Econ 522 – Lecture 15 (Nov 4 2008)
Midterm is not yet graded. Hopefully by the end of the week.
One point I want to quickly revisit.
There are two separate questions we addressed when looking at the question of breach:
when do we expect a promisor to breach?
when would it be efficient for a promisor to breach?
These are different questions, and may have different answers.
We expect a promisor to breach when it’s cheaper to breach than to perform, that is,
when his liability for breach (presumably, the damages he’ll owe) is greater than the cost
of performance. This has nothing to do with efficiency – it’s just one guy, doing what is
best for him, given the options he faces.
Now, the question of efficiency involves summing up everyone’s payoffs – in this case,
the promisor and the promisee. It’s efficient for the promisor to breach when the total
payoffs are higher when he breaches than when he performs. Since it’s usually the
promisor who’s incurring costs and the promisee who’s getting benefits, this will
typically be when the cost of performing is higher than the benefit of performance.
So the challenge, when designing a remedy, is to make the two solutions match – that is,
to make the promisor, when he’s just thinking about himself, breach exactly in the times
when it’s efficient for him to breach. But it’s important to keep the two concepts distinct
– the promisor will do what is in his best interest, and the challenge is to make that match
up with what’s efficient. We’ll face the same challenge with tort law as well.
Today, we begin tort law.
The word tort is from French, and means injury. With contract law, we studied situations
where someone injured you by breaking a promise they had made. With tort law, we will
look at situations where someone harmed you without having made any promises.
A line I like from the Friedman book:
“If someone shoots you, you call a cop. If he runs his car into yours, you call a lawyer.”
The line between the two isn’t always that clear. Some torts are crimes as well – if
someone assaults you, they may go to jail, and you may also sue them for damages.
(Plus, there’s the OJ Simpson case.) We’ll get to criminal law in a few weeks; for now
we’ll focus on the tort side.
Before we get to the material, though, I’d like to do a quick experiment.
As with nuisance law and contract law, our focus in torts will not be on fairness or equity,
but on efficiency, that is, on structuring the law to provide correct incentives for people to
For deliberate harms, this will usually be fairly simple.
For accidental harms, though – traffic accidents and things like that – setting the
correct incentive for avoiding harm is tricky.
Unlike in nuisance law, injunctive relief is often not an option
Much as we’d like to imagine that nobody could run into your car without
negotiating for permission, it doesn’t work that way
Similarly, unlike with contract law, there is no promise or agreement made ahead
of time; something happens, and then we deal with it.
Cooter and Ulen explain tort law as covering all the situations where transaction
costs were too high to allow for bargaining
o It’s impossible to negotiate with every driver for the right to hit you
o it’s not always possible to bargain with an angry drunk about whether or
not he breaks your nose.
Thus, Cooter and Ulen define the essence of tort law, at least economically, as the
attempt to make injurers internalize the externalities they cause, in situations
where transaction costs are too high to do this through property or contract rights.
With contract law, we started out by discussing a “classic” legal theory of contracts (the
bargain theory), and then moved on to an economic analysis. With torts, we’ll do the
The traditional theory of tort liability was developed about 100 years ago. It specified
three key elements of a tort, which must be present for he plaintiff to recover damages:
breach of duty
First, and simplest, the plaintiff needs to have been harmed.
Without harm, there is no tort.
The book gives an example of a gas company that sold gas with a defective
additive, which has no effect on cars with normal carburetors but is dangerous in
cars with turbocharged carbs.
The owner of a car with a normal carburetor might be outraged by the situation,
but since the gas did no damage to his car, he cannot sue.
Similarly, traditional tort theory does not compensate for exposure to risk.
Suppose a manufacturer accidentally exposed a bunch of workers to a chemical
that doctors agree will cause 15% of them to develop liver cancer later in life.
Under traditional tort doctrine, the victims must wait and see who gets cancer
Similarly, if I do something negligent that causes you a 50% chance of dying, but
you get lucky and escape the situation, I’m generally not liable for any damages.
(So far, U.S. courts have been reluctant to allow tort actions on the basis of
exposure to harm, although some believe they should. Thus, safety regulation,
administered by a government agency, is typically used instead of tort law to deal
with exposure to low-probability harms.)
