The body of scholarship using the various tools of economics to assess alternative standards of liability for accidents is enormous.   The purpose of the following essay is neither to provide a comprehensive map of all of the articles and books on the subject nor to resolve the raging debate concerning the relative merits of negligence and strict liability.  The goal, rather, is to illustrate how economists tend to think about the issue of tort liability and to outline a few of the major arguments they have developed. 

Driving a car, operating a nuclear power plant, or using a hair dryer are all activities that produce risk. Almost everything we do produces a risk of some sort accident or harm, and we quite naturally take precautions against many of those harms. Some precautions, like those taken in nuclear power plants, are expensive; others, like not leaving a hairdryer on when we are finished using it, are relatively inexpensive. In deciding what level of precautions to take, we sometimes weigh the cost of the precaution against the benefit (in terms of reduced accident cost) that it will provide. Suppose that I own a house with a fireplace, and the probability (P) of a spark from my chimney setting the roof my house on fire in the foreseeable future is .01. (In other words, there is a one percent chance that a spark will start a fire). If my roof is set on fire, I would suffer a loss (L) of $10,000 in damage. The expected cost of the fire would be its cost if it occurs ($10,000) times the probability that it will occur (.01), that is $100. Assume that I am risk neutral, and that a spark-catching device on my chimney will cost $80 to install. 

Clearly, installation of the device would be both wise from my standpoint and socially efficient (its benefits outweigh its costs). Suppose, instead, that the device cost $200. In that case, installation would be both unwise from my standpoint (because its cost exceeds its expected benefits) and socially inefficient. (I would be better off -- and society would be better off -- if instead of purchasing the device, I bought fire insurance.)  Because, in both of these cases, the self-interest of the affected person and the collective interest of society at large coincide, it may not be necessary for the law to intervene. At least if I am fully informed and attentive (an assumption we will return to later), I will (without any guidance from the state) voluntarily take the socially optimal level of precautions. 

What if (because of prevailing winds in my community) any sparks issuing from my chimney would land, not on my roof, but on my neighbor's? Suppose that the loss to the neighbor in such a case would be $10,000; the probability of the calamity occurring is .01; and the cost (to me) of the spark-catching device would be $80. Here, self-interest and the interests of society diverge. Plainly, it would be socially efficient for me to install the device. But if I am not obliged to pay for the injury sustained by my neighbor, I will not take the efficient precaution. In such situations, legal intervention may be necessary. It is cases of this sort that prompt many economists to think of tort law primarily as a system for inducing people to behave in socially efficient ways -- specifically, in ways the minimize the sum of the costs associated with injuries and precautions taken to avoid injuries.

What standard of liability is most likely to achieve that effect? There are many possible standards lawmakers might choose from, but the two most prominent are negligence rules, under which defendants pay for harms caused by their unreasonable activity, and strict liability rules, under which defendants pay for all of the harms caused by their activities, whether or not that activity was reasonable.Which is best from a deterrent standpoint? Let's consider them in turn. 

I. Negligence

The most common economic definition of negligence can be seen in operation in Judge Learned Hand's famous opinion, U.S. v. Carroll Towing Co., 159 F.2d 169, 174 (2nd Circuit 1947).[3] The event that gave rise to the case were as follows: Several barges owned by the Connors Marine Co. were tied together off a busy Manhattan Pier. The defendant's tug boat, "Carroll," removed one of the lines connecting the barges to the pier. When the remaining lines broke, the barges were washed down-river and sank. No one was aboard the barges when they broke away, and the evidence indicated that had Connors' barge operator (the "bargee") been on board, the barges could have been saved. The question for the Court was whether Connors was liable for failing to have its "bargee" remain aboard. Judge Learned Hand wrote that Connor's liability for the lost barge depended on: 
 

(1) The probability that she will break away; (2) the gravity of the resulting injury, if she does; (3) the burden of adequate precautions. Possibly it serves to bring this notion into relief to state it in algebraic terms: if the probability be called P; the injury, L; and the burden, B; liability depends upon whether B is less than L multiplied by P: i.e., whether B less than PL.

A defendant is negligent if and only if B < PL.[4] This formulation of the negligence rule has come to be known as the Hand Formula. Judge Learned Hand went on to conclude that, taking into account the surrounding circumstances (there were strong winds, the harbor was busy, and the ship's cargo valuable), the risk (PL) outweighed the burden of prevention (B). Consequently, he held Connors liable.[5] 

Graphically, the Hand Formula looks like this: 

The X-axis depicts the amount of care taken by the defendant. The Y-axis represents the cost in dollars -- of taking precautions (line B) or of the expected loss resulting from failing to take precautions (line PL). The upward slope of line B indicates that the marginal cost of prevention increases as more care is taken. The downward slope of line PL indicates that the marginal expected cost of an accident declines as more care is taken.[6] The point at which the combined cost of precautions and accident are minimized is c*, the optimal level of care. 

