Encyclopedia of Law and Economics
Show Less

Encyclopedia of Law and Economics

Edited by Gerrit De Geest

The second, expanded edition of the acclaimed Encyclopedia represents a major update of the most authoritative reference work in the field of law and economics and the nine print volumes are now released online as a single integrated product.The Encyclopedia provides balanced and comprehensive coverage of the major domain in law and economics, including: criminal law, regulation, property law, contract law, tort law, labor and employment law, antitrust law, procedural law, and the production of legal rules. Each theme or volume is overseen by a leading scholar and each of the 166 entries is prepared by an expert in the field, providing an in-depth and authoritative overview of the individual topic, combined with an exhaustive bibliography, allowing users to access and filter the entire corpus of literature in law and economics.As with the print edition, the Encyclopedia is unique in serving both as an entry point and a platform for advanced research. The online edition is enhanced with Elgaronline’s powerful search tools, facilitating the search for key terms across the entire Encyclopedia, whilst the browse function allows users to move seamlessly between the volumes. These elements combine to create a powerful research tool for any researcher or scholar in the field of law and economics.
Show Summary Details
This content is available to you

Chapter 13: Tort law and liability insurance

Gerhard Wagner

[In: Volume 1, Michael Faure (ed) Tort Law and Economics]

13.1 Introduction

Insurance is a form of risk transfer, allowing the party initially bearing risk to shift it to another party more willing or able to bear it. Parties who are in the business of taking on risk initially borne by others are insurers, those who are relieved of such a burden are the insured. In return for the transfer of risk, insurers demand a price, namely the premium due under the insurance contract. Where the risk in question is harm to the body, health or property of the insured herself, the contract is for first-party insurance. Here, the person injured and the person insured are identical. Third-party insurance is a contract that covers potential harm suffered by others than the insured, that is, third parties. The third-party loss is not insured against out of benevolence or other altruistic motives but because the insured party may be liable in damages towards a third party. Expressed in the language of tort law, third-party insurance covers claims of victims against injurers who are liable in damages.1

13.2 The economic rationale of insurance

13.2.1 The demand for insurance Risk aversion

From the insured’s perspective, insurance is ‘an exchange of money now for money payable contingent on the occurrence of certain events’ (Arrow, 1971, p. 134). Insurance owes its existence to risk-aversion and the declining marginal utility of wealth. If the utility derived from financial resources were constant, there would be no risk-aversion, and without risk-aversion the insurance industry would not exist. Risk-aversion denotes the fact that an individual who is presented with a choice between a small, but certain loss and a higher, but uncertain loss of equal expected value prefers the certain outcome over the uncertain one (Shavell, 2004, p. 258; for a more thorough treatment, cf. Zweifel and Eisen, 2003, pp. 59 ff). A risk-averse individual would choose a certain loss p. 378of 1,000 rather than a lottery with a 50 percent chance of losing 2,000.2 The insurance contract transforms the contingency of a relatively large future loss into the certainty of a stream of small losses, that is, the premiums due under the contract. Imagine a case where an individual runs a 10 percent risk of incurring a loss of 100,000 within the next ten years. Leaving interest aside for the moment, this risk may be transformed into a stream of ten payments of 1,000, one due in each of the ten years.

The stream of ten payments of 1,000 each represents the actuarially fair insurance premium. The total payment of 10,000 equals the value of the risk accepted by the insurance carrier, that is, the expected value of the untoward event. However, the insurance industry would not be able to thrive on actuarially fair premiums, even if one ignores the possibility of investing premiums in the capital markets and earning returns on these investments. Running an insurance business involves administrative costs, which must be covered by the sum of premiums earned. Thus, premiums charged in insurance contracts may well be at a level above the actuarially fair premium. Obviously, the mark-up added to the premium for coverage of administrative costs does not destroy the demand for insurance. In the example above, individuals are apparently willing to pay more than 1,000 per year, for example, a premium amounting to 1,200, in exchange for ridding themselves of a 10 percent risk of suffering a loss of 100,000. The ‘surcharge’ of 2,000 to be paid within ten years over and above the actuarially fair premium reflects the strength of risk-aversion and the associated willingness to pay to transform risk into certainty.

