TORT REFORM: the CASE of MEDICAL MALPRACTICE Author(S): PATRICIA M

Home , Malpractice

TORT REFORM: THE CASE OF MEDICAL MALPRACTICE Author(s): PATRICIA M. DANZON Source: Oxford Review of Economic Policy, Vol. 10, No. 1, ECONOMICS OF LEGAL REFORM (Spring 1994), pp. 84-98 Published by: Oxford University Press Stable URL: http://www.jstor.org/stable/23606301 . Accessed: 28/10/2014 16:05

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to Oxford Review of Economic Policy.

http://www.jstor.org

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

TORT REFORM: THE CASE OF MEDICAL MALPRACTICE

PATRICIA M. DANZON The Wharton School, University of Pennsylvania

I. INTRODUCTION tíons, screening and mediation panels, etc. Some of these reforms have moderated the growth innumber In most countries the prevailing rule of liability for of claims and size of awards (Danzón, 1984a, 1986; medical injuries is some form of negligence rule Zuckerman et al., 1990), but fundamental criti (Schwartz, 1992). Many countries, including the cisms of the tort system remain. More radical UK, the US, and Canada, are increasingly dissatis alternatives, including enterprise liability and no fied with this traditional system. In theory, the tort fault systems of compensation, have been pro system is designed to deter medical negligence and posed, drawing partly on the no-fault schemes that to compensate patients injured as a result of negli have beenin operationin Sweden and New Zealand gent care. The evidence suggests that it performs for two decades. these functions imperfectly, at best, and at high cost, including high overhead costs. Although rough In the UK the rapid increase in number and cost of estimates suggest that the frequency and cost of medical claims in the 1980s led to the introduction malpractice claims is several-fold higher in the US of the National Health Service (NHS) indemnity, a than in other countries (Danzón, 1990), there is a form of fault-based enterprise liability whereby the common concern over the frequency of medical Health Authorities assume responsibility for de injuries and claims, and the costs of compensation fending all claims arising from NHS treatment by and of malpractice insurance premiums. employed doctors; similarly, self-governing NHS hospital trusts assume liability for the negligence of Over the last two decades most states in the US have their employees. Other alternatives, including sev enacted some tort reforms for medical malpractice, eral no-fault schemes, have been proposed by the including caps on awards, offset of benefits from British Medical Association, the Royal College of other collateral sources, shorter statutes of limita Physicians, and the NHS (Fenn, 1993Ö).

84 © OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

Since concern over medical malpractice has been Tort liability performs two primary functions. First, triggered by its budget cost to payers and apparent by providing compensation it acts as a source of inequity in compensation, reform proposals tend to insurance. Second, by imposing sanctions on per focus on these features. However, a broader view of sons found negligent, it deters future negligent the real social cost of medical injuries indicates that behaviour. However, if the tort system is evaluated the primary function of a medical liability system on grounds of economic efficiency, then it can be should be quality control ('deterrence'). Compen justified, if at all, only by its performance in deter sation can be provided at lower cost and more ring negligence. Compensation and risk spreading equitably through other public and private insur can be accomplished at lower cost—and arguably ance systems. Thus the tort system and tort reform more equitably—through either public or private must be evaluated in the context of the full network first-party insurance. In the US roughly 40 cents of of systems of quality control and injury compensa the malpractice-insurance-premium dollar reach tion that exist in all countries. In this paper, section the patient as compensation, compared to over 90 II outlines the economic theory of professional cents for large first-party health-insurance pro liability. Section III summarizes the shortfalls be grammes. Much of the difference—about 40 cents tween this theory and the actual operation of mal of the liability-insurance dollar—is spent on litiga practice systems. Sections IV, V, and VI evaluate tion, equally divided between plaintiff and defence. proposed reforms, including traditional tort re Other real but hidden costs of tort liability include forms, the Swedish and New Zealand no-fault time and anxiety costs borne by the litigants, and compensation schemes,, and proposals for an ad liability-induced distortions in medical practice— ' ministrative fault-based system (AMA, 1988) and defensive medicine'. These additional costs of tort enterprise liability. Section VII concludes. liability are worth incurring only if there are offset ting deterrence benefits, in terms of future injuries averted. II. THE THEORY OF TORT LIABILITY Economic models have examined the efficiency of Physicians and other learned professionals—in alternative liability rules in performing this deter cluding architects, attorneys, and accountants— rence function.1 Optimal or efficient investment in have been singled out from other occupations in injury prevention minimizes the total societal cost their professional liability to clients. The tradition associated with injuries, including costs related to al basis for professional liability is negligence. injuries, prevention, litigation, and other overheads.2 Under a negligence rule, the plaintiff must show This requires that, at the margin, a dollar spent on that the defendant owed a duty of care, that he failed prevention saves a dollar of expected injury-related to conform to the required standard of care, and that costs, including overheads. this failure was the proximate cause of the plain tiffs injury. Traditional rules of tort damages pro A fundamental principle of liability rules is that, if vide for full compensation of pecuniary and non all parties are fully informed about risks and con pecuniary damages. Thus, in principle, the law of tracting is costless, then the allocation of resources medical malpractice holds health-care providers to loss prevention will be the same, regardless of liable only for medically-caused (iatrogenic) inju whether the liability rule is caveat emptor (all ries that are caused by negligence; adverse out losses reside with the victim) or strict liability (all comes that are consistent with the normal risks of losses shifted to the injurer) (Coase, 1963). But if customary medical care are the burden of the pa consumers misperceive risks or contracting is cost tient. Nevertheless, most professionals consider ly, then caveat emptor leads to non-optimal acci liability insurance to be a prerequisite of profes dent rates andnon-optimal insurance (Spence, 1977; sional practice. Shavell, 1980). Such asymmetric information pro

