The Economic Effects of the Liability System

<<P>Liability law, or tort law, allocates the costs of accidents among individuals. Accidental injuries are frequent in modern society. One car rear-ends another, damaging property and causing personal injuries; a soda pop bottle explodes, injuring a consumer; a physician misdiagnoses a patient's illness, resulting in medical complications when the illness is ultimately treated. The reach of the tort system has expanded dramatically over the past forty years. A 1983 RAND study shows that, between 1960 and 1979, total tort awards paid in San Francisco County courts rose between 83 and 1,760 percent in real dollars, depending on case type (Shanley and Peterson 1983, table 11).

In theory, tort law has two principal goals. The first is to provide compensation to parties injured in accidents in an effort to provide a form of social insurance against accidental injuries. If individuals were risk-averse, they would seek to insure themselves against accidental injuries caused by others. If market failures prevent individuals from purchasing such insurance on private markets, however, or if we as a society believe that it is inequitable or unfair to let individuals bear the cost of insuring against the carelessness of others, then government intervention in the form of liability laws may be warranted.

The second goal is to provide potential injurers with incentives to avoid accidents for which the social costs of prevention are less than the costs created by the accident--that is, to induce optimal deterrence. Thus the deterrence goal of tort law is designed to induce individuals to internalize the negative externality created by engaging in careless behavior by charging injurers for the accidents that they cause. If taking precautions against harming others is costly, and if potential injurers are not required to pay for the harm that they cause, they will in general take fewer precautions than are socially optimal. By requiring individuals to pay compensation for the harm they cause if an accident occurs, tort law can induce potential injurers to take the efficient or socially optimal level of precaution--the level of precaution where the marginal costs of precaution equal the marginal benefits--by forcing them to internalize the externality.

Despite its expansion, however, the system for allocating accident costs in the United States performs poorly on both the compensation and the deterrence dimensions. The system has high transaction costs and fails to compensate injured parties appropriately. In addition, the system leads to inefficient precautionary care measures. There is evidence that liability pressure has distorted firms' incentives for innovation. In the health care sector, liability pressure has led to defensive medicine--precautionary treatments with minimal medical benefit administered out of fear of legal liability.

Why does the liability system perform as poorly as it does? How might it evolve in the future, and to what extent will such an evolution improve its functioning? Finally, how might alternatives to tort--such as no-fault insurance--provide individuals with insurance against accidental injury while preserving incentives for innovation and appropriate accident avoidance behavior? This essay explores these questions in three sections. In the first section I provide background on the operation of the liability system and discuss its implications for productive efficiency. Although simple models suggest that the current system may be efficient, this section shows that the efficiency implications of the system are at best theoretically ambiguous in more-complex models. In the second section I review recent empirical evidence from my research on the performance of the liability system, which suggests that reductions in the level of liability would likely improve efficiency. The third section concludes by evaluating the implications for social welfare of some suggested avenues for potential reform.

THE U.S. LIABILITY SYSTEM: BACKGROUND

In general, tort claims are adjudicated in state courts according to state laws and state court precedents. Historically, tort law has required four elements for a successful claim: injury, cause, duty, and negligence. First, the claimant must show that he or she actually suffered an injury and a tangible loss. Second, the claimant must attribute his injury to the actions of the defendant, as opposed to the actions of another individual or to nature. For example, was the death of a hospital patient due to the actions of a physician, or due to the patient's illness? Third, the claimant must show that the defendant owed her a duty to take care; the general rule is that individuals have a duty to avoid "misfeasance," but not to avoid "nonfeasance." Fourth, the claimant must prove that the defendant was negligent. Stated simply, this entails showing that the defendant took less care in connection with the cause of the injury than would have been taken by a "reasonable person" in the defendant's position, or, in cost-benefit terms, showing that the costs of precaution would be less than the expected costs of the damage caused by a failure to take care. Taken together, this four-part test of the validity of a tort claim is known as the negligence rule.