Although individuals are different, it is fair to assume risk-aversion as a general attitude of human beings. The marginal utility of money declines, and as a consequence, large losses weigh heavy on the individual, more heavily than is warranted by their expected value. Even individuals who are modestly wealthy and earn well above average would be ruined financially if confronted by an obligation to compensate victims for severe personal injuries. As an ordinary traffic accident suffices to cripple a human being, such mischief may well trigger damages of an amount far too high for individuals to manage. This is the reason why, at least in countries in the p. 379Western world, most families carry liability insurance, either as a part of household contents insurance, or in the form of a separate policy.

However, liability insurance is also pervasive among firms (cf. the data reported by Mayers and Smith, 1982, p. 281). Obviously, most business entities are risk-averse as well. This is puzzling because it is generally assumed that the public corporation owned by a multitude of shareholders with diversified portfolios is risk-neutral in its dealings (Easterbrook and Fischel, 1991, pp. 43 f, 119 ff). Thus, one would expect that corporations choose the option of self-insurance instead of costly market insurance, and to live with whatever risk remained. In reality, however, even large firms favour market insurance over self-insurance. The reasons for this business strategy are manifold (Mayers and Smith, 1982, pp. 281, 283 ff; Zweifel and Eisen, 2003, pp. 156 ff). For one thing, the involvement of insurance companies allows firms to take advantage of the claims management expertise accumulated by insurers. Liability insurers are ‘born’ defendants and repeat players in the arenas of tort litigation. Firms may avail themselves of the expert knowledge of liability insurers by referring claims and cases to them via an insurance contract rather than defending and processing them themselves. This explains why insurance is sometimes bought (and sold) even after the accident has occurred, taking advantage of the claims management or restoration capacities of the insurer and relieving the insured of the risk associated with unforeseen developments ex post (Smith and Witt, 1985, p. 379; Mayers and Smith, 1982, pp. 281, 285 f). Other reasons for the involvement of insurers are that liability insurance reduces the variance in the performance of business enterprises, and that market insurance may be cheaper in terms of administrative costs than the full diversification of investment portfolios. Finally, corporate officers, in contrast to shareholders, are not diversified with their investment of human capital and thus have an adverse attitude towards the risk of loss in general, and the insolvency risk in particular (Easterbrook and Fischel, 1991, pp. 29 f). Liability insurance helps to keep the business alive and to keep managers in their jobs. Consequences for the deterrence function of tort law

The inability of a risk-averse party to transfer the risk to an insurer constitutes a welfare loss in itself. The utility of an individual who has a demand for the equalization of the marginal utility of money but cannot satisfy that demand is lower than in a state of the world where satisfaction is possible.

However, this is not the only disutility flowing from a dearth of insurance. A liability rule that operates in solidum, without insurance cover being available, distorts the very incentives it was designed to create (Shavell, 1987, p. 209; Endres and Schwarze, 1991, pp. 8 f). At least in the p. 380area of strict liability, a risk-averse individual faced with the risk of becoming liable in potentially large amounts will take excessive care, that is, spend more on precautions than efficiency requires. In addition, the individual will also shun dangerous activities to an extent that is undesirable from a social point of view. Insuring liability for fault

The standard assumption in economic analyses of negligence liability is that the standard of care is set at an efficient level by the courts, and that potential tortfeasors act accordingly (Shavell, 2004, p. 180). A regime of negligence liability presents agents with a choice between foregoing the taking of safety measures and thus saving the associated costs in return for becoming liable towards victims, and taking the precautions required by the law in return for ridding themselves of responsibility for any damage caused. Within the economic model, it is always rational to follow the second strategy because, by definition, the duty of care is set at an efficient level where the marginal costs of a unit of care equal the marginal costs of any damage that is prevented.

Knowing all this, agents see to it that care is taken and liability avoided (Shavell, 2004, pp. 180 f). This raises the question why anybody should need insurance coverage for liability in negligence. It would seem that demand for liability should only flourish in areas of strict liability, where the taking of due care does not isolate the potential tortfeasor from damage claims.

