HARNESSING THE AMERICAN MARKETPLACE / 57 Products Liability Reform Jules L. Coleman merican manufacturers continue to grapple with The impact of the current system for compensating consumers for harm done by manufactured goods extends far beyond such celebrated cases as those facing cigarette manufacturers or asbestos suppliers. In 1985-1986, nearly half (47 percent) of all product manufacturers in the U.S. removed product lines form the market place, 25 percent discontinued product research, and 39 percent decided against introducing new products, all as the result of increased exposure to liability. These are striking figures. More importantly, in removing products already on the market, manufacturers are responding to continuing changes in the rules governing liability for their products - changes they obviously could not predict in advance. For had the manufacturers been able to predict that their products would be deemed defective, they would either have altered the design earlier or simply not produced the product in the first place. Such radical changes in the availability of product lines cannot be good thing for either manuf~tcturers or consumers. On the other hand, it is certainly not desirable that dangerous products remain on the market. What is undesirable about the current products liability system is its uncertainty. What the law will deem to be d a n g e r o u s and h o w far it will c o n s i d e r the manufacturer's responsibility to extend are far less predictable than they should or could be. It can come as no real surprise that two words capture the American manufacturer's concerns regarding products liability- exposure and uncertainty. T h e s e two concerns are connected. The manufacturer's concern about exposure ranges over A the costly problem of products liability. both persons and time. To whom is a manufacturer liable? How far in time does responsibility extend? The Aviation Industry Example Problems of exposure and time are nicely illustrated by the current crises in the aviation industry. In 1979, the aviation industry produced more than 17,000 aircraft. In 1987, it shipped only 1,085 - a decrease of more than 90 percent. Specifically, Cessna, which produced 9,000 piston planes in 1979, produced none in 1987; and Beech dropped from 1,214 in 1979 to 195 in 1987. Teledyne Continental, a major engine manufacturer, saw its production drop from 11,000 to 400 in the same period. These figures do not represent a response to decreasing demand for the products generally, since the differential is almost entirely accounted for by increased foreign production- especially by West German companies. Rather, it reflects increased costs that make American products unattractive. Product liability costs now constitute the largest single cost factor in the construction of a new piston engine aircraft. Industry officials allege that if actual product liability costs were added as an average cost per plane to each aircraft produced, it would increase the cost by $100,000 per aircraft. The increase in product liability costs does not stem from increased accident rates. In fact, accident rates have steadily declined. Rather, it stems from the nature of product liability law itself. The problem of exposure is reflected in the fact that even though 85 percent of all accidents are due to pilot error, manufacturers are sued in well over 90 percent of cases involving fatalities or serious injury. There are, moreover, no caps on recovery, and no time limitations. 581 SOCIETY 9 NOVEMBER/DECEMBER 1989 The net result is that, while in 1977, the aviation industry paid $24 million dollars in claims, in 1985, it paid $210 million, though the number of suits remained relatively constant. How did product liability law come to wreak such havoc, and what can be done to reform it? What is Products Liability? At one time, product liability law was really a branch of the law of sales, risk was allocated by contract, and a manufacturer's liability was restricted by the doctrine o f privity to those individuals who were the immediate purchasers o f its products. The net effect of the doctrine ofprivity applied to products was to insulate the product manufacturer from liability to the ultimate users of its products. A manufacturer's potential liability would extend in time no longer than the period of ownership of the immediate purchaser, and often a good deal less. Such a rule restricting the scope of liability over persons and time had two significant benefits for product manufacturers. First, it reduced significantly the costs of doing business. Many of the harms causally connected to the manufacturing process were not held to be the manufacturer's responsibility. Instead they were extemalized, that is, imposed on third parties. As a consequence, manufacturers were not required to bear the full social costs of their activities. Second, the sales contract as a market mechanism of risk allocation enforceable at law provided manufacturers with a great deal o f certainty regarding at least one central element of the costs of production. With confidence and warrant, a manufacturer could determine in advance his exposure to liability and adjust his investments in safety accordingly. The current problem is whether the rule of liability for products can provide both the predictability for rational investment in safety and the incentives necessary for proper investment decisions. We have come a long way from a standard that allowed predictability but encouraged underinvestment to the present set of standards that magnifies uncertainty and, in doing so, undermines possibilities for rational investment in safety. The story of m o d e m products liability law, then, is a tale of confused and failed attempts to make this body of law at once rational and predictable. Modern