In a simple economic model of the tort system, the negligence rule provides incentives for defendants to engage in optimal precautionary behavior. Defendants are held liable for accidents only in cases in which accident avoidance is less costly socially than the accidents themselves. The negligence rule does not, in theory, provide complete insurance; only plaintiffs injured by a defendant's negligent actions receive compensation. At least in part for this reason, tort law has over the past forty years gradually expanded its emphasis on compensation. In many cases (most notably, those involving defective products), strict liability--which holds defendants liable regardless of whether they were negligent--has replaced negligence as the operative tort standard. Compensation has also been expanded through such changes as the switch from contributory to comparative negligence and the rejection of contractual limitations on liability (e.g., Huber 1988). ¹

A tort system with more expansive compensation than that specified by the negligence rule, even a system with strict liability for all claims, could also provide defendants with optimal incentives. Under strict liability, defendants would fully internalize the costs of all accidents. In cases in which accident avoidance imposed fewer social costs than did allowing accidents to happen, defendants would find it in their interests to take care; in cases in which accident avoidance was cost-ineffective, defendants would decline to take care but would compensate plaintiffs for their injuries.

But this model does not capture important features of the real world. The performance of the liability system quickly becomes theoretically ambiguous if the system imposes transactions costs or erroneously assigns liability. Failures in product markets (such as public goods, externalities, or informational imperfections) may also lead an otherwise optimal liability system to provide incentives for nonoptimal precautionary behavior.

The transactions costs of the current liability system are perhaps its best-known drawback. Administration of tort compensation involves courts, attorneys for the injured (plaintiffs), attorneys for the injurer (defendants), and other institutions such as liability insurers. Even in routine tort cases involving a bodily injury (BI) from an automobile accident, Carroll et al. (1991, 18-19) find that the transaction costs of the private parties, in aggregate, accounted for 33 percent of all tort expenditures; if government spending on the courts were included, the percentage would be higher. The average auto BI claim cost $5,474 (in 1989 dollars), of which $1,829 was spent on transaction costs. These transaction costs can result in both the undercompensation of injured parties (because injured parties bear some of the transaction cost burden) and more precaution than is socially optimal (because injurers pay both for the injuries that they cause and for some of the transaction costs).

In addition, because accurate information on the causes and effects of accidents is difficult to obtain, the system inevitably allocates liability erroneously in some cases. Those mistakes reduce the social value of the insurance the system provides to victims by adding uncertainty to the compensation process. Mistakes also create risks for potential injurers--including the risk that they will be held responsible for accidents that they did not cause and the risk that damages will not equal the true costs of the injurer's negligent behavior. In cases of medical injury, only one in fifteen patients who suffer an injury due to medical negligence receives compensation, and five-sixths of the cases that receive compensation have no evidence of negligence (Harvard Medical Practice Study 1990; Weiler et al. 1993). Rather, the primary determinant of whether an injury will receive compensation is the extent of the injury, not the extent of fault (Brennan et al. 1996). If additional investment in safety reduces the probability of mistakenly being held liable, then potential injurers may take socially excessive precautions (Cooter and Ulen 1986; Craswell and Calfee 1986). Excessive care results from the all-or-nothing nature of the liability decision: small increases in precaution above the optimal level can lead to large decreases in expected liability.

Market failures, such as externalities, can lead to the same results. The most-studied example of the interaction between market failure and the liability system involves the case of innovation and R&D, which in general may be either over- or underprovided by private markets. For example, if firms are not able to capture all the social benefits created by their innovations--because, for example, of beneficial externalities to other firms of one firm's R&D--then, from society's perspective, firms may engage in too little R&D, the liability system notwithstanding. In this case, discouragement of innovation, even through an otherwise optimal liability system, may be socially harmful.

A substantial literature has outlined some possible mechanisms through which liability pressure may affect innovation. For example, Viscusi and Moore (1993) model the effect of liability pressure on firms' product safety/innovation trade-off. They observe that the impact of liability on innovation depends on the interaction between innovation and the firm's ability to produce safety. If innovation enables the firm to increase safety at less cost, then innovation will respond positively to liability; but if innovation makes it more difficult for the firm to produce product safety, then innovation will respond negatively to liability.

Other work focuses on the link between liability costs and the adoption of new technologies, rather than the link between liability and invention. For example, liability considerations may discourage a manufacturer from making safety or other product improvements, if juries use such improvements as evidence that the manufacturer's previous designs were defective (Graham 1991). Along these lines, liability pressure may lead to the nonoptimal extension of the service life of equipment, if costly safety features are effectively mandated by the tort system only for new equipment (see Martin 1991 for a discussion of this as it applies to the aircraft industry).

