Over the Last Three Lectures

Econ 522 – Lecture 18 (Nov 132008)

Over the last three lectures…

we introduced the notion of torts, and several possible rules for when an injurer is held liable for the harm he caused
we introduced the notion of precaution – costly actions the injurer (or the victim) could take to reduce the likelihood of an accident – and examined the incentives for precaution created by various liability rules
we introduced the notion of activity level (which can also be thought of as unobservable precaution), and the incentives created by the various liability rules
negligence rules must be accompanied by a legal standard for how much care or precaution is required to avoid negligence; we discussed the rule put forward by Judge Learned Hand, which held that precaution is required as long as it is cost-justified, that is, efficient
and we looked at the effects that errors, both systematic and random, would have on the incentives each liability rule gives
finally, we re-examined the questions of precaution and activity levels in a market setting, where expected liability losses would factor into the price of a product, and considered the outcomes when customers had differing abilities to perceive the riskiness of their choices

Two points I didn’t get to last time:

Cooter and Ulen talk a bit about the different costs of administering different liability systems
Obviously, it’s simpler to prove just harm and causation than to prove harm, causation, and negligence
So once a case goes to court, we expect the administrative costs to be higher under a negligence rule than under strict liability. (More time spent, more witnesses, etc.)
On the other hand, under a negligence rule, many victims will know they have no case, and therefore not bring a lawsuit at all; under a strict liability rule, every accident victim is entitled to damages, so there will be more lawsuits
So strict liability will lead to more cases, but easier cases.
(Obviously, a rule of no liability leads to lower administrative costs than either, since there’s no work to be done.)

C and U also point out the tradeoff between rules (such as the legal standard of care) which are tailored to individual situations, versus broad, simple rules that apply to many situations
As we’d expect, broad, simple rules are cheaper to create and enforce, but will not create perfect incentives in every situation; more specific, detailed, “tailored” rules will be more costly to create and enforce, but will create more efficient incentives.
(In addition, there is the question of who bears some of these administrative costs. In some countries, victims who win at trial also have their legal expenses paid by the injurer; in the U.S., this tends not to be the case.)

Given all the time that we’ve just spent developing a formal economic model and examining its implications, I think it’s fair to step back a bit and ask the question: does the model work?

That is, is there any evidence from the real world that a choice of liability rule affects peoples’ behavior in the way the model predicts?
The usual assumption we make in economics is that if you make something more costly, people will do less of it.
But when people get in their cars, do they really think about the amount they will have to pay in the event of an accident when deciding how fast and how far to drive?
Do people really think about liability rules when deciding whether to get in a bar fight?

This is exactly the question (not the bar fight question, the more general question) addressed in the paper by Gary Schwartz, “Reality in the Economic Analysis of Tort Law: Does Tort Law Really Deter?” He reviews a wide range of empirical studies in different areas of tort law, and comes to the following, not that startling conclusion:

Tort law does affect peoples’ behavior in the direction the economic model predicts, but not as much as a literal reading of the model would suggest

He points out that most of the academic work prior to that point was either implicitly assuming that people behaved exactly as in the model; or pointing out various critiques of the model, and reasons why liability rules would not impact behavior at all; but that the truth lay somewhere in between.

(One of the obvious ways in which the model is “wrong”: the model suggests that, under a negligence rule, injurers will always take the mandated level of care – that is, there will never be any negligence! And yet there are lots of studies showing that negligence is rampant – in auto accidents, in medical malpractice, and in other areas. Nonetheless, studies in a variety of industries show that a greater degree of liability does lead to greater overall levels of precaution.)

Schwartz has a funny line toward the end of the paper. He argues that since people do not respond as precisely to incentives as the model predicts, we shouldn’t spend so much time trying to “fine-tune” the law to achieve perfection:

“Much of the modern economic analysis, then, is a worthwhile endeavor because it provides a stimulating intellectual exercise rather than because it reveals the impact of liability rules on the conduct of real-world actors. Consider, then, those public-policy analysts who, for whatever reason, do not secure enjoyment from a sophisticated economic proof – who care about the economic analysis only because it might show how tort liability rules can actually improve levels of safety in society. These analysts would be largely warranted in ignoring those portions of the law-and-economics literature that aim at fine-tuning.”