Liability Law It was obvious that the doctrine o f privity as applied to products could not survive. Such a law did nothing to protect the ultimate users of products those very individuals for whom the products were intended. The landmark case of MacPherson v. Buick Motor Company is widely credited with weakening the so-called privity limitation, thereby removing product liability from the law of sales and putting it within the law of tort. Liability in tort is not restricted to those with whom one has a contractual or business relationship. One could think about the rules o f tort liability as those goveming relationships among "strangers," those with whom one is unconnected by contract. The idea is simple enough. If two parties have a contractual relationship, then if there are no laws prohibiting them from doing so, the risks each bears in case something fails to go as planned should be settled in advance. The law can be viewed as providing a set of default rules that can be appealed to in the event parties fail to agree, or which they might contract around if there is agreement on allocation of risk. These private arrangements about risks and costs can only be made when both parties are in a position of contracting at low cost, and legal rights and duties to one another are inexpensive to determine. Uncertainty will hamper bargaining or contracting even if other transaction costs are low; and not being in a position to contract at low cost makes contracting around rules too expensive to be rational. One might view the entire law o f torts as setting out the rights and responsibilities we have to one another as regards safety-related risks when our relationships are fundamentally noncontractual, when the costs of allocating risk by contract are too high. For these rules to provide us with meaningful incentives, they must be predictable. For them to provide us with optimal incentives, they must be rational. The rejection of the sales contract as a vehicle for allocating risk, therefore, rests on two assumptions. The first assumption is that the costs of contracting, of allocating risk by market forces, are too high; that, in effect, the market will allocate product risk inefficiently. If we left the allocation of risk to manufacturers and consumers, we would either get far too little or too much investment in safety. The usual assumption is that we will get too little. The second assumption is that substituting imposed tort standards is more likely to lead to predictable, rational investments in safety. Two questions naturally emerge. First, is the move from the contract to the tort solution warranted? Is there sufficient evidence that leaving risk alloca- HARNESSING THE AM ERICAN MARKETPLACE / 59 tion decisions to the market will be inefficient? Second, if courts were ultimately correct in abandoning contract for tort, has tort law provided the needed rationality and predictability? Let us take up these questions in reverse order. The initial question to address in this respect is whether in moving from the law of sales to the law of torts, the courts have provided a body of liability rules for products that is both predictable and rational. Or is it, instead, a body of law, as the statistics on product withdrawal suggest, that is unable to guide manufacturers consistently about their legal responsibilities and which encourages sweeping and deep confusion in product offerings? The greatest danger to an insurance market is radical uncertainty - precisely what the current products liability regime has given us In tort law, a distinction is made between strict and fault liability. When a rule of strict liability applies, a defendant will be held liable if the plaintiff can e s t a b l i s h a c a u s a l c o n n e c t i o n b e t w e e n the defendant's conduct and the victim's harm. Under negligence, the victim must establish not only a causal connection between conduct and harm, but a fault in the defendant's conduct as well. A person is at fault in causing harm only if he does so either intentionally, recklessly, or negligently. Accidents are not intentional, and most of the law of accidents, therefore, is devoted to negligent wrongdoing. Negligence is failure to exercise the care of a reasonable person of ordinary prudence. What is Prudent Behavior? E c o n o m i s t s have s u g g e s t e d that the care reasonable persons should take can be given a precise economic characterization. Assume, first of all, that there is some probability of harm from a product and that the costs to the potential victims are known as well as the costs of preventing harm. If the costs of preventing harm exceed the expected costs to the potential victim (the costs to the person harmed multiplied by the likelihood of its occurrence), it would not make economic sense to incur those prevention costs. An agent who failed to make inveslments in safety beyond the expected costs to the victim would not be negligent. Where the costs of prevention are less than the expected costs of the harm, failure to take precautions would constitute negligence. Failing to spend $200 to prevent an accident whose expected cost is $100 is reasonable and, therefore, not negligent. Failing to spend $50 to prevent the same accident would be unreasonable and negligent. In the event harm results from negligence so defined, the victim's losses would be the injurer's fault, and he would be justifiably held liable for them. If an injurer is not required to take precautions against unexpected harm, he is benefitted; that is, he saves the costs of prevention. We might think of the negligence test as a cost-benefit test. The injurer is negligent if the expected costs of the harm exceed his benefits from risking its occurrence. The test is social in that the determination of whether the