1 See, for example. Brown (1973), Shavell (1980). 2 have normative content to the extent that 'Optimal' and 'efficient' are used here in this technical sense. They only efficiency is a major goal of social policy, without implying that it is or should be the only policy objective.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

vides a rationale for the professional liability of cally biased. With uncertain legal standards, a learned professionals. In the case of medical care, if negligence rule may create non-optimal deterrence patients are less well informed than providers about incentives (Craswell and Calfee, 1986), including the benefits and risks of alternative treatments and incentives for 'defensive medicine'; many valid cannot readily monitor the quality of care deliv claims are not filed and many invalid claims are ered, then the rate of risky procedures and care per filed. Uncertain legal standards lead to a demand procedure may be non-optimal.3 for liability insurance. Perfectly experience-rated liability insurance would not interfere with deter In principle, a negligence rule of liability can rence. But in practice experience-rating is very correct these distortions and create incentives for crude, both because insurers lack the necessary efficient care and risk-taking, under certain condi information and possibly because of political pres tions. These conditions include that courts set the sures. A rule of strict liability in theory eliminates standard of due care at the efficient level, that the need for courts to define due care. But determin damages be optimally set, that providers be hable ing whether an injury was caused by medical care, for failure to obtain informed consent, and that suits rather than by the underlying disease, would re be brought and compensation awarded if and only quire a similar inquiry, as would the no-fault rules if negligence occurs.4 Efficient deterrence incen that exclude 'normal risks' of medical care. More tives can, in theory, also be achieved by a rule of over, a strict liability rule is more vulnerable to strict liability, whereby providers are hable for ah court errors in setting damages and requires admin injuries caused by medical care, regardless of neg istration of many more cases. Thus, once the as ligence. sumptions of perfect information are abandoned, the choice between liability rules becomes ambig However, negligence and strict liability differ in uous a priori and we must turn to empirical evi their allocation of risk, number of claims, and dence on costs and benefits. overhead costs. Under strict liability ah iatrogenic injuries would be compensable through tort. This is Unfortunately, however, accurate empirical evi inefficient if tort compensation is more costly to dence on key components of the costs and benefits administer than first-party compensation. By con of alternative liability regimes is unavailable. Most trast, under a perfectly functioning negligence rule problematic is the measurement of deterrence ben there should be no negligence and no claims, since efits—the injuries averted because liability makes by definition it is cheaper to prevent injuries that providers more careful.5 It has proved impossible to would be deemed neghgent than to pay for the distinguish empirically between efficient liability resulting damages (Shavell, 1982). Injuries that induced changes in medical practice (deterrence) optimally are notprevented could be covered through and wasteful defensive medicine. private first-party or social insurance.

The neghgence system operates in practice very III. CONCERNS WITH THE TORT differently from this theoretical ideal, primarily SYSTEM because the decision-makers—courts, doctors, patients, liability insurers—lack the perfect infor (i) Rising Claim Costs mation that is assumed by the models (Danzón, 1991a). Because courts lack perfect information Although medical malpractice liability has existed about appropriate care, the standards apphed in for centuries, such actions were rare until the late practice are unpredictable and possibly systemati 1960s. In the US from the early 1970s to the mid 3 This holds if act as prediction providers self-interested income-maximizers. Altruism, professional or ethical concerns, or other quality-monitoring mechanisms may modify the result (Danzón, 1991b). 4 for lack of informed consent is to Liability necessary control the rate of risky procedures. Strictly, these conditions are sufficient but not could in necessary. Efficiency theory be obtained if deviations are offsetting, e.g. a shortfall in claims is offset by higher damage awards. 3 It is often that the lack of of argued experience-rating liability-insurance premiums undermines the deterrence potential of tort and that reforms should therefore focus on liability compensation. But even with flat-rated premiums or under the NHS in the UK, the and costs indemnity psychological reputation of liability may still deter carelessness, albeit bluntly.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

1980s malpractice-claim frequency increased at malpractice claims in the US appears to fall far more than 10 per cent a year, and claim severity short of the number of negligent injuries; for other (average payment per paid claims) increased at countries there is no evidence, but the shortfall is twice the rate of general inflation. Claim frequency likely to be even larger. Two detailed studies of is now about 13 claims filed per 100 physicians per hospital records in California (Mills et al., 1977) year, down from a peak of 16 in 1986. This unex andNew York (Weiler et al., 1993) have concluded is pected surge in claim costs precipitated 'crises' in that the incidence of negligently caused injury liability insurance markets in the mid-1970s and just under one per 100 hospital admissions. How mid-1980s, which in turn led many states to adopt ever, both studies used a broad definition of injury tort reforms designed to reduce claim costs, includ and make no attempt to define negligence by weigh ing caps on awards, collateral source offset, and ing marginal costs and benefits of additional pre shorter statutes of limitations. Some of these re cautions. The New York study defined as an forms—inparticular, caps on awards—have slowed iatrogenic injury 'any disability caused by medical the rate of growth in costs (Danzón, 19846,1986; management that prolonged the hospital stay by at Zuckerman et al., 1990). Others appear to have least one day or persisted beyond the patient's the of the negligible or unexpected effects. For example, if release from hospital'. This begs question arbitration reduces the costs of dispute resolution it appropriate level of care, the implied duration of broad may increase the claim filings and number of stay, and expected outcome. Given these that almost patients compensated; however, this increases budg definitions, it is perhaps not surprising et costs. 60 per cent of the injuries were minor, or that the number of claims filed was less than one-tenth of as de During the 1980s the rate of increase in number of the number of negligently caused injuries claims and size of payments was at least as rapid in fined by the study. the UK and Canada as in the US. But in 1987 mismatch between claims physicians in the US were still five to six times However, this apparent deter more likely to be sued than physicians in Canada and injuries does not necessarily imply that and the UK, and awards for comparable injuries rence incentives and compensation are too low. were several times largerintheUS (Danzón, 1990). Compensating small claims through the tort system However, this overstates the difference in real is probably not cost-effective, given other lower Overall incentives compensation to victims, because the attorney's cost compensationmechanisms. and contingent fee (typically one-third of the award) is for care depend on the pecuniary non-pecuniary subtracted in the US and because medical costs are penalties from the claims that are filed, on provid on other shifted to public health-care systems in the UK and ers' risk aversion, and quality-assurance to Canada.6 The increase in malpractice premiums mechanisms.7 The ratio of claims negligent outpaced the increase in claims costs in the UK and injurieswasmuchhigherfor serious injuries: roughly for three such and Canada, particularly for surgeons, as the medical one claim is filed every injuries defence unions introduced rate differentials across one in six is paid (Weiler et al., 1993). Given an the of suit is substan specialties and attempted to shift from pay-as-you iatrogenic injury, probability if there is a valid basis for a claim. go to partial funding of incurred liabilities. The tially greater Harvard concluded that squeeze of sharply rising premiums but constrained Although the study many evidence of a reimbursement under public health systems gener of the claims filed lacked medically this could reflect the limited infor ated intense pressure for reforms. caused injury, mation available to the reviewers. In other studies more complete information, independent re (ii) Mismatch between Claims and Injuries using viewers have concluded that negligence was cer in 31 cent of cases and not The high cost of malpractice claims is not by itself tainly present roughly per in 44 cent, with the remainder uncertain evidence of system malfunction. The number of present per 6 This is not More recent estimates for the UK show the rate of new claims per 100 hospital doctors at 10.5 (Fenn, 1993ft). 100 since the US to all medical specialties, including directly comparable to the US rate of 13 claims per doctors, figure applies to be sued than primary care doctors who are much less likely surgical specialists. 7 to have a low of suit but The optimal tort award for With costly litigation, it may be optimal probability high penalties. or other mechanisms are effective (see Spence, 1977). deterrence purposes is lower if market forces quality-assurance partially