In theory, harm has a simple economic interpretation: a downward shift in the victim’s
utility. Like in the hairy hand case, we can imagine a victim has some utility function
over health and money, with some sort of indifference curves:
A car accident might lower both my wealth (because my car is worth less) and my health
(because I got whiplash), moving me to a lower indifference curve.
Perfect compensation should restore me to my original level of utility.
Money damages are typically how this is done.
Obviously, some amount of money can get me back to my original level of wealth
(by repairing whatever was broken on my car).
If there is medical care available that would restore me to my original level of
health, then the cost of this treatment would get me back to my original level of
health as well, so I’d be back where I started.
Of course, there are some harms that can’t be undone – the amputated leg in the
experiment, or other permanent injury
but there may be some amount of money that would compensate for it, by getting
me back to the same indifference curve I started out on.
Just like in the hairy hand case, we can work out perfect damages in this way.
So that’s the theory – the notion of “perfect compensation.”
In reality, courts have traditionally been very willing to compensate victims for
tangible losses – medical costs, lost income, damaged property
but less willing to compensate for intangible losses or losses that are difficult to
measure, such as emotional harm, pain and suffering, or loss of companionship
Over the years, however, American courts in particular have expanded the list of
compensable harms to include many of these intangibles.
One area where there is a huge disparity between tangible harm and actual harm is
in the death of a child
If a drunk driver kills a child, there is no loss of income to the child’s family; if
anything, they avoid the expense of raising the child.
This used to be a difficult problem for courts, as they could not find a tangible,
measurable harm on which to base compensation.
Allowing compensation for emotional distress and loss of companionship solves
In addition, once we start thinking about the incentives caused by tort liability, setting
liability closer to the actual harm done sets better incentive for harm avoidance.
On the other hand, intangible harms are extremely difficult to calculate.
Reasonable people might disagree dramatically about the emotional harm of
losing a leg, much less a child.
In addition, some people are horrified at the very idea of putting a dollar value on
someone’s life; and clearly, there is no amount of money that would make most
parents indifferent to losing a child.
So while compensation for intangible harms is an appealing idea, it leads to a
great deal of disparity in awards, across different courts, or even within the same
court, for similar injuries.
That’s the first element of a tort under the traditional view – harm. The second element
that must be present is causation – that is, the defendant must have caused the plaintiff’s
Here, things can get pretty tricky. Suppose the factory where I work exposes me
to some chemical that increases my risk of some disease from 1% to 1.2%. I
come down with the disease. Did the chemical cause it?
The law distinguishes two types of causes. The first is “cause-in-fact.” The test
for cause-in-fact is generally the “but-for test”. That is, “but for the defendant’s
actions, would the harm have occurred?”
There’s an example in the Friedman book that shows how tricky even this simple
rule can be.
o Three friends go out hunting, one of them scares up a pheasant, the other
two turn and shoot and both, mistakenly, at the same exact time, shoot the
first one, one through the head, one through the heart.
o Each one is very apologetic to the man’s widow, but say, “If I hadn’t shot
him, he’d still be dead.” Did either of them cause his death? But for
either one’s actions, he’d still have died.
Another problem with the “but-for” test is that it doesn’t distinguish between
recent actions and more distant actions.
I punch a guy in the face in a bar. But for my punching him, his nose wouldn’t
have been broken.
Also, but for his parents having conceived him, he wouldn’t have been alive to be
punched in the face, and the harm wouldn’t have occurred. Are his parents liable?
But for him getting a job in my town and moving there three years ago, he
wouldn’t have been in the bar to be punched in the face – is his employer liable?
So for a tort to exist, the defendant’s act must not only be the cause-in-fact of the harm, it
must also be the proximate cause.
Proximate just means close – there’s no rigid definition for how close the cause
One famous case, Palsgraf v Long Island Railway (1928), addressed this problem.
The facts of the case were:
Plaintiff [Mrs. Palsgraf] was standing on a platform of defendant’s railroad after
buying a ticket to go to Rockaway Beach. A train stopped at the station, bound
for another place. Two men ran forward to catch it. One of the men reached the
platform of the car without mishap, thought he train was already moving. The
other man, carrying a package, jumped aboard the car, but seemed unsteady as if
about to fall. A guard on the car, who had held the door open, reached forward to
help him in, and another guard on the platform pushed him from behind. In this
act, the package was dislodged, and fell upon the rails. It was a package of small
size, about fifteen inches long, and was covered by a newspaper. In fact it
contained fireworks, but there was nothing in its appearance to give notice of its
contents. The fireworks when they fell exploded. The shock of the explosion
threw down some scales at the other end of the platform many feet away. The
scales struck the plaintiff, causing injuries for which she sues.