benefits outweigh the expected costs is made by a public body, usually a jury, sometimes a judge, and occasionally an administrative or regulatory agency. Strict Liability and Negligence Under the standard of strict liability, an injurer will be held liable - the victim's losses shifted to him whether or not he has acted negligently, whether or not the costs of prevention exceed the expected costs of the harm. This fact might lead one to think that, in principle, a rule of strict liability will be harder on potential injurers than would be a negligence rule. In one sense, it is, but in another important sense, it is not. The strict liability rule, like its negligence counterpart, requires a cost-benefit calculation. Under strict liability that calculation is made exclusively by potential injurers, not by a public body, such as a court. If negligence is the standard, then its determination requires a collective decision. That decision, in tum, requires information including the injurer's schedule of benefits and the victim's damage schedule. In court, both parties, after the fact of the harm, have incentives to misrepresent, to understate or to exaggerate the costs and benefits. Under strict liability, an injurer already has special access to his benefit schedule, thereby increasing the likelihood of correct cost-benefit evaluations being made. Since, under strict liability, the injurer will be liable after the fact no matter what the cost-benefit calculation is, the injurer is led to make the calculation before he or she engages in a risky activity. Under both strict and fault liability, injurers are motivated to invest in safety only when doing so is 60 1 S O C I E T Y 9 NOVEMBER/DECEMBER 1989 cost-justified. In that sense, strict liability is no harsher on injurers than is negligence. The difference between the two, then, is that under fault liability, when neither injurer nor victim is at fault, the loss falls on the victim. In strict liability that loss becomes the injurer's responsibility. In that sense strict liability is harsher on injurers. With respect to safety investments, both strict and fault liability impose the proper incentive on manufacturers. They differ with respect to the distribution of risk for injuries that are in a suitable sense no one's fault. Moving Toward Strict Liability M o d e m product liability law is governed by a mixture of strict and negligence liability. The standard of liability in Mac Pherson was negligence, and it was not until the 1960s that strict liability came to provide a basis for recovery. Like most developments in the law, the change from negligence to strict liability was slow, and in a way tortured. The first significant m o v e m e n t from negligence to strict liability involved increasing reliance on the doctrine of res ipso loquitor. Under that procedural rule, the mere fact of an injury could be taken to be presumptive evidence of negligence. Rather than a consumer having to show a manufacturer's negligence, a manufacturer would be forced to establish his innocence. Even a consumer's own negligence in using a product could b e c o m e evidence o f the manufacturer's negligence. For surely a manufacturer could foresee that sometimes products would be misused. Failing to guard against that misuse would be further evidence of negligence. With no apparent way to defeat liability, manufacturers were held, in effect, strictly liable even though courts took pains to argue on negligence theories. The tuming point came in Escola v. Coca Cola Bottling Company. The opinion of the court was written on a negligence theory, but a brilliant concurrence by Justice Traynor advanced the view that the real and correct standard of liability for products is strict. Traynor argued the view that the goals of tort law are primarily economic ones: deterrence and insurance. Rules of liability should be designed to minimize risk and to spread it maximally over persons and time. For Traynor, pursuing both goals dictated a strict liability standard. His argument has captured the imagination of judges and scholars alike for its elegance and coherence. The argument is as follows. W h e n neither manufacturer nor consumer is negligent, who should bear the costs? As between victims and manufacturers, the latter are better suited to obtain insurance at a lower cost and to spread those costs maximally over persons and time. Manufacturers are better suited to invest in safety precautions or to determine which risks are worth taking. Liability ought strictly to be imposed upon manufacturers without regard to their negligence. Pursuing either end - deterrence or insurance - calls for a standard of strict liability. By the time the substantive rule of strict liability was openly imposed on manufacturers, the combination of forces yielded a role very much like absolute liability: liability for all harms without regard to the negligence o f the product user. In this regard, manufacturers were being asked to insure consumers for all safety-related risks. Tort law became, in effect, a vehicle by which the courts created a private insurance scheme. Manufacturers insure productusers - period. Design Defect Tests In another sense, m o d e m product liability law has maintained vestiges of negligence. In order to impose liability on a manufacturer some threshold requirement had to be met. This threshold was to be spelled out in terms o f a manufacturer's failure to measure up to some standard of conduct either internally or externally imposed. These standards were represented by "manufacturing" and "design" defect requirements. A manufacturer's product is said to suffer a manufacturing defect if it fails to perform as the manufacturer represents it. The negligence component is represented by design defect standards. And here lies the