Moral hazard due to informational imperfections can also lead otherwise optimal liability rules to be welfare reducing. Markets for health care provide a case in point: most patients' care is financed through first-party health insurance, and most physicians' financial liability for acts of negligence is financed through third-party malpractice liability insurance. On one hand, increasing financial penalties for physicians above the expected costs of medical injuries may be welfare improving, if moral hazard from malpractice insurance leads physicians to take too few precautions. On the other hand, moral hazard from health insurance may be more important. Despite their insurance against the financial risks of malpractice, physicians may employ costly precautionary treatments in order to avoid nonfinancial malpractice penalties, such as fear of reputational harm, decreased self-esteem from adverse publicity, and the time and unpleasantness of defending a claim. If this is true, physicians may have the incentive in general to provide precautionary medical care that has social marginal costs greater than social marginal benefits because neither they nor their patients bear the costs of care in any particular case. In this situation, even compensation the expected costs of medical injury may result in social losses due to defensive medicine.

THE EFFICIENCY EFFECTS OF THE LIABILITY SYSTEM: EMPIRICAL EVIDENCE

In theory, then, transaction costs, informational problems, and market failures mean that the implications of the liability system for productive efficiency are theoretically ambiguous. For this reason, empirical evidence is essential to determining appropriate tort liability policy. This section reviews some of my recent empirical research on the effects of the liability system. The strategy of this research is to compare trends in economic and social outcomes from states that changed the rules governing their liability systems with trends in outcomes from states that did not, controlling for other determinants of the outcomes in question. Differences in trends between the two types of states provide an estimate of the effect of the change in rules, under the assumption that there are no unobserved state-level factors correlated both with changes in liability laws and changes in outcomes; I discuss the validity of this assumption below.

Table 1 presents the nine changes in liability law that I analyze, a brief description of each change, the types of claims affected by each change, and a categorization of the theoretical effect of each change on the level of liability. I categorize the reforms in two ways: the direction of their hypothesized effect on liability and the form of their effect on liability. For example, caps on damages decrease the expected level of liability by truncating the distribution of the value of claims. In contrast, mandatory prejudgment interest increases the level of liability by imposing interest on damages from the date of the injury or the date of the filing of the lawsuit, instead of the date of the court's judgment (see Kessler 1996 for both a theoretical and empirical explanation). Both these reforms have a direct effect on liability. Direct reforms include changes in laws that specify statutory increases or reductions in awards: caps on total or noneconomic damages, collateral source rule reforms (which require damages to be reduced by all or part of the dollar value of collateral source payments to the plaintiff), abolition of punitive damages, and mandatory prejudgment interest. Indirect reforms include changes that affect awards only indirectly, such as reforms imposing mandatory periodic payments (which require damages in certain cases to be disbursed in the form of an annuity that pays out over time), caps on attorneys' contingency fees, and abolition of joint-and-several liability for total or noneconomic damages, creation of a patient compensation fund, and imposition of comparative negligence.

Table 2, which assesses the impact of liability on one important determinant of social welfare--labor productivity--presents estimates from Campbell, Kessler, and Shepherd (1998) for selected high-liability industries of the effect on productivity of liability-decreasing and liability-increasing reforms adopted between 1972 and 1990. We measure productivity with the natural logarithm of gross state product (GSP) per worker, where the GSP in an industry in a state is equal to gross output (sales and receipts and other operating income, plus inventory change) minus intermediate inputs (consumption of goods and services purchased from other industries or imported). The estimated effects in table 2 control for state-fixed effects; for time-fixed effects, where the time effects are allowed to differ for states adopting decrease and increase reforms effective before 1972, for states adopting decrease and increase reforms effective 1972-1990, and for sunbelt states; for contemporaneous, once-, and twice-lagged political and interest-group characteristics of states; and for contemporaneous, once-, and twice-lagged economic and resource-base characteristics of states. Thus, the estimates in table 2 represent the percentage difference in productivity growth over time in states that changed their liability laws, relative to productivity growth in states that did not, controlling for time-invariant and time-varying characteristics of states.

The leftmost two columns present estimates from models that exclude controls for the state/time-varying political and interest-group characteristics of states; the rightmost columns present estimates that control for political, economic, and resource-base variables (see Campbell, Kessler, and Shepherd 1998 for details). Because these models use data on inputs and outputs aggregated to the state level from individual firms of different sizes, each state/year/industry observation is weighted by the level of real state GSP.