The conclusions drawn from the economic model of tortious liability are in conflict with empirical evidence. Liability insurance is a strong line of the insurance industry everywhere, and nowhere is it confined to strict liability but always it includes negligence liability within its scope. In most jurisdictions, liability for negligence still forms the core of the law of torts or delict such that demand for insurance is largely driven by the risk of becoming liable in negligence.

The explanation for this outcome involves the behaviour of both courts and agents (Shavell, 2000, pp. 171 f). One reason why tortfeasors are being found negligent in court is that judges make errors in defining the level of due care and applying it to the facts of a given case. Real-life judges may feel sorry for the plaintiff and think that imposing liability would not overburden the defendant – perhaps precisely because he does not have to meet the judgment out of his own pocket anyway, but will shift the costs onto a liability insurer. Judges may suffer from hindsight bias and thus overestimate the possibility of foreseeing the harm ex ante. They may err in applying the economic calculus to the facts of the case at hand, or they may even reject the economic approach altogether and impose liability whenever any precaution available at the time of action was not taken, instead of asking whether such precaution was cost-effective or not.

p. 381Secondly, even where the standard of care is calculated and applied in the correct way, it might not have been observed by the agent who later becomes the defendant. Individuals may act irrationally because they lack the time to think things through and to plan their actions, because they lack information or rely on distorted information, because they are overwhelmed by emotion or for other reasons. The best chance for the standard of care to be observed is where firms make informed decisions within a well-designed planning process for safety measures of a technical or organizational nature. Outside this area, decision processes are oftentimes simple, brutish and short. Non-pecuniary losses

It is received wisdom within the economic analysis of law that non-pecuniary losses generate no demand for insurance because such losses do not cause a shift in the utility function of money (Rea, 1982, pp. 35, 36 ff; Danzon, 1984, p. 521; Priest, 1987, p. 1546; Schwartz, 1988, pp. 363 ff; Shavell, 1987, pp. 228 f, 2004, pp. 269 ff: Ott and Schäfer, 1990, pp. 567 f). Because the marginal utility of money does not change from the status quo ante to the state of injury, it does not make sense to transfer money from the status quo ante to the state of injury through the means of insurance.

Simple and clear as this argument may be, the reality is more complicated. For one thing, many non-pecuniary losses create an additional demand for money because oftentimes compensation for monetary loss is incomplete. Secondly, the individual may be able to compensate for the non-pecuniary loss by engaging in expensive forms of consolation, such as sea cruises and the like (Croley and Hanson, 1995, pp. 1813 ff). Thirdly, the fact that there is no or very little first-party insurance for non-pecuniary loss may be due not to lack of demand but to lack of supply. Insurers may shun such policies because of adverse selection, moral hazard and high administrative costs (cf. Section 13.4; cf. also Schwartz, 1988, p. 365; Croley and Hanson, 1995, pp. 1848 ff). Sometimes, however, the high costs of measuring the non-pecuniary loss may be overcome by agreeing on a lump sum due once an injury is diagnosed. Life insurance essentially works this way.

Be that as it may, in the case of liability insurance, coverage for non-pecuniary loss is indispensable. A potential tortfeasor has a demand for coverage irrespective of whether the harm for which he may become liable is of a pecuniary or a non-pecuniary nature. Therefore, the problem – if it is one – must be solved by legislation in the area of tort law and not within the sphere of insurance law. In theory, it would be possible to exclude tortious liability for non-pecuniary harm, but such a move would destroy incentives to take care and to adjust activity to the efficient level (Shavell, 2004, p. 382p. 272). Again, it is suggested that the government should be left to monitor care levels and activity levels in order to levy fines on agents that diverge from the efficient standard (Shavell, 2004, pp. 272 ff). If this suggestion were followed through, society would have to finance a system of tort liability and liability insurance for pecuniary losses, and a system of fines to deter the infliction of non-pecuniary losses. It seems highly questionable that such a duplication of deterrence systems would really lead to the desired savings in administrative costs (see, Section 13.3).