problem. There is no one design defect test applicable in all jurisdictions, nor are any of the tests easy to administer. Most importantly, the results under these standards are not at all predictable. In a recent article, Alan Schwartz reviews critically four separate design defect tests and finds them all wanting. These four tests are (1) the expectation test, (2) the Learned Hand test, (3) the risk/benefit test and (4) the regulatory/compliance test. Under the expectation test, a manufacturer can be liable if its product is less safe than it is reasonable for a consumer to expect it to be. A manufacturer is liable under the Learned Hand test if it is negligent in taking precautions. Liability under the risk/benefit test is a bit more complex. A firm's design is defective if, among all the feasible designs available, the finn fails to adopt the one which maximizes net benefits. In the HARNESSING THE AM ERICAN MARKETPLACE 161 regulatory/compliance scheme, the appropriate standards are set by a regulatory agency and the court's job is to enforce those standards. Failure to c o m p l y with the operative safety standard is negligent; compliance is merely evidence of nonnegligence, not constitutive of it. The rule of strict liability in conjunction with design defect tests wreak havoc within the manufacturing sector of the economy The expectation test is not helpful, because, in the absence of some other test which tells us what level of safety it is reasonable to assume, the test is empty. The Learned Hand test is merely the economic negligence test and reintroduces all the administrative costs of gathering the relevant information about the benefit and damage schedules of manufacturers and consumers. That administrative problem is magnified under the general risk/benefit test. How are juries to determine from among all possible feasible designs the one that maximizes social welfare? What do juries know about product design altematives? The regulatory/compliance test does not negate uncertainty because a manufacturer cannot know w h e t h e r c o u r t s w i l l c o n s i s t e n t l y rule t h a t regulatory/compliance will free the firm from liability. Together, the rule of strict liability and the design defect tests have wrought havoc within the manufacturing sector of the economy. But is Modern Products Liability Rational ? Strict liability, as Traynor noted, makes manufacturers insurers of consumers, though a manufacturer can theoretically recover these insurance costs in product prices. By adopting strict liability, tort law has created a system of third party insurance. First party insurance, the type consumers are most familiar with, is purchased to protect against something untoward happening to the purchaser. Third party insurance is used to protect against something untoward happening to someone else for which the purchaser may be legally or otherwise responsible. The problem with third party insurance is that it creates problems of "adverse selection" and "moral hazard" far in excess of those created under a first party system. People will be more inclined to pursue any hospital test or treatment as a result of an accident, regardless of cost, if someone else is paying for it. But if a person has to pay for the service or buy the relevant insurance, he or she would be inclined to consider undergoing only cost-justified treatments. Third party schemes are also much more expensive to administer. So great is the difference in cost between first and third party insurance, that George Priest, America's leading insurance scholar, has recently argued that tort liability's imposition of a third party scheme to allocate product safety risks is the primary source of the recent insurance crisis. The logic of Traynor's argument leads manufacturers to purchase insurance for consumers that consumers themselves would not rationally choose. The design defect tests, which constitute a partial retreat from the role of strict liability, turn out to be extremely costly to administer. These tests are so diverse in their requirements as to give little guidance to manufacturers seeking to determine the level of investment in safety they should make. When applied individually, design defect tests will result in wildly uncertain outcomes. In short, the modem solution has failed to provide what we seek, a principled, rational and predictable body of law regulating product safety. The absence of a predictable and rational legal standard is reflected in the current insurance crunch. The greatest danger to an efficient insurance market is radical uncertainty, and that is precisely what the current products liability regime has given us. The question, then, is whether there exists a set of policy recommendations that would add an element of rationality and certainty to the law governing product safety. Inthis regard, it is important to reconsider whether abandoning the contract mechanism for allocating risk was warranted in the first place. Suggestions for Reform The most fundamental feature of m o d e m products liability is the rejection of the market in favor of a tort solution. This rejection presupposes that "contracts" between manufacturers and consumers do not or cannot adequately address issues of safety, and that a solution must be introduced in the form of a set of externally imposed standards. The very idea of a tort or regulatory solution to the allocation of risks regarding product safety is predicated on a presumed market failure. Contracts betweenmanufacturers and consumers would be generally inefficient, thus a 62 / SOCIETY 9 NOVEMBER/DECEMBER 1989 state-imposed solution would produce efficiency gains. It is not at all obvious that the contract mechanism for allocating risk between manufacturers and consumers - the market solution - would be grossly