The court ruled that even though the Long Island Railroad guard’s actions in
pushing the passenger may have caused her injuries – he pushed a passenger who
dropped a package of fireworks which went off and caused the scales to fall on
her – his actions were too remote to be considered the proximate cause.
As I said, though, there’s no precise legal definition for how close the action must
be to the harm.
Another example from the Friedman book:
I run into a friend on the street, and we stop to talk for a minute. We say
goodbye, go our separate ways, and twenty seconds later, he’s crushed by a
falling safe. Clearly, but for talking to me, he would still be alive – he would
have passed that spot well before the safe fell. Did I cause his death? Am I
Friedman claims this hypothetical is based on a real case:
“A tree fell on a moving trolly, injuring passengers. One of them sued. He
succeeded in demonstrating that in order for the trolly to be where it was when the
tree fell on it the driver had to have driven faster than the speed limit at some
point during the trip. Breaking the law is per se negligence, so the driver was
legally negligent whether or not his driving was actually unsafe. If he had not
driven over the speed limit, the trolly would not have been under the tree when it
fell, so, the plaintiff argued, the driver’s negligence caused the injury.”
Thankfully, the court was not stupid, and the plaintiff lost. The court held that the
driver’s negligence “had not caused the accident in the legally relevant sense.” I’m not
really sure what that means, but it seems pretty clear that innocent actions which led to an
accident by chance, but did not increase its likelihood, should not be punished. In this
case, the driver’s speeding was just as likely to prevent the accident (by passing that point
before the tree fell) as to cause it. If we think about tort law as an attempt to set the right
incentives, there is no reason to discourage the driver’s actions.
This could be done in one of two ways. One, we could make him liable for the accident,
but make him reverse-liable (have him receive a large reward) if his speeding caused him
to miss a falling tree. Of course, it might be very hard to prove when he’s avoided an
accident; really much simpler just to not make him liable in either case.
That’s two elements of a tort under the traditional theory. The third element is breach of
duty, and is sometimes necessary and sometimes not.
When harm and causation are sufficient for a tort to exist, the rule is referred to as “strict
Under the common law, strict liability tends to be the rule for unusually
The book gives the example of a construction company blasting with dynamite –
the company is liable for any harm caused by the blasting.
In most cases, however, there is a third element that must be demonstrated in
addition to harm and causation.
It must be shown that the defendant breached a duty he owed to the defendant,
and that this breach led to the harm.
If I punch someone in a bar, I breach a duty not to punch them.
When an injurer breaches a legal duty, he is “at fault,” or “negligent.”
A liability rule requiring all three elements – harm, causation, and fault – is referred to as
a “negligence” rule.
Under a negligence rule, a defendant can argue that an accident occurred despite
his best efforts to prevent it.
That is, under a negligence rule, even if I caused you harm, I’m not liable if I
satisfied the applicable standard of care and the accident happened anyway.
(For accidents, the “duty” owed to the victim is the duty of care – when that duty
is breached, that is, when I am careless or negligent, I am liable; when I meet that
duty, even if you get hurt, I’m not liable.)
Of course, this leaves the question of who determines the applicable standard of care –
that is, how careful you have to be to not be considered negligent.
In some cases, there may be a clear yes/no test – the swimming pool had a
lifeguard on duty or it didn’t.
In many cases, there isn’t. Is driving a car 50 MPH on a particular road
negligent? What about 51? 52?
How is this standard of care determined?
In some cases, the government imposes safety regulations that set the standard –
speed limits for highway driving, the requirement that bicycles have brakes,
workplace regulations, and so on.
Some are left vague – the definition of “reckless driving” may depend on the
road, the time of day, the weather.
In these situations, the common law tends to focus on the duty of reasonable
Like foreseeable reliance, this is the level of care that a reasonable person would
have taken under the circumstances.