The estimates in table 2 show that states that change their liability systems to reduce levels of legal liability experience greater increases in productivity than states that do not, measured in terms of constant-dollar GSP per worker. In particular, states adopting decrease reforms experience approximately 1.7 percent greater aggregate productivity growth than states that do not. The magnitude and statistical significance of this result is robust to the inclusion of controls for states' political and interest-group characteristics. Evaluated at the mean value of states' labor productivity over the 1972-1990 period, this finding suggests that decrease reforms are associated with a $603 increase in GSP per worker per year, in 1987 dollars.

Conversely, in several industries, states that change their liability laws to increase levels of liability experienced lesser increases in productivity than states that do not. However, states adopting increase reforms do not experience significantly less productivity growth in aggregate than states that do not; in addition, the estimated effect of increase reforms across industries is not as robust to the inclusion of political and interest-group controls.

Decrease reforms improve productivity the most in the manufacturing and finance, insurance, and real estate sectors (FIRE). According to estimates calculated controlling for political and interest-group factors, states that adopt decrease reforms experience 2.7 percent greater manufacturing productivity growth (statistically significant at the 5 percent level) than states that do not. Evaluated at the mean value of manufacturing labor productivity over the 1972-1990 period, this finding suggests that decrease reforms are associated with a $1,015 increase in GSP per worker per year, in 1987 dollars. Industries experiencing smaller positive but still statistically significant effects of decrease reforms include hotels and lodging places and wholesale trade. States that adopt decrease reforms experience 1.9 percent greater productivity growth in the hotel sector (standard error 1.1) compared with states that do not and 1.0 percent greater productivity growth in the wholesale trade sector (standard error 0.5).

Increase reforms reduce productivity the most in the amusement/recreation sector (subject to high levels of premises liability) and the transportation sector (subject to high levels of auto liability). Estimates calculated controlling for political and interest-group factors indicate that states that adopt increase reforms show 5.1 percent lower productivity growth in the amusement and recreation sector (standard error 1.1) and 2.9 percent lower productivity growth in the transportation sector (standard error 0.7) compared with states that do not. Industries experiencing smaller negative but still statistically significant effects of increase reforms include wholesale trade (subject to product liability) and retail trade (subject to premise liability) of approximately 1-2 percent.

These results are consistent with the hypothesis that marginal reductions in liability from a maximal level improve productive efficiency. However, there are two important alternative hypotheses that are also consistent with these findings. The key issue in identifying the impact of liability reforms on productivity is the potential endogeneity of reforms. If states that adopt liability reforms also have other unobserved characteristics that affect productivity, then the observed association between reform and productivity may be due to these unobserved factors, rather than a causal link. Ideally, we would estimate the impact of reforms by instrumental variables (IV) methods; however, IV identification requires the assumption that some political or economic determinants of reforms do not affect productivity, an assumption with no strong theoretical basis. To address the potential endogeneity of reforms, we control for a wide range of time-varying and time-invariant characteristics of states that may affect productivity and the propensity to reform, and we allow different types of states to have different underlying time trends in productivity.

To confirm that these estimates represent a causal connection between liability reforms and productivity, we estimate in Campbell, Kessler, and Shepherd (1998) how the effect of reforms varies across industries and over the years since the adoption of reforms. First, we find that productivity growth is most responsive to liability reform in industries likely to be subject to the most common sources of commercial liability insurance claims and tort cases: auto accidents, unsafe premises, and defective products. Second, we find that the long-run effects of liability reform on productivity growth (two or more years after the reforms' effective date, generally three or more years after enactment) are greater than the short-run effects of liability reform (no more than two years after the effective date, generally no more than three years after enactment). If the relationship between liability reform and productivity were causal, then one would expect reforms to have greater long-run than short-run effects, because of the time it would take for reforms to change firms' and individuals' behavior.

The observed relationship between reforms and labor productivity also may not reflect a true causal effect, to the extent that GSP per worker is not an accurate measure of performance. For example, GSP per worker may not include all the costs of production. Specifically, firms from states with relatively low levels of liability may have relatively low costs because they do not bear the true costs of production; this could cause a positive association between observed productivity and liability reform even if reform results in the inefficient deployment of resources into externality-intensive uses. GSP per worker also may not measure true value added. To the extent that doctors and hospitals practice defensive medicine--that is, administer treatment with minimal medical benefit out of fear of legal liability--labor productivity in health care would be positively correlated with measures of liability pressure, if increases in liability caused increases in treatment intensity and treatment intensity were positively correlated with labor productivity. Indeed, the weak relationship in table 2 between liability reform and productivity in the health care sector may reflect exactly this measurement problem.