inefficient. Nor is it clear that a general tort or regulatory solution would produce relative net benefits as compared with the market alternative. The usual argument in support of the market failure theory is that consumers have imperfect information regarding (a) the rights and responsibilities made explicit in their contracts regarding risk allocation; (b) the relative costs of various contracts; and (c) the actual risks involved. This argument, however, contains several flaws. First, contract clauses may be difficult to read or comprehend. The solution to this problem is not to reject contracts, but to enhance comprehensibility. Second, consumers probably do not search adequately for the best contracts, but there is no reason to believe that this will lead to suboptimal contracts. It is just as likely that firms will provide the desired clauses, but exploit the low level of search by recapturing more than their marginal costs for providing these terms. Finally, consumers may underestimate or overestimate risks. If they do so systematically, contracts between consumers and manufacturers are unlikely to be efficient. It is not at all obvious, however, that courts and juries are better positioned to make the appropriate risk/benefit calculations. The information demands on courts and juries in applying any risk/benefit test are extraordinary. Even if systematic imperfect information threatens the efficiency of the market solution for allocating safety risks between consumers and producers, it is not clear that the current approach can do much better. The market-failure argument for rejecting contract in favor o f tort law is inadequately persuasive to support a total rejection of the market/contract as a risk-allocation mechanism. One possible reform of the current law would be to consider ways o f reintroducing the market as a central instrument in the allocation o f safety risks. This could be accomplished in either o f two ways. First, the tort approach could be eliminated entirely in favor of sales contracts. The argument for this change would be that the tort solution creates more problems than the contract approach, whatever its shortcomings. Contract law is the lesser evil. Such a change is at the present time infeasible. A second, more attractive solution would be to combine the contract and tort approaches. The sales contract can be reintroduced as an option in allocating risk by allowing parties to contract around tort rules. In this way, tort rules can apply as default provisions in the event contracting parties cannot efficiently or rationally allocate risk among themselves. Any return to the contract model will increase predictability since the conditions under which a manufacturer is held liable are set out with some considerable degree of certainty in the contract. The rationality of the allocation and distribution of risk by contract depends on the rationality and knowledge of consumers and producers. Consumers may be adequately knowledgeable about the nature and scope of risk, and that is why we cannot rely entirely on markets. Rational actors will sometimes need to avail themselves of rules set by third parties - tort rules. So it is natural to mix contract with tort approaches. But if the current tort approach has led to crises, surely that problem will not be eliminated by the modest change of reintroducing contract as an option. Something has to be done about the tort rules themselves, even if they are to apply only as default provisions. What should those rules be? One suggestion is that the rules of tort law be those which "idealized" consumers and producers - fully rational and informed - would have chosen. The law of tort, of which product liability law is a part, governs obligations regarding safety among strangers. Strangers are persons who, in this sense, lack a contractual relationship. They are not in contracting relationships because the transaction costs of contracting are too high. All auto drivers could, in theory, contract e x a n t e with everyone who is put at risk by their operating an automobile. Such a contract would give the driver a right to put others at risk at some price acceptable to each person endangered. The costs of finding and contracting with everyone who might be put at risk by driving are prohibitive, however, and no contracting occurs. Instead, a rule of liability is imposed in the event someone is injured by motoring. What should this rule o f liability be? One way of answering tiffs question is to imagine that the relevant parties - motorists and pedestrians - could contract with one another at low cost. What contract would they agree to? w h e n contracting or bargaining is too costly to be rational we might adopt institutional arrangements that replicate what a contract between HARNESSING THE AM ERICAN MARKETPLACE 163 the parties would have produced. To determine what roles should apply among strangers we should ask ourselves, "What rules would they have created contractually had transaction costs not made their doing so impossible?" Ex Ante Contracts Let us call the strategy for reform that looks to the preceding question the e x a n t e contract approach. The case for applying the e x a n t e contract to the problem o f product liability reform is especially strong. First, product liability law is a branch o f tort law, and one way of approaching tort law generally is to see the rules it sets forth as rational only if they reflect what individual parties would have contracted to had not the costs of doing so been prohibitive. Second, m o d e m products liability law has its roots in the law of sales, the law in which one's liability is determined by the mix of contract and privity. In effect, the rejection of privity imposes a framework within which ultimate users (and not just