He also points out, since “fine-tuning” may not work, that simple rules start to make more sense. He looks at the example of worker’s compensation in the United States. Worker’s compensation holds the employer liable (whether or not he was negligent) for the economic costs of on-the-job accidents, while leaving the victim bearing all non-economic costs such as pain and suffering. Schwartz argues:

“Analyzed in incentive terms, this regime of “shared strict liability” takes for granted that there are many steps that employers can take, and also many things that employees can do, to reduce the work accident rate. Yet workers’ compensation disavows its ability to manipulate liability rules so as to achieve in each case the precisely efficient result in terms of primary behavior; it accepts as adequate the notion that if the law imposes a significant portion of the accident loss on each set of parties, these parties will have reasonably strong incentives to take many of the steps that might be successful in reducing accident risks.”

Many of the objections Schwartz points out in his paper – reasons that people may not respond to liability laws in the way the “standard model” predicts – can be seen as violations of what Cooter and Ulen refer to as the “core assumptions” of the model. Specifically, the model as we’ve explained it so far assumes:

Decision-makers are rational
There are no regulations in place beyond the liability rule
There is no insurance
Injurers pay damages in full (for example, they can’t run out of money and go bankrupt)
Litigation costs are zero

We can relax each of these assumptions in turn, and see what effect this should have.

Assumption 1. Rationality

Cooter and Ulen give two examples of ways in which the rationality assumption may be violated.

The first is on the basis of a growing literature in behavioral economics that says that many people systematically misperceive the value of probabilistic events. That is, a number of experiments have shown that when people evaluate probabilistic events, they make choices that are not compatible with the usual expected-utility framework.

One classic example of this comes from a classic paper by Daniel Kahneman and Amos Tversky, called “Prospect Theory: An Analysis of Decision under Risk.”
They found that given a choice between a 45% chance at $6,000 and a 90% chance at $3,000, most (86%) of their sample chose the latter; but given a choice between a 0.1% chance of $6,000 and a 0.2% chance of $3,000, most (73%) chose the former.
Under the standard expected-utility setup, either u(6000) is twice as high as u(3000) or it’s not; here, people were clearly doing something other than maximizing expected utility; and it seems to do not with how they evaluate the value of money, but how they evaluate probability.
More recent work by the same authors – cited in the textbook – argues that people tend to overestimate the likelihood of events with well-publicized, catastrophic results, like accidents at nuclear power plants – the resulting panic makes the few that occur stick in peoples’ minds, so they imagine them to be more frequent than they actually are.
(There’s also a chapter in Freakonomics about how people fixate on the “wrong” risks – that is, people freak out about very unlikely events (leading, say, to lots of regulations about flame-retardant childrens’ pajamas), while ignoring much more likely risks that seem more commonplace, such as swimming pool accidents.)
All these examples build the case that maybe people don’t make perfectly rational expected-gain tradeoffs the way we expect them to. Given that, we wouldn’t expect people to correctly trade off the expected incremental cost of probabilistic accidents, – p(x)’ A, against the certain cost of increased precaution, w.

Cooter and Ulen consider the implications of this in a setting of bilateral precaution, accidents with power tools. Power tools can be designed to be safer, and they can be used more cautiously. However, suppose consumers underestimate the likelihood of a power tool accident. (People assume that any product on the market must be very safe, so they exercise no caution whatsoever.)

A negligence rule with a defense of contributory negligence is common for product liability. This would lead chainsaw companies to design chainsaws that are perfectly safe (or at least, efficiently safe) as long as they are not used negligently. Under perfect rationality, this would lead consumers to take efficient care in using them, and all would be well. But if irrational consumers underestimate chainsaw risk, this would lead to too many accidents.

On the other hand, a strict liability rule (along with the manufacturer knowing that its consumers will be negligent) will lead chainsaw manufacturers to design even safer chainsaws, which are less likely to cause accidents even when used recklessly; in a world with irrational consumers, this is a good thing.

(Of course, this is almost completely analogous to the ridiculous-sounding example we used a few days ago, of forcing car manufacturers to design seatbelts that buckle themselves. Oh, well.)

The second type of irrationality Cooter and Ulen consider is unintended lapses, that is, accidental negligence. Rather poetically, they point out that “many accidents results from tangled feet, quavering hands, distracted eyes, slips of the tongue, wandering minds, weak wills, emotional outbursts, misjudged distances, or miscalculated consequences,” all of which they summarize as “lapses”. That is, people try to exercise due care, but once in a while, they fail.

The example they give is from a world without cruise control. The speed limit on a road is 70, and so driving faster than that constitutes negligence. A driver intends to drive 65, but from time to time his mind wanders and he looks down to find himself driving 73. If one of these times, he’s in an accident, he’s liable.

(On the other hand, a driver who sets out to drive 75, but mistakenly finds himself doing 67 when he hits someone, is not liable, obviously.)