If you took less care than a reasonable person would have, you are liable; if you
took more care, you are not.
(The civil law tends to rely less on standards of “reasonableness”, and tries to be
more specific in what duties are owed and what level of care is required.)
Under a rule of strict liability, proof of negligence is unnecessary; proving harm and
causation is enough to establish liability. Under a rule of negligence, you must prove
harm, causation, and negligence.
Interestingly, the movement between these two rules has not always gone in the same
Strict liability was the usual rule in much of Europe before the 1800s.
By the beginning of the 1900s, negligence became the usual rule
but in the second half of the 1900s, strict liability experienced a rebirth, especially
for manufacturer liability in American consumer products.
Manufacturers in the U.S. are now held liable for harms caused by defective
products, whether or not they were at fault – we’ll see examples of this shortly.
(The Friedman book spends a surprising amount of time on the problem of who
should be responsible when a bottle of Coke explodes, which I guess used to be a
pretty big problem before they switched away from glass bottles.)
(OK to end here if short on time. If not…)
Obviously, the more carefully I drive, the less likely I am to hit you.
If transaction costs are low, we can negotiate ahead of time, not for my right to hit you,
but for how carefully I’ll drive.
Obviously, this doesn’t work in practice; too many drivers, and too hard to “commit” to
the correct level of care.
But, what will happen to me in the event I hit you will affect how carefully I drive; that
is, tort law creates incentives, and I will decide on my behavior as a response.
As we said earlier, the essence of tort law is to use liability to get injurers to internalize
the externalities their behaviors cause.
So far, we’ve seen two possible liability rules:
A strict liability rule, under which you’re liable for any harms which you caused
A negligence rule, under which you’re only liable for harms you caused if you
were negligent, that is, if you breached a duty, typically the duty of due care
That is, under negligence rule, you are not liable for harms you caused if you
satisfied a legal standard for care/precaution
so a negligence rule is also dependent on establishing a legal standard for how
much precaution is necessary.
Under the common law, this standard is often reasonable care, that is, the
amount of precaution a reasonable person would have undertaken.
Next, we introduce an economic model of precaution and harm, and look at the incentives
that different liability rules create.
To begin with, suppose there is some type of accident that may or may not happen – I
may run into you in my car, a manufacturer may accidentally use lead paint in its toys,
etc. There is also some amount of precaution that can be taken that makes the accident
x denote the level of precaution
p(x) the probability an accident occurs, given that level of precaution
(so p(x) is decreasing in x – more precaution = fewer accidents)
A be the cost of an accident
w the cost of each incremental unit of precaution (could be money, time,
Then the cost of precaution is wx, and the expected cost of accidents is p(x) A.
We can graph these against x, the level of precaution:
The total social cost due to this type of accident is the expected cost of accidents, plus the
resources spent in prevention:
Social Cost = wx + p(x) A
So the efficient level of precaution will be the level of precaution that minimizes this.
(Add social cost to the graph)
Calculus: to minimize wx + p(x) A, we take the derivative w.r.t. x, and find
w + p’(x)A = 0
or at the efficient level of precaution, x*,
w = – p’(x*) A
(LHS is marginal social cost of precaution; RHS is marginal social benefit; optimal level
sets marginal cost = marginal benefit)
Under reasonable assumptions…
When x < x*, -p’(x) A > w; the marginal social benefit of more precaution is
bigger than its cost, and so more precaution is better.
When x > x*, -p’(x) A < w; the marginal social benefit of more precaution is less
than its cost, and so less precaution would be optimal.
Note that we haven’t yet specified who is taking this precaution. In many cases, it is
mainly the injurer who can take precautions to reduce the likelihood of accidents. In
some cases, the victim can also take precautions. The book gives a bunch of examples
where both the injurer and the victim has some influence on the likelihood of an accident:
Accident Injurer’s precaution Victim’s precaution
Faulty electrical wiring Manufacture wiring more Fireproof the house
causes house fire carefully
Moving car hits parked car Drive more safely Park car in safer location
Car hits pedestrian Drive more safety Walk more safely
Software fails Better design of software Back up data at risk*
Exploding coke bottles Improve quality control Handle bottles carefully
Medicine causes side Improve warning label Study warning
effects level/interactions carefully
(* This one doesn’t really fit their setup, since it really lowers the cost of each accident,
not the probability of an accident occurring)
They then use this general setup to look at different liability rules.