immediate purchasers) are treated as if they had a contractual or quasi-contractual relationship with the manufacturer. Cigarette litigation may someday prove extremely problematic We might then view products liability as an extension, rather than a rejection of the contract model, in which case the problem is to specify the terms of a rational contract between manufacturers and consumers. What would such a contract look like? By analyzing potential rules of liability in a hypothetical contract model, we insure that the rules of tort used as default provisions in failed or incomplete contracts allocate risks in ways contract law would. There is no discontinuity between the contract and tort approach. Rather, the tort rules extend the c o n t r a c t approach, e n h a n c i n g the overall rationality and predictability of the scheme of risk allocation. Justice Traynor was right when he noted that manufacturers and consumers are concerned primarily with two aspects of the risks related to product s a f e t y - t h e reduction of risk, and its distribution. We will refer to the first of these as deterrence and the second as insurance. The questions that need to be asked are, "How would rational actors distribute and seek to reduce product safety costs in an optimal contract?" and "How would adhering to that allocation scheme alleviate problems inherent in the current regime?" Answering these questions promises a rational and predictable tort scheme for defective products. The Social Insurance Function of Products Liabifity One advantage of strict liability is that is serves a private social insurance function. Manufacturers who are better situated to purchase insurance do so, then pass the costs to consumers, each of whom in effect buys insurance and pays a very small premium as part of a product's overall costs. This cost, presumably, is far less than what a consumer would have to pay to purchase insurance on his or her own, if in fact, an individual could purchase sufficient insurance. But does this procedure reflect how manufacturers and consumers would distribute the costs o f insurance among one another is they could do so by contract? The answer is yes and no. This may well be the way manufacturers and consumers would distribute insurance costs, but would consumers and manufacturers seek insurance against all risks? Under current law, all risks are insured. But in response to the question, rational consumers would probably not seek insurance against all risks. Consumers appear to insure against pecuniary losses, but are less likely to insure against nonpecuniary ones. What is the difference? The difference is usually explained in terms of the marginal utility of the dollar. Individuals do not secure the same utility from each dollar they possess. If someone has a great deal of money now, but expects to have very little in the future, some of his current dollars would be worth more to him later than they are worth now. A rational agent will try to f'md a way to shift current dollars to the future, when they will be worth more. One way of doing this is to purchase insurance. It follows that rational people will seek insurance - move money from the present to the future - when the agent's marginal utility from these monies will increase in a future period. Pecuniary losses increase a person's marginal utility of money. If at some future point someone is unable to work as a result of an accident, that person's income drops and the value of each dollar relative to its current value may increase. To equalize expected utility, dollars are shifted from "now" until "later" by purchasing insurance. 641 S O C I E T Y 9 NOVEMBER/DECEMBER 1989 Rational actors will tend to buy insurance against pecuniary losses. If one point of a tort liability rule imposed on product manufacturers is to provide insurance to consumers, then a rule which reflects the contract that rational consumers would choose is one that provides full recovery for pecuniary losses. Nonpecuniary losses, in contrast, do not increase the marginal utility of money. For that reason rational actors are unlikely to insure against them. The problem is to identify which losses are nonpecuniary. Several commentators have suggested that "pain and suffering" awards fall into the category of recompense for a nonpecuniary loss, and that such damage awards should, therefore, be eliminated. In support of this claim, some have noted that insurance companies do not sell "pain and suffering" insurance on a first party basis. The absence of available coverage in a competitive industry suggests insufficient demand. This fact supports the hypothesis that rational actors do not guard against nonpecuniary losses, and also reinforces the intuition that pain and suffering constitute nonpecuniary costs. One caveat should be mentioned with regard to the inferences just drawn. The absence of a market for first party insurance for nonpecuniary losses does not necessarily demonstrate the absence of demand. Insurers may be disinclined to offer the insurance, whatever the extent of the demand, because they fear fraudulent claims. These considerations suggest that manufacturers should not be required to provide coverage for nonpecuniary losses precisely because, in the absence of a tort law rule that provides it for them, rational consumers do not demand such protection. One additional suggestion for revision in current practice is to reduce drastically, or to eliminate entirely, recovery for nonpecuniary losses. This proposal is interesting because it shows the tension in all reform proposals as well as the difficulty inherent in any proposal. It typically evokes the response, "Since nonpecuniary losses are caused by the manufacturer, why should he not pay for these costs?" How we resolve the question of who bears the