Analysis of industry-specific microdata can address these measurement concerns. Table 3 and figures 1 and 2 (based on the analysis in Kessler and McClellan 1996) explore the effect of liability reform in the health care sector with patient-level data on treatment intensity, health care costs, and patient health outcomes. The table and figures are based on longitudinal data on all elderly Medicare recipients hospitalized for treatment of a new heart attack (acute mycardial infarction: AMI) or of new ischemic heart disease (IHD) in 1984, 1987, and 1990, matched with information on tort laws and state characteristics from the state in which the patient was treated. (Because AMI is essentially a more severe form of the same underlying illness as IHD, we can assess whether reforms affect more or less severe cases of a health problem differently by comparing AMI with IHD patients.) We study the effect of tort law reforms adopted during the 1985-1990 period on total hospital expenditures on the patient in the year after AMI or IHD to measure intensity of treatment. We also model the effect of tort law reforms on important patient outcomes. We estimate the effect of reforms on a serious adverse outcome that is common in our study population: mortality within one year of occurrence of the cardiac illness. We also estimate the effect of tort reforms on two other common adverse outcomes related to a patient's quality of life: whether the patient experienced a subsequent AMI or heart failure requiring hospitalization in the year following the initial illness.

The main hypothesis that we test is as follows. If reductions in medical malpractice tort liability lead to reductions in intensity but not increases in adverse health outcomes, holding constant individuals' characteristics and other state political and regulatory influences, then medical care for these health problems is defensive--that is, doctors supply a socially excessive level of care due to malpractice liability pressures. Put another way, tort reforms that reduce liability also reduce inefficiency in the medical care delivery system to the extent that they reduce health expenditures that do not provide commensurate benefits. We assess the magnitude of defensive treatment behavior by calculating the cost of an additional year of life or an additional year of cardiac health achieved through treatment intensity induced by specific aspects of the liability system. If liability-induced precaution results in low expenditures per year of life saved relative to generally accepted costs per year of life saved of other medical treatments, then the existing liability system provides incentives for efficient care; but if liability-induced precaution results in high expenditures per year of life saved, then the liability system provides incentives for socially excessive care.

Table 3 previews the econometric results in the figures by reporting conditional means for expenditures and mortality for patients from states adopting and not adopting direct reforms, not controlling for patient demographic characteristics or the political or regulatory environment of states. Table 3 reflects several well-known facts about treatment of heart disease in the United States. Real resources spent on hospitalization for heart disease grew dramatically everywhere in the United States. For example, expenditures for elderly patients with heart attack grew between approximately 17 and 24 percent in real terms from 1984 to 1990, depending on the patient's state of residence. Coincident with this growth in resource use was a dramatic improvement in average mortality from heart disease. In 1984, an elderly American had approximately a 40 percent probability of dying within one year of suffering a heart attack; by 1990, despite the fact that the population had aged slightly, the probability of dying within one year of a heart attack had fallen to approximately 35 percent--fully a 12.5 percent decline in one-year mortality (.35-.40/.40), in only seven years. Thus, the average expenditure-benefit ratio of the increased treatment for heart attack care in the 1980s was approximately $50,000 per year of life saved²

However, the marginal expenditure-benefit ratio of the additional increase in care attributable to high levels of medical malpractice liability pressure was much higher. As table 3 suggests, patients from states without direct reforms experienced substantially greater growth in expenditures on their heart disease, without experiencing much greater rates of improvement in their health outcomes, as compared to patients from adopting states. Expenditure growth was slower in the reform compared to the nonreform states for AMI, and trend differences were slightly greater for IHD. In contrast, mortality trends on average were quite similar for reform and nonreform states. These results suggest that doctors practice defensive medicine and that direct reforms to the liability system improve productivity in health care, in terms of achieving reductions in resource use with no adverse effect on output (e.g., patient health). These simple comparisons do not account for any differences in trends in patient characteristics across the state groups, do not account for differences in the political and regulatory environments of states, and do not account for any effects of other potentially correlated reforms. Nonetheless, they anticipate the results that follow.