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

(Farber and White, 1991). For claims with negli parties' influence over the outcome, with tittle if gence, the probability of payment was 0.64, and the any loss in efficiency of deterrence and compensa average payment was $258,000; for claims without tion. negligence the probability of payment was 0.24 and the was $65,900. This average payment suggests (v) High Overhead Costs that the most extreme criticisms of the tort system as a random lottery are exaggerated. A final area of concern is the high cost of litigation and implied high overhead rate on patient compen relative to other mechanisms. (iii) Unequal Compensation sation, compensation The fact-finding undertaken in liability systems is Another common criticism is that tort awards often worth incurring only if there are offsetting deter provide very unequal compensation for similar rence benefits. While this is unproven it is plausible injuries. However, although equal compensation (see Danzón, 1985) and there is some empirical for similar injuries might be appropriate if compen evidence of significant deterrence effects. Weiler et sation were the sole purpose of the tort system, al. (1993) find evidence that the proportion of deterrence may require unequal payment for simi injuries attributable to negligence was lower in larinjuries. Theory and empirical evidence suggest hospitals facing a higher probability of being sued, that the settlement process adjusts payments for the given a negligent injury.9 Extrapolating, they con degree of negligence, even though comparative clude that the proportion of negligent injuries per fault is not the typical rule, and this is consistent hospital admission would be 80 per cent higher if with efficient deterrence (and with some defini tort liability were eliminated. This is probably a tions of fairness). Several studies confirm that the lower bound on the deterrent effect of tort liability, disposition of claims conforms to some degree to assuming that the average deterrent effect exceeds legal rules (Danzón and Lillard, 1983; Färber and that marginal effect observed from cross-sectional White, 1991). Nevertheless, considerable unpre analysis. Moreover, elimination of tort liability dictability remains and this undermines deterrence, would probably result in some reduction in other creates incentives for defensive medicine, and con quality assurance and professional monitoring sys tributes to volatility in liability-insurance markets. tems, that have been strengthened in response to liability. Thus, in practice, liability and other qual efforts be not substi (iv) Inappropriate Compensation ity-control may complements, tutes. Another valid concern is that awards for pain and suffering, which account for a large and probably growing fraction of malpractice pay-out, may ex IV. TRADITIONAL TORT REFORMS ceed levels necessary for optimal compensation.8 Theory cannot determine optimal compensation All of the perceived defects of the status quo— for irreplaceable loss, but there is a strong presump imperfect deterrence, imperfect experience-rating tion that payments for pain and suffering are too of liability insurance, imperfect compensation, and high in the US (Cook and Graham, 1977; Danzón, high litigation costs—are ultimately attributable to The 19846). unpredictability of these awards un imperfect information on the part of courts, plain dermines their value for patient insurance and for tiffs, and providers. Although imperfect informa deterrence; it also contributes to volatility in liabil tion under caveat emptor is a rationale for provider ity-insurance markets. Scheduled limits on awards liability, changing the liability rule does not of for non-economic loss, related to the plaintiffs age itself create better information. The practical choice and injury severity, are used explicitly in countries is thus between imperfect alternatives. In evaluat such as Sweden. This type of reform is likely to ing proposed reforms and alternatives, the practical reduce litigation, by reducing uncertainty and the question is whether they are likely to improve

8 Five cent of claims account per for 50 per cent of dollars paid in compensation (Danzón and Lillard, 1983). 9 The Harvard et study (Weiler al., 1993) did not find statistically significant evidence that a higher risk of suit reduces the absolute number of negligent injuries. However, there are statistical reasons why such an effect may be hard to detect.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

efficiency in deterrence, compensation, and admin er alternatives would be more cost-effective. The istration, or at least improve one dimension without Swedish model has been adopted in Norway and loss along others. Finland and has been suggested in the UK (see Fenn, 1993a) and Canada. The Swedish and New Zealand models are often cited as the (i) Traditional Tort Reforms illustrating potential savings from a no-fault, i.e. causation Some modifications of traditional rules would plau only, test of compensability (for example, Weiler, sibly offer net benefits, at least in the US (Danzón, 1991), by analogy with workers' compensation 1985). In particular, annuitized payments (but with systems. However, analogies between these mod amounts fixed at claim disposition) and scheduled els and workers' compensation or proposals for limits on awards for non-pecuniary loss, based on strict enterprise liability are misplaced. A causa injury severity and the plaintiff's life expectancy, tion-only rule of liability is neither necessary nor a are consistent with optimal insurance and would sufficient for low litigation and overheads as Zea reduce litigation with minimal if any loss in deter percentage of premiums in Sweden and New rence. Collateral source offset significantly reduces land. Nevertheless, the experience of these two cost internalization and is therefore, in theory, less systems is instructive.11 desirable than subrogation; however, since subrogation may entail higher transactions costs, (i) The Swedish Patient Compensation Insurance determining the optimal mechanism for eliminat ing double compensation is an empirical question, The Swedish Patient Compensation Insurance (PCI) the answer to which may differ across countries. was established in 1975 by voluntary contract of Proposals for screening and mediation panels, in between medical providers and a consortium tended to streamline claim disposition, may simply insurers to pre-empt the threat of statutory expan increase delay and costs, unless significant penal sion of tort liability.12 Although patients retain the ties are imposed for appeal from their decisions to right to sue in tort under traditional negligence the courts. Adopting the English rule for allocation rules, tort claims have been extremely rare until of court costs could reduce frivolous suits. Howev recently. A key feature of the Swedish model is Pa er, in order to protect risk-averse plaintiffs, defence decoupling of compensation and deterrence. while costs should be applied against the plaintiff s attor tient compensation is provided by the PCI, handled ney, if paid on a contingent basis, rather than the discipline of medical providers is by There is against the individual plaintiff. the Medical Responsibility Board (MRB). notransmissionofinformationbetweenthem, which These reforms are less relevant to the UK and other is said to be necessary to elicit the doctors co European countries to the extent that damages are operation with the PCI. already constrained by schedules or implicit rules, model is its payments from public health-care systems and oth The superficial appeal of the Swedish er social insurance are netted out of the tort award relatively low budget cost and administrative over (collateral source offset), rules of discovery and head rate, and its widespread acceptance by medi increase in procedure are less prone to exploitation,10 and rules cal providers. After an intended initial 21 for cost allocation discourage frivolous suits. claims, claim frequency has stabilized at about per 100 physicians per year, compared to 13-16 claims per 100 physicians in the US; roughly 40 per V. NO-FAULT ALTERNATIVES cent of these claims receive compensation in both countries. But the PCI costs roughly $2.38 per in Even if the most extreme criticisms of the tort capita, or 0.16 per cent of health-care costs whereas medical insurance system are exaggerated, the question remains wheth Sweden, malpractice