The simplest two rules to analyze are the cases of strict liability and no liability.
Case 1: No Liability
Suppose first that it’s the victim who can take actions to prevent the accident – say, a
bicyclist who could wear reflective clothing and use a light when riding at night
Let x be the level of precaution taken by the victim, and p(x) the probability of an
The victim incurs the full cost of the accidents, A, if there is no liability; so he chooses a
precaution level to minimize
w x + p(x) A
which leads him to a level of precaution xv* which solves
w = - p’(x) A
(since the left-hand side is his marginal cost of more precaution and the right-hand side is
his marginal private benefit)
But this is exactly the efficient level of precaution from before. So a rule of no liability
causes the victim to bear the entire cost of the accident, and therefore leads the victim to
select the socially optimal level of victim precaution.
But on the other hand, suppose it is the injurer whose actions determine how likely
accidents are – for example, x can be a measure of how cautiously you drive. Under a
system of no liability, the injurer doesn’t bear any costs from accidents. So he simply
tries to minimize wx. Which means when there is no liability, the injurer has no
incentive to take costly precautions, so he sets x = 0, which is lower than the social
So a rule of no liability leads the injurer to select an inefficiently low level of
We can put these two results into a table:
Legal Rule Victim’s Precaution Injurer’s Precaution
No liability Efficient Zero
Next, we consider Case 2: Strict Liability
Suppose we are in a world with strict liability, where in the event of an accident, the
injurer will have to pay damages D to the victim. And suppose that compensation is
perfect – D = A, so the damages truly compensate the victim for the harm.
Go back to the example of reflective clothing. Now the bicyclist knows that if he’s in an
accident, he’ll incur harm of A and receive damages of D, so now the costs he faces are
w x + p(x) (A – D) = wx
which he minimizes by setting x = 0 – under strict liability, the victim doesn’t bear the
cost of accidents, so he takes no precautions.
Is this efficient? No. The total social cost of accidents is
wx + p(x) A
the transfer of money from injurer to victim is just a transfer. So it would be efficient to
set a higher level of precaution.
So a strict liability rule leads the victim to select an inefficiently low level of
But on the other hand, if it’s the injurer who affects the probability of accidents, strict
liability ends up working pretty well. The injurer knows if there’s an accident, he’ll have
to pay damages of D, so he minimizes
wx + p(x) D
and with D = A, this is
wx + p(x) A
So he’s facing the same problem as minimizing the total social cost of accidents – so he
chooses x efficiently!
So a strict liability rule leads the injurer to take efficient precaution.
And we can update our table:
Legal Rule Victim’s Precaution Injurer’s Precaution
No liability Efficient Zero
Strict liability Zero Efficient
That’s what happens when there’s only one party that can take precautions to prevent an
accident; which is referred to as unilateral precaution. Our results so far suggest that
when only the victim is able to take precautions to prevent the accident, a rule of no
liability is preferable; and when only the injurer is able to take precautions, a rule of strict
liability is preferable.
(This is exactly analogous to the problem we saw in contract law, of allocating a risk to
the lowest-cost avoider. When one party – in this case, victim or injurer – can take
precautions and the other cannot, that first party is the low-cost avoider. Making them
bear the cost of the accident leads to efficient unilateral precaution.)
However, there are lots of cases of bilateral precaution – situations where both parties
have the possibility of precaution. In that case, the probability of an accident is a
function of both parties’ precaution levels, and the social cost of accidents is
SC = wv xv + wi xi + p(xv, xi) A
And in that case, things get more complicated. (You’ll see an example of this on the
Recall with contracts, we saw that there was breach remedy that set efficient incentives
for both the promisor and the promisee. Cooter and Ulen refer to it as the paradox of
compensation. Here, we find the same thing: it’s hard to find a rule that sets the correct
incentive for precaution by both the injurer and the victim.
However, in tort law, there is a trick. We already defined a negligence rule, under which
the injurer is liable for harm only if he did not take certain precautions. That is, in tort
law, we can make damages depend not only on whether an accident occurred, but also on
the level of precaution that the parties took ahead of time. This may allow us to get both
And this is where we’ll start Thursday.