In determining what level of care the courts should require, two points should be kept in mind. First, courts should not merely examine the cost of the accident that occurs without taking into account the probability of such a loss occurring. Returning to our initial example, if my neighbor's house catches on fire at the cost of $10,000, then my failure to take a precaution costing $9,999 will appear negligent if the court looks only at the loss that actually occurred.[7] Yet the expected cost of the accident was only $100. Unless the court considers both the probability of the loss and its magnitude, it will prescribe a level of care that above the optimal level, producing an inefficient result.

Second, the court must consider the marginal cost and benefits of additional precautions, not the total costs and benefits of all precautions. Suppose that spark-catching devices were prohibitively expensive, but an alternative set of precautions were available to me: I could paint my roof (or my neighbor's roof) with fireproof paint -- thereby reducing the expected cost of a spark accident from $100 to $10 -- for a cost of only $30. A second coat of fireproof paint could reduce the expected cost of a spark accident to zero for another $30. If a court looked at the total cost and benefits of two coats of fireproof paint, the total precautions would appear efficient because the total cost ($60 of paint and labor) is less than the total benefit ($100 in reduced risk). However, if we separate the two decisions -- the choice to apply a first coat and the choice to apply a second cost -- it becomes apparent that the latter is inefficient; it provides only a $10 marginal benefit for a $30 marginal cost. The general point: courts concerned with economic efficiency should look at the marginal cost of each level of precaution. 

[For two contemporary applications of the Hand Formula, see Judge Posner's decisions in U. S. Fidelity & Guaranty Co. v. Plovidba, 683 F.2d 1022 (7^th Cir 1982) (involving the reasonableness of boat owner's unloading procedures resulting in the accidental death of a longshoreman) and McCarty v. Pheasant Run, Inc., 826 F.2d 1554 (7^th Cir 1987) (involving the reasonableness of hotel security precautions where woman suffered beating and attempted rape)]. 

What action will the parties take if the court uses a negligence rule?[9] Assuming that the level of care required by the courts is equal to the optimal level of care, the defendant will naturally take the optimal level of care to avoid liability. What about the plaintiff? Will he become careless, knowing that the defendant must either take optimal precautions or pay for any injuries caused by his failure to do so? Probably not. Not all losses can be readily compensated with money. For example, no amount of money can offset the utility losses a plaintiff suffers from his own death. The loss of a loved one or the pain and suffering from bodily injury are other examples of non-compensable losses. In addition, plaintiffs are often risk averse with respect to relatively large losses and tend to prefer uniform levels of utility over time. A risk averse individual would prefer a constant utility of 10 over time to a utility of -5 (due to injury) in one period and a utility of 25 (including compensation for the previous injury) at a later time. For these reasons, in a negligence regime, plaintiffs, as well as defendants, are likely to continue to take care -- although whether the level of care they take is optimal has yet to be determined. 

So far, the negligence standard is looking pretty good from an economic perspective. On further reflection, however, negligence rules prove vulnerable to some serious economic criticisms. See Guido Calabresi & Jon Hirschoff's Toward a Test for Strict Liability in Torts, 81 Yale L.J. 1055 (1972). Four are especially strong: 