10 and several countries. Schwartz (1992) describes differences in procedural rules between the US, Japan, European 11 see Danzón For a more detailed description and evaluation of the Swedish and New Zealand systems, (1993,1994). 12 few ten a received Proposals for statutory expansion of liability grew out of concern that very patients (roughly year) compensation under traditional tort liability.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

premiums are about 1 per cent of (higher) health The PCI's low overhead percentage is not the result expenditures in the US—thus more than a tenfold of using a causation-only test for compensability. difference. Administrative overheads are 14-18 per Although the PCI is often called no-fault, this is cent of total PCI premiums, compared to roughly misleading. From the patient's perspective, the 60 per cent in the US. This low overhead rate is criteria of compensability are quite similar to a often cited as evidence of the potential savings from custom-based negligence rule. An injury is switching from a negligence rule to a no-fault compensable if (i) it occurred with 'substantial (causation-only) rule of compensability for medi probability' as a direct consequence of medical cal injuries (Weiler, 1991), analogous to the strict intervention" and (ii) either the treatment was not liability of employers for workplace injuries under medically justified or the injury could have been workers' compensation. avoided by performing the treatment differently. Normal and even most abnormal risks of standard However, these inferences are based on a misunder medical care are explicitly excluded.17 But from the standing of the PCI. The low budget cost of the PCI, provider's perspective, the PCI is truly no-fault and despite the higher claim frequency, reflects prima no-liability. The PCI eliminates all reference or rily two factors. First, the collateral offset rule shifts inquiry into fault, requires no proof of negligence most of the wage loss and medical expense to other by an individual provider, and entails neither finan social insurance programmes, thereby undermin cial nor reputational consequences for the provider. ing cost internalization and general deterrence.13 This 'no-fault' scheme bears no resemblance to This is cost-shifting, not real cost reduction. Pay strict liability, either in theory or as applied to ments through the PCI vastly understate the true workers' compensation and product liability. cost of compensating iatrogenic injuries in Swe den. The low expenditure on litigation reflects the fact that neither party has strong incentives to oppose or Second, awards for non-economic loss are below appeal the insurer's decision. Physicians have no those in most other European countries and roughly personal stake in the outcome, so generally co one tenth of those in the US.14 Underlying this operate rather than oppose compensation. Patients difference is the much less pro-plaintiff tort regime face low expected net benefits from appealing to the in Sweden.15 Since the PCI is a voluntary alterna review panel or to arbitration, and are probably tive, it must offer plaintiffs an expected pay-off, net uninformed about the appeal process, which is of costs, that at least matches their expected tort closed to the press and public and has ruled in recovery, in order to deflect tort claims. Thus other favour of the insurers in 90 per cent of cases.18 Thus countries that have more generous tort systems the primary factors contributing to the low over could not adopt the Swedish model or other volun head percentage are the elimination of all links tary contractual alternatives and expect to realize between compensation and deterrence, and the comparably low expenditures. Indeed, out-of-court modest level of patient rights, compared to a US tort settlements (which are one contractual alternative) plaintiff (although not necessarily compared to a already offer some of the gains that might be tort plaintiff in Sweden). Other contributing factors expected under a voluntary Swedish model.16 are the simple claim-filing process; administration

13 'General deterrence' refers to the internalization of injury costs to the responsible activity or industry; it operates via effects of and demand prices elasticity. 'Specific deterrence' refers to internalization to the individual responsible; it operates by changing individual incentives for prevention. 14 The mean for non-economic was payment loss $3,800 in 1987, with a maximum of $117,070. Nevertheless, payments for non-economic loss account for 74 roughly per cent of total PCI payments, because economic loss is covered through collateral sources. 15 other have Among obstacles, plaintiffs allegedly difficulty obtaining the expert testimony required under the custom based negligence rule. 16 Of course if the Swedish model were adopted as the mandatory alternative, eliminating the right to sue in tort, then the need to match tort would not be a compensation binding constraint and benefit levels could be set at any level. 17 This is discussed in detail in Danzón (1994, forthcoming). "•Since 1992, major panel decisions and all arbitration decisions are published.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

by a monopoly consortium of insurers, which elim tutes, casual evidence from several countries sug inates insurers' incentives to compete by vigorous gests that they are complements. However, as insti ly opposing claims or experience-rating premi tutional health-care purchasers—health insurers in ums;19 and lack of competition and provider-specif the US, fundholding GPs in the UK, sickness funds ic accountability for costs in the health-care system, in Germany—become increasingly active as con of the which makes providers more willing to tolerate sumer surrogates in monitoring quality care, flat-rated premiums,20 despite significant geographic rationale for tort liability may diminish. differences in claims experience. It remains to be seen whether the PCI will survive the 1992 reforms (ii) The New Zealand Accident Compensation of the Swedish health-care which have system, Corporation introduced more competition and provider account ability. The New Zealand Accident Compensation Corpo ration (ACC) was established in 1974 as a compre Although the PCI database on iatrogenic injuries hensive no-fault compensation system for victims could, in principle, be used for risk-prevention of' personal injury by accident', including 'medical purposes, in practice the information collected is misadventures'. Unlike the Swedish PCI, the ACC insufficient Moreover, although clinics and hosp pre-empts tort actions for such injuries. Between itals are informed about their claims experience, the 1975 and 1989 total expenditures under the ACC cent a or responsible individuals and sometimes even the grew at a nominal rate of over 20 per year, infla nature of the injury are not identified. roughly 6 per cent a year after adjusting for tion. This exceeds the average rate of inflation of Patients can file a claim with the MRB if they feel US malpractice premiums over the same period. that their treatment was negligent or contrary to the However, these two figures are not strictly compa reflects all code of medical practice. They bear their own filing rable because the New Zealand figure cannot costs and receive no compensation. Providers may injuries. Trends in costs of medical injuries because med be sanctioned by a reprimand or warning, but this be distinguished from other injuries, in the ACC has no financial consequence and probably at most ical injuries have not been identified a minor reputation effect. There are roughly six database. MRB claims per 100 physicians per year, of which one in six receives some sanction. Thus the ratio of In 1992 very significant reforms of the ACC were MRB sanctions to paid PCI claims is less than one enacted. In particular, the rules for medical injuries that are in ten—a rough measure of the loss in potential restore criteria of compensability quite deterrence that results from decoupling compensa similar to traditional negligence rules, but without tion from medical discipline. restoring the tort system for claims adjudication. The problems under the original ACC that led to The main lesson from the Swedish PCI experience these reforms are instructive. is that a sufficient and, possibly, a necessary condi defined tion for low overhead costs and provider co-opera 'Personal injury by accident' was broadly include and mental tion in patient compensation is to forgo all links by the ACC statutes to 'physical dental and between compensation and injury prevention. damage caused by medical, surgical, was to Whether or not the loss in deterrence outweighs the first aid misadventure'. The original intent risks of medical care but reduction in litigation costs is an empirical ques exclude illness and normal fall outside the tion, the answer to which may differ across coun to include medical injuries that normal but not limited to tries, depending on their tort systems and on the realm of risk, including In adverse costs and effectiveness of other systems of quality those caused by negligence. practice, have either low or control. Although in principle tort liability and outcomes that very probability have been considered com other systems of quality control should be substi unexpected severity