risks of nonpecuniary losses depends on how we view the purposes of tort law. If we start out with the premise that parties causing harm should be liable for all the losses they cause, it appears as if manufacturers should bear those risks. However, if we begin with the premise that tort rules emerge only because the market fails and that a sensible tort scheme would reflect an ideal market solution, we come to a dif- ferent conclusion. In this case, because consumers would not purchase insurance against nonpecuniary losses, it follows that manufacturers should not be required to provide that insurance. Reducing Risk Consumers are at least as interested in reducing risk as they are in insuring against it. How then do we reduce risks to consumers and who should be responsible? Returning to Escola, the California Supreme Court argued that not only were manufacturers better suited to distribute risk (via insurance) but that, as between them and consumers, producers were better suited to reduce risk, or to decide which risks were worth reducing. Manufacturers can better reduce risk, but how much risk reduction do consumers want? Consumers do not want to purchase more safety than is rational. If safety prevention costs are accurately reflected in product prices, consumers will want manufacturers to engage in optimal safety prevention activities. The consumer's demand for efficient investment in safety can be met by a variety of very different tort liability rules. Several commentators have shown that both negligence and strict liability with the defense of plaintiff's contributory negligence are efficient in the relevant sense. Either a negligence rule or a strict liability rule can, in theory, produce efficient investments in safety. Consider an accident causing $100 of harm. Spending $50 successfully to prevent it would be rational; spending $200 would not be. Assuming we were "risk neutral," we would want a potential injurer to spend anything up to $99 to prevent the harm, but nothing above $100. Both negligence and strict liability give us that result. U n d e r negligence, a m a n u f a c t u r e r will be negligent if she does not spend $99 to prevent $100 of damage. If she does not spend $99, she will be liable for $100, and therefore she will pay $100. But it will make sense for her to pay $99 rather than $100, and she will do so. Under strict liability the manufacturer will .be required to pay the $100 no matter what. In that case, if causing harm is a product of a course of conduct costing less than $100 to avoid, she simply will not engage in it. We get the same result. On the other hand, if it costs $150 to prevent the harm, then under negligence, we will not want an agent to spend this $150. And so the accident occurs. The same is true under strict liability. The injurer will HARNESSING THE AM ERICAN MARKETPLACE / 65 bear the $100 costs no matter what. But if the activity that causes it would cost $150 to prevent, she will forego prevention. The incentives for safety- at least with respect to injurers - are the same under both rules. A liability regime seeking to provide the safety consumers would have bargained for by contract can choose either strict or negligence liability. A tort law approach to products liability increases uncertainty and the costs of doing business Modem products liability is a mixture of strict and negligence hability with that distinction reflected by the difference between manufacturing and design defect tests. The manufacturing defect test is a form of strict liability. Manufacturers are liable if their products fail to measure up to their own standards. Liability that is strict in this sense is desirable on several grounds, not the least of which is its concreteness and predictability. To choose only a manufacturing defect test is to choose strict liability; to choose a design defect test is to take the negligence approach to safety. The greatest degree of uncertainty and unpredictability arises in the apphcation of design defect standards. Consumers want optimal safety and manufacturers want predictability. Both of these can be accomplished by jettisoning all design defect tests in favor of a simple rule of strict liability that allows defendants to defeat claims against them by establishing the contributory negligence of product users. Manufacturers will then be liable for all harms their products cause unless the injuries result from product misuse or inadequate care by consumers. Such a rule is efficient, knowable, and far easier to administer than any of the alternatives for assessing design defect. Not only does this rule of strict liability with contributory negligence replicate the ideal contract, it is also fair. Unlike the current regime, such a rule allows manufacturers to escape or to reduce liability when injuries are wholly or partially the fault of victims. The current practice, in contrast, is weighted towards getting questions to a jury and framing those questions in ways that unduly burden manufacturers. Instead of asking whether the victim took inadequate care in maintaining or using a product, the current system asks whether the manufacturer could have foreseen that a consumer might misuse or negligently use a product. Under the negligence rule, a victim's negligence has the perverse effect of enhancing rather than defeating his case for recovery. The design defect tests may well have reduced rather than increased the consumer's incentive to take care. The kind of product liability law needed is one that increases both predictability and rational investment in safety. A rule of strict liability with defenses of victim misuse is both predictable and efficient. It distributes risks in just those ways consumers and manufacturers would by contract if information about risks were easy to come