1. The administrative costs of a negligence rule are higher than those of a strict liability rule. To apply a negligence rule, a court must first determine the level of care that would have been optimal under all of the circumstances, and then whether the parties met that level of care. To apply a strict liability rule, all a court must determine is whether or not the defendant caused the plaintiff's injuries. The latter is much simpler -- and thus cheaper -- for the parties as well as the court.
2. Another argument against the use of the negligence rule is that, because of the increased expense of bringing a suit, more plaintiffs will choose not to file suit. This could result in underdeterrence because the defendant will not bear the full cost of his risky activity. Suppose, for example, that a defendant's factory pollutes a stream and causes a loss of $10 to be suffered by each of 100 residents who live down stream. The pollution can be prevented by a filter that costs $750. The total cost of the defendant's activity is $1000. Plainly, installation of the filter would be efficient. Because PL exceeds B, a court applying a negligence standard would find the defendant liable for failure to install the filter. To avoid that liability, we would expect the defendant to engage in the socially optimal behavior -- i.e., install the filter. However, what if the defendant knows that, if he continues to spill, only sixty percent of the residents can afford to bring suit? Under those circumstances, the defendant's expected liability will be only $600. Paying that judgment would be cheaper than installing the filter. Consequently, he is likely to fail to take the optimal level of precautions.
3. Courts may err in determining the level of proscribed care. If courts set the level of proscribed care too high (and force defendants to observe it, instead of allowing them to "punch and pay"), efficient activity will be deterred. If courts set the proscribed level of care too low, inefficient injuries will occur. For a detailed discussion of the problems of over- and underdeterrence, see Craswell, Deterrence and Uncertain Legal Standards 2 J. Law, Econ & Organization 279 (1986). The parties may also err in predicting the level of proscribed care that will be required by the courts. See Mark F. Grady's A New Positive Economic Theory of Negligence, 92 Yale L.J. 799 (1983). Such errors by either the court or the parties will result in an inefficient outcome.
4. Finally, a pure negligence rule may not address cases where the plaintiff is the least cost avoider of the accident. For example, if the plaintiff can avoid a loss of $10 by taking a precaution which costs only $2, and the defendant can prevent the same loss by taking a precaution that costs $9, we would prefer the plaintiff to take the $2 precaution rather than have the defendant take the relatively inefficient $9 precaution. This last problem can, however, be addressed through the use of a properly formulated contributory negligence doctrine. Question: will the contributory negligence doctrine currently in force in most American jurisdictions have this effect?

II. Strict Liability

Under a strict liability rule, the defendant pays for the injury his conduct causes the plaintiff regardless of whether the defendant was negligent. That is, the defendant pays for both negligent and non-negligent injuries. Famous cases employing strict liability rules include Rylands v. Fletcher, L.R. 3 H.L. 330 (1868), Ploof v. Putnam, 81 Vt. 471 (1908), Vincent v. Lake Erie Transp. Co., 109 Minn. 456 (1910) and Bolton v. Stone, 1 K.B. 201 (1950)[provide links to all]. 

As with a negligence rule, a defendant who is held strictly liable will take the optimal level of care to minimize his expenses. If a precaution costs less than the expected cost of the injury it will prevent (i.e., the cost of the injury times the probability it will occur), the defendant will take the precaution rather than pay for the injury. The defendant will not take an inefficient precaution, because if the cost of the precaution is more than the benefit it would provide in terms of injury prevention, it is cheaper for the defendant to pay for the resulting injury than to take the precaution. Thus under either a strict liability or a negligence rule, the defendant will take the optimal level of care. 

One advantage of a strict liability rule -- as mentioned above -- is that it reduces administrative costs. The courts need not determine what level of care should be required of defendants nor whether defendants have met the prescribed level of care. A strict liability rule thus can produce an efficient result in a situation in which a court would be unable to determine what precautions should be taken by the parties. A strict liability rule also reduces litigation costs to parties and may encourage settlement of lawsuits by reducing uncertainty about probable judgements. The cost savings of each trial under a strict liability rule may be offset, however, by an increased number of suits brought. Thus it is not necessarily the case that a strict liability rule will always be cheaper to administer than a negligence rule. 

A third advantage of a strict liability rule is what has been described as the "insurance function" of strict liability. Holding a company strictly liable for the injuries created by its products permits the company to spread risk to all the consumers who purchase the product. If a product will injure one consumer in a thousand, a strict liability rule will impose upon every consumer of the product 1/1000 of the resultant loss (in the form of slightly higher prices), rather throwing the burden entirely upon the person who is non-negligently injured. An ancillary benefit of such a regime is that the prices of products will fully reflect all of the costs associated with the harms the products cause, rather than only those costs due to the manufacturer's negligence. The resultant pattern of product prices will encourage consumers to select safer products and avoid dangerous ones. 

There are several potential arguments against the use of strict liability in torts: 
 