" be if Sweden the EC. This monopoly structure would probably have to changed joins 20 that are for and of the public The PCI is financed by premiums paid by county councils, responsible financing provision and other Premiums are assessed on a health-care system in Sweden, and by private physicians, dentists, paraprofessionals. flat per capita basis, regardless of claims experience.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

pensable. This is similar to the criterion of 'unin tions of procedures, and injuries related to lack of tended and unexpected' adverse consequences pro informed consent, misdiagnosis, ortreatmentomis posed by Weiler (1991), which was rejected by the sions, unless resulting from negligence.22 founders of the PCI as unworkable (Oldertz, 1988). The 1992 reforms eliminated the shifting of costs to Defining compensability in terms of an event that the public health-care system and internalized to is unexpected or of unexpected severity suppresses the ACC all medical costs incurred by beneficiar but does not eliminate the need to determine wheth ies. Previously, medical expenses incurred by ACC er the care was appropriate. If 'expected' is defined beneficiaries were largely borne by the public health as a statistical probability, this depends on the level system, except that the ACC paid directly for of care delivered, relative to the condition of the services in private hospitals, co-payments, and particular patient. If a subjective measure of 'ex services not covered by the public system. This pected' is used, this presupposes some notion of cost-shifting undermined the ACC's incentives to informed consent and is surely impossible to deter monitor claim duration. mine ex post. Rulings and commentators have sometimes used objective criteria, sometimes sub Overhead costs are less than 10 per cent of total jective.21 expenditures and payment is prompt. However, far from indicating efficiency, this simply reflects the Difficulties in implementing this definition led to ACC's practice of accepting over 80 per cent of numerous proposals for change, including relying claims as filed, relying largely on physicians as on ICD-9 (International Classification of Diseases) gatekeepers to certify that a claim is a 'personal definitions of injuries. Others urged extending the injury by accident' and, in cases of permanent system to all incapacity, arguing logically that a no disability, that continued benefits are necessary. fault scheme cannot inequitably distinguish among But physicians have no incentive to oppose claims; victims with similar conditions, on grounds of the indeed, until recently physicians could benefit from cause of their injuries. This objection applies to any certifying a claim, since the ACC paid higher fees system that provides compensation selectively to than did the NHS and ACC compensation acted as victims of medical injury, but with no deterrence a bar to a tort suit. Thus this mechanism of claims rationale for the discrimination. adjudication may have saved overhead costs but has contributed to the rapid escalation of total The 1992 reforms adopted a far more restrictive claims costs. The ACC databases did not identify definition that goes a long way towards restoring a iatrogenic injuries. Thus premiums could not be negligence standard of compensability. 'Medical levied on medical providers and the frequency and misadventure' is now defined as 'personal injury causes of medical injuries could not be monitored resulting from medical error or medical mishap'. for risk-management purposes. Again, such econo 'Medical error' is 'the failure ... to observe a mizing on overheads may be 'penny wise but pound standard of care and skill reasonably to be expected foolish', skimping on budget costs but with higher in the circumstances'. 'Medical mishap' is deter real social costs. mined largely on the basis of' rarity and severity' of the outcome, specifically, less than a 1 per cent Prior to 1992 the costs of iatrogenic injuries were probability of occurring, provided that the injury hidden in the payroll and general taxes that financed severity exceeds a threshold. This category specif all injury compensation to workers and non-work excludes ically abnormal reactions and complica ers, respectively. The 1992 reforms authorize the

21 Venell notes that in one case to (1992, p. 4) appealed the High Court, 'Bisson J. appeared to move away from the previous that if the objective approach, risk was one that was known to the medical profession then it was not medical misadventure. He a which adopted subjective approach involved looking at things from the point of view of the victim (and her medical Duncan refers to an 'which is advisers).' (1984) injury unexpected and undesigned by the person injured'. 22 'It has been apparent that difficult questions of causation manifest themselves when the task is to establish a causal link rather as in to than, negligence, having prove that a potential tortfeasor has failed to attain an appropriate standard of care, to which the subsequent damage was causally linked' (Venell, 1992). The suggestion is thus that causation is more, not less to difficult establish, once the element of negligence is removed.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