by and interpret. Providing for Nonpecuniary Losses Two final problems remain. Justice Traynor held the view that deterrence and insurance goals can be pursued coherently and simultaneously, and he t h o u g h t strict l i a b i l i t y was the a p p r o p r i a t e mechanism for doing both. But the argument presented here may raise doubts about this assumption. If we start with the contract model, then consumers should not be compensated by manufacturers for nonpecuniary losses. If manufacturers are not required to compensate their consumers fully, then they will not have to bear the full social costs of their conduct. They will be led to underinvest in safety. There is, at least in theory, a solution to this dilemma. Questions of liabihty should be separated from questions about recovery. This distinction between reasons for holding someone liable and reasons for compensating someone helps us to see our way clear of the dilemma. Manufacturers can be required to compensate victims for all pecuniary costs, but deny victims recovery for nonpecuniary ones. At the same time, fines can be imposed on manufacturers equal to the nonpecuniary costs caused by their products. The imposition of fines moves the proceeding away from the law of tort as we know it, creating a role for a regulatory or administrative agency. We may be able to pursue both the insurance and deterrence goals but not with only one institution such as the tort law. This solution may strike some as controversial and odd. Why should manufacturers pay fines to regulatory agencies who are not victims of defective products? Why not just compensate the tree victims? If we allow victims to recover both pecuniary and nonpecuniary damages through the courts, we do not create a need for regulatory intervention in the 661 SOCIETY 9 N O V E M B E R / D E C E M B E R 1989 process. This would create an additional saving in administrative costs. To achieve deterrence, we must make manufacturers liable for the full costs of their conduct, and that includes both pecuniary and nonpecuniary costs. We can simply allow victims to recover both. On the other hand, if we decide that victims should not be allowed to recover nonpecuniary losses, then we have to f'md some other way of making manufacturers pay the difference. Manufacturers should not be required to provide coverage for nonpecuniary losses We am tom in two directions. On the one hand, it seems "fair" that victims should be compensated for their losses - all losses. On the other hand, why should consumers secure insurance through product liability for losses they would not themselves ordinarily insure against? I offer no solution to this problem, and instead highlight the factors relevant to resolving it. If we begin by viewing manufacturer's liability from the perspective of the market model, then we simply must reject victim recovery for nonpecuniary loss. The argument for allowing consumers to recover nonpecuniary losses requires the formulation of a different basis for product liability, perhaps one grounded in principles of corrective justice. ExtraordinaryCases This leads to a final suggestion: ordinary and extraordinary cases should be examined separately. In the extraordinary case, the damage affects so large a population, sometimes over such a long period of time, that the ordinary tort solution of case-by-case adjudication seems inappropriate. It is too costly and inconsistent. It may lead to significant injustices. Cigarette litigation may someday prove extremely problematic. If court decisions ultimately go against the tobacco industry, hundreds of thousands of individuals, not just this year, but every year for some considerable period of time, will have claims to advance against the industry. Litigating these cases individually does little to enhance deterrence or insurance. Nor will bankrupting cigarette manufacturers do much to provide compensation for people who will suffer from smoking-related illnesses 20 years from now. In these mega-injury cases we should also separate the aims of compensation from those of deterrence. Rather than pursuing both through tort channels, adopting a regulatory approach to deterrence and creating a fund for compensation is a possibility. The present tort system leads to radical changes in manufacturing product decisions like the widespread and deep removal of product lines on a year-to-year basis. A tort law approach to products liability increases uncertainty and, therefore, the costs of doing business - costs ultimately borne by consumers. Weakened defenses under the current system do little to encourage consumer cam in the maintenance and use of products, thus increasing the manufacturer's exposure to liability without a consequent increase in safety. By reinstituting the market as a vehicle for allocating risk, we reintroduce a low-cost instrument. Private arrangements are less cosily than public ones. This is not to say that all risk should be imposed by the terms of private contracts. After all, not every consumer is knowledgeable about the relevant risks. But allowing a market solution permits consumers and manufacturers who want to contract around the tort rules to do so. One purpose of the law then is to provide a framework for private arrangements, not simply an alternative to private arrangements. There are no easy answers to this complicated problem. These proposals am offered in the spirit of open and free public debate. The goal is to get concemed citizens and officials to think about the problems systematically and to be open to a variety of possibilities. We cannot permit a system that is so confusing to manufacturers, and ultimately costly to consumers, to continue unaltered. Jules L. Coleman is Professor of Law and Philosophy and lecturer in Political Science at Yale University.