Cooter and Ulen’s discussion here is weirdly moralistic – they seem to take the position, both that speeding is somehow immoral, and that “not wanting to speed” is somehow more important than actually not speeding. They point out that a driver who realizes he may occasionally lapse will rationally target a level of precaution higher than the legal standard, to lessen the frequency of these lapses taking him below the legal standard x~. (This is exactly the same effect as the overprecaution we expect as a result of random uncertainty about the exact legal standard.)

As they point out, however, a liability rule that required intentional negligence, rather than accidental negligence, would be almost impossible to enforce – proving intent is even harder than proving negligence, which was already harder than proving harm and causation – and would likely lead to most injurers avoiding liability altogether, leading to no incentives for precaution. They give the rather creepy notion that GPS in cars will eventually allow us to distinguish the habitual speeder from the “accidental” speeder, and then move on.

We’ll go out of order and consider bankruptcy next.

We’ve said all along that strict liability causes an injurer to internalize the expected harm done by accidents, leading to efficient precaution.

However, consider a situation in which a firm’s liability is more than its net worth, that is, more than the total value of the company. The firm has no way to come up with the damages owed; so it declares bankruptcy. Thus, bankruptcy places a limit on the damages that can be paid.

But if the damages that will actually be paid are less than the actual harm, then the firm is not internalizing the full cost of accidents; as a result, the firm will take inefficiently little precaution.

The book considers the example of a hazardous waste disposal company. If the company intends to stay in business forever, it will be very careful in transporting hazardous waste, in order to avoid accidents/liability. On the other hand, it might take a different strategy: dump recklessly, earn short-term profits, pay them out to shareholders, remain undercapitalized, and expect to go bankrupt once an accident occurs and someone sues.

Aside from limiting liability, bankruptcy also has a social cost, since the intangible assets of the company – its reputation, its employees’ firm-specific knowledge – are destroyed.

An injurer whose liability is limited by bankruptcy is referred to as being “judgment-proof”, that is, they are immune to judgments beyond a certain level. (We’ll return to this concept later.) There is no perfect solution to the distortions that this causes, but there are some ways to reduce them, one of which is regulation, which is the third extension we consider.

The next extension they consider to the “standard” model is of settings which are governed by both a liability rule and safety regulations. For example, fire regulations may require a store to have a working fire extinguisher, and fines may be issued to stores that, upon inspection, fail to meet regulations; but if a fire in the store injures a customer, the store may still be liable.

Rather than give a general model of how liability and regulation interact, Cooter and Ulen discuss a few determinants of the plusses and minuses of each.

Administrators who regulate only a single industry can acquire the detailed technical knowledge needed to set safety standards efficiently, while a court might have trouble acquiring this level of knowledge on a wide range of industries. In these settings, courts can adopt the legal standard set by safety regulators; with both standards set the same, “potential injurers will conform to that standard in order to avoid both ex ante fines and ex post liability.”

However, they give arguments why a court may feel industry regulators might set safety standards either too high or too low.

If regulators are susceptible to political pressure from powerful firms, safety standards might be set too low to help them avoid liability
On the other hand, corrupt regulators might set standards too high, to ensure that bribing them would be cheaper for businesses than complying with the rules
Standards could also be set high to protect incumbent firms from new competition

Thus, courts may choose to deviate from regulated safety levels in setting the legal standard for care. When safety regulations and liability law impose different standards, firms will tend to follow the higher standard, to avoid both liability and fines.

As we just saw, when liability exceeds an injurer’s total wealth, the injurer goes bankrupt, but cannot be held liable for the full amount of the harm
In settings where damages would bankrupt a firm, expected damage payments would be lower than p(x)A, since damages would be limited to an amount less than A.
This would lead to insufficient precaution under a strict liability rule
However, regulations which hold a firm to the efficient level of care avoid this problem, since large fines could be assessed to firms in violation of safety standards before an accident occurs
Thus, in industries where severe accidents are likely to bankrupt firms, safety regulation may work better than liability in encouraging precaution.

Regulation may also be better than liability when accidents impose only a small harm on a large group of people: since going to trial is costly, it may not be worth it for victims suffering only a small harm, and firms might escape liability because nobody finds it worthwhile to sue.
(Class action lawsuits also get around this problem – we’ll get to that later.)
In these cases, liability alone might also lead to insufficient precaution, while regulation can enforce the efficient level of care.

Going back to the fundamental assumptions we’ve been making in tort law… If I drive more carefully, I cause fewer accidents. If I face greater liability when I cause accidents, I choose to drive more carefully.