Figures 1 and 2 present regression-adjusted trends in hospital expenditures and patient health outcomes for elderly heart attack and ischemic heart disease patients (respectively) from states enacting direct reforms between 1985 and 1987, versus all other states. The trends in the two types of states coincide exactly in the initial year because the levels of expenditures and health outcomes in the figures are calculated controlling for fixed differences across states, for time-varying state political and regulatory characteristics, and for patient demographic characteristics: patient age, gender, black or nonblack race, and urban or rural residence. These estimates also control for the presence of indirect reforms at the time of treatment and therefore isolate the impact of direct reforms on defensive practices.³

Figures 1 and 2 confirm that the simple descriptive statistics presented in table 3 are not an artifact of differences across states that are correlated with both direct reforms and medical treatment patterns. For example, the top panel of figure 1 shows that expenditures on heart attack treatment grew 7 percent more rapidly in states without direct reforms, as compared to states adopting direct reforms between 1985 and 1987; this difference was statistically significant at conventional levels. Mortality trends in these two types of states, in contrast, were virtually identical and not statistically distinguishable. As calculations in Kessler and McClellan (1996) show, the expenditure/benefit ratio for a higher-pressure liability regime is more than $500,000 per additional one-year AMI survivor in 1991 dollars; even a ratio based on the upper bound of statistical confidence intervals around the estimated effect of reform-induced treatment on mortality translates into hospital expenditures of more than $100,000 per additional AMI survivor to one year. Results for outcomes related to quality of life--that is, rehospitalizations with recurrent AMI--also showed no consequential effects of reforms.

Results for patients with IHD are qualitatively similar to those just described for AMI (see figure 2). IHD expenditures also grew rapidly between 1984 and 1990. Direct reforms led to somewhat larger expenditure reductions for patients with IHD than with AMI, possibly reflecting the fact that IHD is a relatively less severe form of heart disease for which more patients may have "marginal" indications for treatment. The effects of reforms on IHD outcomes are again small. Thus, direct liability reforms appear to have a relatively larger effect on the expenditure-benefit ratio of IHD treatments.

We estimated several additional models, discussed in detail in Kessler and McClellan (1996), to confirm the validity of these results. As with the research on the effect of reforms on productivity, the key issue is the endogeneity of state-level legal reforms. We estimated models controlling for statute of limitations reforms, to assess whether there were unobserved characteristics of states that were correlated both with the propensity to adopt legal reforms generally and with medical treatment patterns, health care costs, and health outcomes. Statute of limitations reforms (which require that patients alleging injury file suit relatively sooner than was traditionally required) should not have any effect on the treatment of elderly patients for heart disease because medical injury in this population is immediately apparent. We found that statute of limitations reforms had neither an economically nor a statistically significant effect on expenditures and outcomes, consistent with the hypothesis that there these results are not biased by unobserved state-level factors that are correlated with law reforms and health care decisions.

CONCLUSIONS AND AVENUES FOR FURTHER REFORM

There is widespread agreement that the liability system accomplishes neither its compensation nor its deterrence objectives. Previous work has shown that the system is neither sensitive nor specific in its provision of insurance against accidental injury. In addition, my recent empirical research suggests that reforms that decrease liability, particularly direct reforms that statutorily specify a liability decrease, improve productive efficiency. States that adopt liability-decreasing reforms experience approximately 1-2 percent greater aggregate productivity growth than states that do not, measured in terms of gross state product per worker. Analysis of microeconomic data from the health care sector confirms this effect. Elderly Medicare beneficiaries with cardiac disease from states adopting direct reforms have lower growth in the resources used to treat their illness, without worse health outcomes.

Although these studied reforms improve efficiency, they do little to improve the performance of the system in terms of the compensation goal. Caps on damages, for example, limit awards to those individuals with the most serious injuries. For this reason, researchers and policymakers have suggested a wide range of largely untried reforms, some advocating radical changes to the allocation of responsibility of injuries, that seek to address both compensation and deterrence goals. These reforms can be divided into three classes. The most gradual class of reforms retains the current system of trial by judge and jury but adds additional guidelines or other structure to the legal process. Alternative dispute resolution (ADR), the second class of reforms, retains a fault-based system of allocating damages but replaces the traditional judicial system with mediation or arbitration. The most radical reforms propose no-fault insurance for injuries, often coupled with some administrative mechanism for allocating fault.