ACC to establish a medical misadventure account, investigation. The true overheads of an insurance or funded by premiums paid by registered health-care accident compensation scheme include not only the professionals, with experience-rating andno-claims measured overheads, but also the deadweight loss bonuses. This is consistent with the shift towards from unnecessary injuries and inappropriately com greater autonomy and accountability of health-care pensated claims (Danzón, 1992). Unfortunately providers as part of the reform of the health-care this is not observable, but in the ACC it is likely to sector.23 Whether the ACC will in fact exercise its be very high. authority to assess medical providers remains to be seen. In the past it has compressed employer premi ums rather than exploit its full statutory authority to VI. OTHER PROPOSED use experience-rating. As in Sweden, lack of com ALTERNATIVES TO TORT petition—the ACC is a public monopoly insurer— is a necessary condition for the survival of flat-rated Reform of several key dimensions of the malprac premiums. The 1992 reforms also authorized the tice system have been proposed: the rule defining a ACC to report potentially negligent medical compensable injury (causation only, low probabil misadventures to the appropriate disciplinary body. ity/high severity, medical error); provider responsi However, since reporting is apparently discretion bility (individual, enteiprise, none); measure of ary, the effect of this clause also remains to be seen. compensable damages (schedules, deductibles, collateral source offset); forum of adjudication The New Zealand experience under the original (administrative agency rather than traditional tort ACC structure illustrates pitfalls to be avoided trial); and financing (premiums paid by individual rather than providing a useful prototype that other providers or health-care organization, broader tax the countries might copy. The original definition of a es). Some of these changes could be applied to its other compensable event raised practical and philosoph traditional tort system, while retaining ical issues that led almost inevitably to proposals to features—for example, the NHS indemnity scheme doctors with a form expand the system to cover incapacity. However, replaces liability of individual retains tradi the huge budget costs of such a system and the of enterprise liability, but otherwise difficulty of defining incapacity led to the restora tional tort rules; some, such as scheduled damages source tion of a quasi-negligence criterion of patient com for pain and suffering and collateral offset, In evaluat pensation, but without provider-specific liability, are already in place in several countries. in that resembles in some ways the PCI criterion. ing comprehensive changes, such as the NHS demnity scheme or proposed alternatives, it is As in the Swedish PCI, the low administrative costs important to identify the problems that the scheme and evaluate the scheme in should not be interpreted as a measure of efficiency. is intended to address of Rather, low overheads reflect the elimination of all the context of its overall effect on the social costs deterrence and true overhead links between compensation and deterrence. The injuries, including visible costs. causes of medical injuries are not investigated and costs, not merely the more budget US there is no feedback to the individual providers that Here I comment briefly on two comprehensive the NHS scheme in the are responsible for the injuries. The proposed intro proposals and on indemnity duction of experience-rated premiums for physi UK. cians is likely to raise providers' opposition to and hence raise and patient compensation litigation (i) The AMA Administrative Fault-based System overhead costs. As in Sweden, the elimination of all has an administrative provider liability, explicit or implicit, is crucial to The AMA (1988) proposed that would remove the non-adversarial adjudication of claims. In addi fault-based system (AFBS) to a new tion, in New Zealand the very low overhead per medical malpractice claims from the courts 'Medical centage reflects the rapid increase in claims pay specialized administrative Malpractice The aim is to extend ments (the denominator) owing to minimal claims Review Board' in each state.

23 and of care. have Like the NHS reforms, the goal of these reforms was to separate financing provision hospital Hospitals and must to services to regional been reorganized as 'Crown health enterprises', with autonomous boards, compete provide health authorities.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

compensation to more patients; streamline admin Standards; the Board has ultimate authority over istration, including screening out invalid claims these functions. more promptly; strengthen deterrence and spread the costs of compensation more broadly. To strengthen professional discipline, the Board would operate a clearing-house for reports from The AFBS retains a fault-based rule of liability, several sources, including settlements and awards modified to define the standard of care in terms of in malpractice cases, hospital reviews, reports from a' of range reasonableness', based on the standards other physicians (who are required to report sus of a and prudent competent practitioner in the same pected incompetence, impairment, and drug or or similar circumstances. The traditional contribu alcohol dependence of their colleagues), and other tory negligence rule is replaced by comparative state disciplinary actions. The Board can also in negligence. Whereas, traditionally, compensation vestigate reports of substandard performance from is in denied if the principle relative contributions of several sources, including members of the public the patient's underlying condition and the provid and, following a full due process proceeding, may er's actions were 55 and 45 per cent, respectively, impose sanctions including fines and licence revo under the proposed standard the provider would cation. pay 45 per cent of the damages. Apportioning would extend lower damages compensation to more The proposal does not specify in detail how the and could be considered more patients, equitable to system would be financed. The optimal system of and it is also patients providers; consistent with financing would depend on the extent to which the efficient deterrence of injuries.24 Damage rules are system implicitly includes pure social insurance reformed to include an indexed schedule of pay components, for example, as a result of offering ments for non-economic on the loss, depending free legal aid to patients whose claims pass the life and patient's expectancy, collateral source off initial screea Although the incremental deterrence set. Future would be damages annuitized, but the value of provider-specific premiums might be small, amount is determined at time of claim disposition given the direct feedback from claims adjudication to maximize incentives for rehabilitation. to the disciplinary process, the incremental cost may also be small, in terms of increased incentives The administrative procedures for claims adjudica for providers to oppose claims. tion are intended to be less costly, permitting access to those with valid greater claims, while This administrative approach resembles in some out non-meritorious claims. screening Neverthe ways the Swedish PCI, but with important differ less, the remains with process adversarial, attorney ences. The AMA proposal retains and, in some representation, as a condition of necessary protect respects, reinforces links between compensation of both ing rights patients and defendants. Patients and deterrence. Unlike the PCI, the AMA retains whose claims are deemed valid on initial screening the notion of individual provider fault. Both limit are offered free assistance. In order attorney to compensation to injuries caused by medical error both the encourage settlement, plaintiff and ('avoidable' injury in the case of the PCI), defined defendant(s) would be to make blind required set relative to customary medical practice. Both use tlement offers prior to the and would be hearing, written clarification of the criteria of compensability, subject to sanctions if an offer that is they rejected although the PCI rules are much more detailed. By not bettered at the The significantly hearing. hear adopting a comparative negligence standard, the ing resembles a traditional trial, except that it is AMA adopts a more expansive definition of causa adjudicated by an examiner in medical tion experienced and hence expands the number of potentially claims, rather than or malpractice judge jury; ap compensable injuries. peal is to the appellate courts, but on rules of law only. Thus the court cannot review the facts or the Because the AMA retains provider liability, pro finding of liability in a case or set medical particular vider incentives to oppose claims remain. In part 24 Haddock and Curran (1985) show that a standard is comparative negligence potentially efficient if applied conditional on a violation of the standard of care. This is in the AMA which implicit proposal requires fault in addition to the comparative measure of causation.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