1. Strict liability has also been criticized on moral grounds by Richard Epstein in A Theory of Strict Liability, 2 J. Leg. Stud. 151 (1972) and Ronald Dworkin in Is Wealth Value?, 9 J. Leg. Stud. 191 (1980).
2. Some economics have criticized strict-liability regimes on the ground that they will overdeter risky behavior. The risk of overdeterrence may be particularly strong in areas of new or developing technology. If companies bear the risk of all foreseeable and unforeseeable harms, they may be deterred from pursuing innovations whose risks are relatively unknown. This, in turn, may result in under-investment in new technology. For a more discussion of overdeterrence, see Richard Posner's Strict Liability: A Comment, 2 J. Legal Stud. 205 (1973).
3. Strict liability rules may also provide incentives for strategic behavior by plaintiffs. Consider the case of a farmer who knows a railroad will be held strictly liable for all livestock its trains run over. The farmer may let more of his livestock wander near the tracks rather than incur the expense of keeping them penned. In situations where both the plaintiff and the defendant should take care, a pure strict liability rule is inefficient. The danger of encouraging strategic behavior may be mitigated by risk aversion on the part of the plaintiff, especially if the plaintiff may suffer physical injury. In addition, a comparative or contributory negligence rule, used in conjunction with a strict-liability standard, can largely remove the incentive for strategic behavior -- although it will also increase the costs of administering the regime.
   Finally, strict liability rules may have regressive effects in practice. When the victim of a non-negligent injury receives compensation under a strict liability rule, the cost of the liability award typically is borne ultimately by two groups: those who must pay a higher price for the product and those who can no longer afford the product and must chose substitute goods. The increase in the price of a product due to strict liability is borne equally by all those who use the product regardless of their income. A fixed increase in the price of a product affects people with low incomes disproportionately more than it does people with higher incomes. For this reason the "price effect" of strict liability is sometimes considered a regressive tax.

III. The Effects of Liability Rules on Activity Levels

The probability that an injury of a particular sort will occur is affected, not just by the degrees of care exercised by defendants and plaintiffs, but also by the frequency with which they engage in their respective behaviors. For example, the likelihood that a hiker will be shot by a hunter is affected, not just by hunters' care in shooting and hikers' willingness to wear orange vests, but also by the frequency with which hunters hunt and hikers hike during hunting season. In principle, there is no reason why courts should not consider activity levels of plaintiffs and defendants when assessing the reasonableness of their conduct -- i.e., when applying a negligence or contributory negligence standard. In practice, however, courts rarely do so -- largely because they are unable to determine the optimal level of any activity. 

Even though, in general, the courts do not examine the activity levels of parties, the standard of liability they select will indirectly affect those activity levels. Suppose, to illustrate the point, that cars and 18-wheel trucks sometimes collide at intersections (perhaps because of bad weather or mechanical defects) even though the drivers of both vehicles take optimal precautions and that in such collisions only the cars (and their drivers) are injured. If a negligence standard is used to determine liability arising out of such collisions, both car and truck drivers will of course have an incentive to take optimal care -- to observe speed limits, watch for other vehicles, stop at stop signs, etc. In such a system, risk averse car drivers will also have an incentive to reduce their activity levels -- i.e., to drive less in hopes of avoiding uncompensated (because nonnegligent) injuries. Under a negligence rule, truck drivers, however, will have an incentive to maintain high activity levels even when, from a social-welfare standpoint, those levels are inefficient. Knowing that they can escape liability altogether by taking the prescribed level of care, the truck drivers will continue driving as long as the marginal benefits of additional trips exceeds their marginal costs -- excluding the costs associated with the resultant injuries.[10] 

A strict liability rule, by contrast, will discourage excessive activity levels by truck drivers. Forced to pay for all injuries they "cause," they will decline to take trips whenever the resultant savings (in terms of decreased liability) outweighs the potential profit. However, risk averse truck drivers may reduce their activity levels more than is socially desirable. Under a strict liability rule, on the other hand, car drivers may drive more than optimal amounts. Knowing that the truck drivers will have to pay for all injuries, they may undertake socially wasteful trips. However, this effect is likely to be mitigated by car drivers' desire to avoid pain and other non-compensable losses. 

The care and activity level effects of various rules are summarized on the following table: 

Is There An  Optimal  Level of:	No Liability	Strict Liability	Negligence	Strict Liability  with Contributory  Negligence
Care by  Truck Driver?	.
Truck Activity?	.		.
Care by Car Driver?		.		.
Car Activity?		.		.

Note that a "no liability rule" operates as a strict liability rule pointed against the plaintiff rather than the defendant.

The general lesson on this comparison is that -- from the standpoint of activity-level effects -- whether a negligence standard or a strict-liability standard would be better depends on (a) the relative risk aversion of the parties and (b) who is most likely to be the best activity-level modulator. If defendants can best adjust the level of their activities, we might wish to use a strict liability rule. If plaintiffs are the best activity-level modulators, we would tend to prefer a negligence rule. For further discussion of the effects of legal rules on care and activity levels in unilateral and bilateral accidents, see Steven Shavall's Strict Liability versus Negligence 9 J. Legal Stud. 1, 2-3 (1980).