Guidelines

Guidelines are a commonly suggested mechanism for improving the process of resolving medical malpractice claims, although the general principle behind them could be extended to other types of tort claims. Medical practice guidelines specify appropriate treatments for patients in particular clinical circumstances. Guidelines would affect mainly the fourth element of a tort claim: the determination of negligence. Traditionally, physician negligence depends on a jury's finding of noncompliance with community standards of care, as interpreted by one or more expert witnesses. This relatively unstructured inquiry has the potential to lead to inconsistent or unpredictable application of the negligence rule (Kinney 1995). Statutory reform to state liability law could allow defendants to use compliance with practice guidelines to establish either an absolute or a rebuttable presumption of due care; conversely, guidelines-based reforms could allow plaintiffs to use noncompliance with guidelines to establish either an absolute or a rebuttable presumption of negligence.

By systematizing the standard of care, guidelines may enable the liability system to process cases more quickly, more economically, and with fewer errors. In doing so, they may both improve compensation and reduce inefficient precautionary care. However, design and implementation of a well-functioning system of guidelines is difficult. In health care, for example, ex ante specification of the relationships between illness and appropriate medical treatment decision making would be at best extremely complex and would have to change rapidly with medical technology. Even the best-designed system of guidelines would likely require expert testimony on a case-by-case basis to aid in interpretation and application.

Alternative Dispute Resolution

Under another sort of proposal, states would replace the right to sue for certain types of injuries in tort with mandatory binding alternative dispute resolution, such as arbitration or mediation. ADR proposals generally transfer power to resolve claims into an administrative system with a specialized expert fact finder/decision maker who operates under fewer constraints than civil court judges. In this way, the goals of ADR are similar to those of guidelines: to provide a more rapid and accurate means of delivering compensation and apportioning responsibility for injury. ADR may offer substantial promise. But, to the extent that an ADR system seeks to preserve all the evidentiary and due process rights that the parties would have in a tort case, it would be less likely to offer substantial advantages.

No Fault

No-fault systems are the most radical suggestion for reform of the liability system. No fault would also limit or remove individuals' right to sue for certain injuries and instead compensate them according to a schedule of damages and/or an administrative hearing, generally at a more modest level than occurs in tort, regardless of the fault of the alleged injurer. Most no-fault proposals are coupled with an additional administrative system that seeks to monitor the behavior of potential injurers in order to preserve incentives for appropriate accident avoidance.

A well-functioning no-fault system offers the obvious advantage of improving compensation and reducing transactions costs while still achieving deterrence. The most serious drawback to no-fault compensation systems is their expense--arising out of no-fault's goal of compensating everyone who is injured, instead of only those injured negligently. Indeed, the expense of a no-fault system is proportional to its success in compensating injured parties, particularly the severely injured. One response to this is to compensate only less-severely injured parties through the no-fault system and allow more-severely injured parties to sue in tort. Many states have implemented limited no-fault approaches to compensating people injured in automobile accidents. However, although limiting the reach of a no-fault system may reduce its costs of operation, it would also reduce its benefits (see Carroll et al. 1991, especially chapter 4, for discussion).

There is no simple way to achieve the twin goals of the liability system. Among other factors, transaction costs, informational imperfections, and other market failures lead the U.S. liability system to perform poorly. These same factors, however, limit the ability of more radical, untried reforms to improve compensation for and deterrence of accidental injury. Further study of existing reforms, and experimental implementation of untried reforms, will improve our understanding of the best approaches to liability reform. In particular, identification of the mechanism by which direct reforms improve productivity--how direct reforms improve incentives to take optimal precaution--may suggest how new reforms can preserve the efficiency benefits of existing reforms and also improve compensation.

TABLES AND FIGURES

TABLE 1.
Liability Reforms Used in Analysis

TABLE 2.
Effects of Liability Reforms on Productivity for Selected Industries, 1972-1990 (standard errors in parentheses)

TABLE 3.
Hospital Expenditures and Mortality Outcomes in States with and without Direct Reforms for Elderly Medicare Beneficiaries with Heart Disease, 1984-1990

FIGURE 1.
Regression-Adjusted Expenditures, Mortality, and Subsequent Illness of Elderly Heart Attack Patients in States with and without Direct Reforms

FIGURE 2.
Regression-Adjusted Expenditures, Mortality, and Subsequent Illness of Elderly Ischemic Heart Disease Patients in States with and without Direct Reforms

NOTES

¹ Contributory negligence bars plaintiffs who are at all at fault from recovery; comparative negligence apportions damages on the basis of fault.