this reflects the intent to use the adversarial process but not sufficient condition. Excluded are injuries to eliminate frivolous claims. The AMA offers free resulting from the normal risks of medical care and legal representation to claims that pass an initial imperfect cures of the underlying condition. Thus screen, and permits patients to represent them in contrast to some earlier no-fault models, ESL selves at that screening, whereas the patient's sub recognizes that attributing an adverse outcome to mission to the PCI (and subsequent appeals) must medical care is complex. Because the patient enters be in writing only, unless special permission is the health-care system in less than perfect health, received for oral representation. The AMA process the issue is to identify the incremental harm caused is public and would certainly be subject to contin by the medical system, recognizing that some ad ued public scrutiny for fairness to patients, whereas verse outcomes reflect the inevitable natural course the PCI faces no public scrutiny. For all these of the underlying disease, and that appropriate reasons, the AMA approach is unlikely to yield low medical care entails some positive risks of traumat per capita budget cost and overhead cost compara ic injury, in order to reduce the risk of deterioration ble to the PCI. However, it offers greater deterrence from the underlying condition. and more generous rights to plaintiffs, partly re flecting the political requirement that any serious A simple causation-only strict liability system would reform proposal must offer gains to both sides, create incentives for providers to avoid risky treat relative to the status quo in that country. ments, even though these treatments might on balance be optimal, because providers would be The PCI experience suggests that this administra penalized for bad outcomes but not rewarded for tive alternative would be held to some accountabil good outcomes.25 To counteract this, the fault prin ity because of the implicit or explicit threat that, if ciple must be retained for errors of omission. it operated unfairly, the tort system would be re established. The Swedish experience indicates that, Weiler argues that the causal inquiry is far less if providers prefer the administrative alternative, difficult than is the additional fault judgement, they will design it such that patients are at least as citing evidence from the Harvard study and from well off under the administrative alternative as they Sweden and New Zealand: 'in Sweden and New would be under the tort system. In that case the Zealand, the two countries that have provided no administrative alternative is clearly a Pareto-im fault compensation for medical injuries... it has provement: it survives only as long as both sides are been possible to draw a causal dividing line without better off. any pronounced administrative burden for the no fault programs as a whole.' As argued above, the lower administrative cost in Sweden (ii) Elective Strict (No-fault) Liability (ESL) percentage and New Zealand results from decoupling of deter from use of a causa This proposal, modelled on the workers' compen rence from compensation, not also sation system (Weiler, 1991; Weiler et al., 1993), tion-only standard of compensation (see full cost would empower hospitals and other health-care Danzón, 1993, 1994). Since ESL retains costs organizations to offer an administrative alternative, internalization, litigation and administrative in return for a waiver from common law tort liabil would certainly be higher than under the PCI or ity. The proposal is for an elective system initially, ACC. in order to gain experience before moving to wider in administrative implementation. Any savings that ESL realizes costs would result primarily from the use of an and scheduled Victims of iatrogenic injury would be compensated administrative disposition process These can be made without regard to provider negligence or fault. The damage payments. changes from fault to a rule of strict intent is a system of strict enterprise liability, without switching as illustrated the AMA Sim analogous to the liability of employers for workplace liability, by proposal. fromthe individual injuries. However, medical causation is a necessary ilarly,if channellingliability

25 reflected the true benefits to In In theory, this bias would not occur if fees for medical services expected patients. practice, are informed and feet are constrained by fees do not accurately reflect expected benefits because patients imperfectly regulation.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

doctor to the health-care enterprise offers real effi Claims process is accurate or at least unbiased. If in ciency savings, this could be done within the con practice the higherclaims rate forsurgical specialties text of a fault-based rule, and could be done by reflects the fact that surgical errors are more obvi voluntary contract. Consistent with this, the Kaiser ous than medical errors, some inter-specialty cross chain of health maintenance organizations (HMOs) subsidies may be justified. assumes liability for all practitioners within its organization. Not surprisingly, such contractual Although the NHS indemnity may be viewed as a election of enterprise liability is confined to fully form of fault-based enterprise liability, efficiency integrated, exclusive HMOs, whose providers treat gains are likely to be realized only as liability is only HMO patients and whose patients receive all transferred to self-governing trust hospitals and their treatment from HMO providers. In looser, fundholding GP practices. Enterprise liability is non-exclusive networks, which are far more com intended to increase deterrence and reduce over mon because they offer greater choice and flexibil head costs by placing liability on the single party ity to providers and patients, enterprise liability that has the information and the authority necessary could add administrative costs and reduce account tomake decisions with respect to risk management. ability, contrary to the intent of the proposal. Employer liability for workplace injuries fits this model, as do fully integrated HMOs that are exclu If the NHS internal (iii) The NHS Indemnity sive for providers and patients. market is effective, the role of district health author In 1990 the NHS introduced an indemnity scheme ities is as purchasers on behalf of patients. Hospi for its employees, whereby the Department of tals and, in particular, the self-governing trust hos Health assumes liability for all treatments provided pitals, are intended to be autonomous entities that by its employees.26 GPs* medical malpractice sub assume responsibility for the cost and quality of the scriptions (premiums) have always been reimbursed services that they deliver. Thus placing liability on as a practice expense, but hospital doctors paid their district health authorities is at odds with the separa own subscriptions. This system broke down in the tion of purchasers and providers, whereas transfer 1980s with the rising cost of claims. The Medical ring liability to self-governing hospitals and Defence associations were forced not only to raise fundholding GP practices could realize efficiency subscriptions across the board but also to introduce gains. Similarly, proposals in the US to transfer specialty differentials, in order to meet the threat of liability from individual physicians and hospitals competitive entry by commercial insurers, and this to health plans, would probably reduce deterrence exacerbated the cost increase for the high-risk and increase administrative cost if applied to tradi specialties. Canadian specialists similarly experi tional fee-for-service plans and loose networks enced a treble shock in the 1980s: an overall rate HMOs (independent practice associations). Enter increase owing to rising claims costs was exacer prise liability is likely to be efficient only for fully bated by a move to specialty-specific rates and pre integrated health-care systems such as staff model funding rather than pay-as-you-go financing, pre HMOs, where the health plan has the information cipitated by the threat of competitive entry into the and authority to manage the delivery of care. liability insurance market. This contrasts with the Swedish experience, where flat rating and partially pay-as-you-go financing have persisted because VII. CONCLUDING COMMENTS the PCI is operated by a monopoly insurer consor tium. As noted earlier, Swedish entry into the EC There is no simple solution to the problems of may disrupt this tranquil monopoly. Although professional liability. The rationale for profession switching to specialty rating—a rough proxy for al liability arises from asymmetric information claims experience—is temporarily disruptive, in between patients and providers. But changing the the run it is long generally consistent with efficient liability rule transfers decision-making to courts internalization of costs to activities that generate and liability insurers which also lack good informa high injury costs. However, this presumes that the tion. Unpredictable and sometimes erroneous deci

