrom) is accessible from the
web page.

VII. The Paradox of Efficient Punishment:

A. Why not hang them all?
B. Why we should not hang them all. Efficient punishments risk setting off a rent-
seeking struggle, with some people trying to appropriate other people's physical,
human and biological capital.

Forward to Part II of the Outline

Back to the Table of Contents

Back to the class home page