² Based on average expenditures for AMI in 1984 of $10,881; average expenditures in 1990 of $13,140; average mortality in 1984 of 35.4 percentage points; and average mortality in 1990 of 39.9 percent. Thus, the average expenditure-benefit ratio of the incremental intensity supplied from 1984 to 1990 was equal to (13,140;ms10,881)/(.399;ms.354) ;eq $50,200.

³ The estimates do not separately control for increase reforms because neither of the increase reforms are relevant to this study. The most common and important increase reform--comparative negligence--does not apply to medical malpractice cases (patients are never contributorily negligent). In addition, no state adopted or repealed prejudgment interest during the 1985-1990 study period.

REFERENCES

Brennan, Troyen A., C. M. Cox, and H.R. Burstin. "Relation between Negligent Adverse Events and the Outcomes of Medical-Malpractice Litigation." New England Journal of Medicine 335, no. 26 (December 26, 1996): 1963-67.

Campbell, Thomas, Daniel P. Kessler, and George Shepherd. "The Link between Liability Reforms and Productivity: Some Empirical Evidence." Brookings Papers on Microeconomics, 1998.

Carroll, Stephen J., James S. Kakalik, Nicholas M. Pace, and John L. Adams. "No-Fault Approaches to Compensating People Injured in Auto Accidents." RAND Report R-4019-ICJ, 1991.

Cooter, Robert D., and Thomas S. Ulen. "An Economic Case for Comparative Negligence." New York University Law Review 61 (1986): 1067-1110.

Craswell, Richard, and John Calfee. "Deterrence and Uncertain Legal Standards." Journal of Law, Economics, and Organization 2 (1986): 279-303.

Graham, John D. "Product Liability and Motor Vehicle Safety." In Peter Huber and Robert E. Litan, eds., The Liability Maze: The Impact of Liability Law on Safety and Innovation. Washington, D.C.: Brookings Institution, 1991, pp. 120-90.

Harvard Medical Practice Study. Patients, Doctors, and Lawyers: Medical Injury, Malpractice Litigation, and Patient Compensation in New York. Report of the Harvard Medical Practice Study to the State of New York. Cambridge, Mass.: The President and Fellows of Harvard College, 1990.

Huber, Peter. Liability: The Legal Revolution and Its Consequences. New York: Basic Books, 1988.

Kessler, Daniel P. "Institutional Causes of Delay in the Settlement of Legal Disputes." Journal of Law, Economics, and Organization 12 (winter 1996): 432-60.

Kessler, Daniel P., and Mark McClellan. "Do Doctors Practice Defensive Medicine?" Quarterly Journal of Economics, May 1996, pp. 353-90.

Kinney, Eleanor D. "Malpractice Reform in the 1990s: Past Disappointments, Future Success?" Journal of Health Politics, Policy, and Law 20 (1995): 99-135.

Martin, Robert. "General Aviation Manufacturing: An Industry under Siege." In Peter Huber and Robert E. Litan, eds., The Liability Maze: The Impact of Liability Law on Safety and Innovation. Washington, D.C.: Brookings Institution, 1991, pp. 479-99.

Shanley, Michael G., and Mark A. Peterson. "Comparative Justice: Civil Jury Verdicts in San Francisco and Cook Counties, 1959-1980." RAND Report R-3006-ICJ, 1983.

Viscusi, W. Kip, and Michael J. Moore. "Product Liability, Research and Development, and Innovation." Journal of Political Economy 101 (1993): 161-84.

Weiler, P.C., et al. A Measure of Malpractice: Medical Injury, Malpractice Litigation, and Patient Compensation. Cambridge, Mass.: Harvard University Press, 1993.

Much of the research discussed in this essay was jointly conducted with Thomas Campbell, Mark McClellan, and George Shepherd. I would like to thank them for working with me and expanding my understanding of the impact of the liability system. Without them, this work would not have been possible. Funding from the Hoover Institution is gratefully appreciated. However, all errors and misstatements are mine alone.