1 This is discussed in detail in Fenn and Dingwall (1990).

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions P. M. Danzón

sion-making by these parties creates incentives for combined with an elective enterprise liability op filing invalid claims, defensive medicine, and in tion and, for the US, the English rule for allocating vestments in litigation to influence the outcome. costs applied to the plaintiffs attorney if cases are The ideal reforms would improve the accuracy of taken on a contingent-fee basis. This maintains cost the decision-making process, structure benefits ac internalization to parties responsible for injuries cording to sound insurance principles, and impose and hence preserves deterrence incentives, while sanctions for abuse of the system. This in turn reducing some of the uncertainties of the traditional should assure efficient deterrence and compensa common-law rules that encourage wasteful litiga tion. tion. No-fault schemes such as the Swedish model, that reduce litigation expense by eliminating all My personal judgement is that an administrative attempt at deterrence, reduce to systems of social fault-based system, with scheduled payments for insurance that single outvictims of medical injuries non-economic loss, written clarification of the rules for special compensation. Such reforms have little for determining economic loss (e.g. inflation and basis in equity. They may shift costs from health discounting), and written criteria of compensability, care budgets but ultimately may lead to higher real is the most promising alternative. This could be social costs of iatrogenic injuries.

REFERENCES

AMA (American Medical Association) (1988), A Proposed Alternative to the Civil Justice System for Resolving Medical Liability Dispute: A Fault-Based Administrative System, Chicago, American Medical Association/ Specialty Society Medical Liability Project. Brown, J. P. (1973), 'Toward and Economic Theory of Liability', Journal of Legal Studies, 2,323. Coase, R. (1963), 'The Problem of Social Cost', Journal of Law and Economics, 3(1). Cook, P. J., and Graham, D. A. (1977), 'The Demand for Insurance and Protection: The Case of Irreplaceable Commodities', Quarterly Journal of Economics, 91,143. Craswell, R., and Calfee, J. (1986), 'Deterrence and Uncertain Legal Standards', Journal of Law, Economics and Organization, 2(2). Danzón, P. M. (1984a), 'The Frequency and Severity of Medical Malpractice Claims', Journal of Law and Economics, 27,115. — (1984b), 'T ort Reform and the Role of Government in Private Insurance Markets', Journal ofLegal S tudies, 13(3), 517. — (1985), Medical Malpractice: Theory, Evidence and Public Policy, Cambridge MA, Harvard University Press. — (1986), 'New Evidence on the Frequency and Severity of Medical Malpractice Claims', Law and Contemporary Problems. — (1990), 'The 'Crisis' in Medical Malpractice: A Comparison of Trends in the United States, Canada, the United Kingdom and Australia', Law, Medicine and Health Care, 18(1-2). — (1991a), 'Liability for Medical Malpractice', Journal of Economic Perspectives, 5(3), 51-69. — (19916), 'Liability for Medical Malpractice: Incidence and Incentive Effects', unpublished paper. — (1992), 'Hidden Overhead Costs: Is Canada's System Really Less Expensive?', Health Affairs. — (1993), 'The New Zealand Accident Compensation System', unpublished paper. — (1994), 'The Swedish Patient Compensation System: Lessons for the U.S.', Journal of Legal Medicine, forthcoming. — Lillard, L. (1983), 'Settlement out of Court: The Disposition of Medical Malpractice Claims', Journal of Legal Studies, 12(2). Duncan, V. M. (1984), 'The New Zealand Experience and Possibilities for the Future', in J. L. R. Davis and D. F. Partlett (eds), Personal Compensationfor Injury, Proceedings of a Seminar sponsored by the Faculty of Law, Australian National University, August. Farber, H. S., and White, M. (1991), 'Medical Malpractice: An Empirical Analysis of the Litigation Process', RAND Journal of Economics, 22,199-217.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions OXFORD REVIEW OF ECONOMIC POLICY, VOL. 10, NO. 1

Fenn, P. (1993a), 'Compensation for Medical Injury: A Review of Policy Options', in C. Vincent, M. Ennis and B. Audley (eds), Medical Accidents, Oxford, Oxford University Press (forthcoming). — (1993h), 'Estimating the Cost of Compensating the Victims of Medical Negligence: Evidence to the Conference Litigation Working Party', — Dingwall, R. (1990), 'The Problems of Crown Indemnity', in Health Care UK 1989, Newbury, Policy Journals. Haddock, D., and Curran, C. (1985), 'An Economic Theory of Comparative Negligence', Journal ofLegal Studies, 14(1). Harvard Medical Practice Study (1990), Patients, Doctors, and Lawyers: Medical Injury, Malpractice Litigation, and Patient Compensation in New York, report of residents and fellows, Cambridge, MA, Harvard University Medical School. Mills, D. H., Boyden, J. S., and Rubsamen, D. S. (1977), Report on the Medical Insurance Feasibility Study, (sponsored jointly by California Medical Association and California Hospital Association), San Francisco, Sutter Publications. Oldertz, C. (1988), 'The Patient, Pharmaceutical and Security Insurances', in C. Oldertz and E. Tildefelt (eds), Compensation for Personal Injury in Sweden and Other Countries, Juristforlaget,Stockholm, 51-78. Schwartz, G. T. (1992), 'Product Liability and Medical Malpractice in Comparative Context', in P. W. Huber and R. E. Litan (eds), The Liability Maze: The Impact of Liability Law on Safety and Innovation, Washington DC, The Brookings Institution. Shavell, S. (1980), 'Strict Liability vs. Negligence', Journal of Legal Studies, 9(9). — (1982), 'On Liability and Insurance', Bell Journal of Economics, 13(120). Spence, M. (1977), 'Consumer Misperceptions, Product Failure and Product Liability', Review of Economic Studies, 64,561-72. Venell, M. (1992), Paper presented at the Oxford Conference on Medical Malpractice. Weiler, P. C. (1991), Medical Malpractice on Trial, Cambridge, MA, Harvard University Press. — Hiatt, H. H„ Newhouse, J. P., Johnson, W. G., Brennan, T. A., and Leape, L. (1993), A Measure of Malpractice: Medical Injury, Malpractice Litigation and Patient Compensation, Cambridge, MA, Harvard University Press. Zuckerman, Bovbjerg, R. R., and Sloan, F. A. (1990), 'Effects of Tort Reforms and Other Factors on Medical Malpractice Insurance Premiums', Inquiry, 27,167-82.

This content downloaded from 165.123.111.89 on Tue, 28 Oct 2014 16:05:29 PM All use subject to JSTOR Terms and Conditions