13.2.2 The supply of insurance

Insurers never cover risks as such and as a whole. Rather, they use definitions, exclusions, deductibles and ceilings in order to carve out a well-defined portion of total risk, leaving the rest in the lap of the insured. To the extent that the insurer has taken on risk from the insured, he does not remain idle and simply ‘bear’ the risk but works to make it disappear, as far as this is possible. The tools employed to this end are pooling and sub-division of risks. Pooling: the law of large numbers

Insurers assemble pools of risks that are similar but independent from one another in the sense that the realization of one risk does not increase the probability that another risk may materialize at the same time or during the same period. In compiling a pool of homogeneous and independent risks, the insurer avails himself of the workings of the Law of Large Numbers. The Law of Large Numbers says that, in repeated, independent trials, with the same probability of a particular outcome in each trial, the actual incidence of that particular outcome will converge more closely with the probability of that outcome as the number of trials increases, or, expressed slightly differently, the average of the observed outcomes approaches the expected outcome of a single trial as the number of trials gets larger.

A common illustration is the rolling of a die. The probability that any of the six numbers will result is 1/6, and the expected value of one trial is 3.5. Predictions of the outcomes of one or a small set of trials are impossible to make with a reasonable degree of certainty. However, if the number of trials is increased dramatically to, say, 60,000, the accuracy of predictions increases as well. It is highly likely that around 10,000 of the 60,000 rolls will result in a 1, another 10,000 in a 2, and so on.

Now imagine an insurer selling a policy covering the risk associated with rolling dice. The insurer charges a premium equal to the expected outcome, that is, 3.5, in return for the promise to pay the amount of the outcome, that is, 1, 2, 3, 4, 5 or 6. If the portfolio consists of a single contract, the maximum loss to the insurer is 2.5 (6 − 3.5) and its maximum gain is p. 383again 2.5 (3.5 − 1). What the outcome is – 1, 2, 3, 4, 5 or 6 – is a matter of chance. The risk of loss or gain must be borne by the insurer. If, however, the insurer has a portfolio of 60,000 contracts, the Law of Large Numbers predicts that the ratio between the sum of the outcomes of the 60,000 trials, divided by the number of trials, will equal 3.5. If this is true, then the sum of all outcomes must be 210,000, and the risk has disappeared. The pooling together of the 60,000 independent contracts seems to have transformed the uncertainty of outcome in every single trial into a certainty with regard to the sum of all trials. The risks have cancelled each other out. Simple as it is, the Law of Large Numbers forms the backbone of the insurance business (Stapleton, 1995, p. 821). Subdivision of risk

It would still be wrong to conclude that it is the whole story. What must not be overlooked is that the creation of a large pool reduces the risk associated with every single policy (so-called relative risk) but it increases the degree to which a possible outcome may diverge from aggregate expected loss (so-called absolute risk). While it is true that a high number of trials reduces the likelihood that the aggregate outcome diverges from the expected loss, it tends to increase the range for the absolute difference between the amount of the expected outcome and the amount of any given real outcome (Samuelson, 1963; see also Hellwig, 1995). This is the reason why it would be too simplistic to assume that insurers eliminate risk by creating large pools. Even after having done this, insurers carry a residual risk characterized by a combination of low probabilities and high stakes.

Furthermore, the concept of a large pool of similar but independent risks is an idealistic abstraction. In reality, although risks often differ from one another in certain respects, they are not fully independent of each other, and the risk pool may not be large enough to eliminate even the relative risk of the pool (Borch, 1990, pp. 112 f.; Arrow, 1963, pp. 941–73). Obviously, insurance is able to function even without the Law of Large Numbers. It is possible to insure against single events. A famous example is an insurance cover provided by Lloyd’s against the risk that a Scottish Whisky distillery would have to pay £1 million to the person able to capture the monster in Loch Ness, as it had promised to do in public. Although there are many lochs in Scotland, the particular risk in question was of a stand-alone nature. This did not deter Lloyd’s from granting coverage in return for a premium of £2,500. Within Lloyd’s, a relatively large group of underwriters took the risk on their books, with each underwriter covering no more than 7.5 percent of total risk, and many less than this share. Thus, instead of pooling the risk, Lloyd’s divided it among a group of underwriters.

The same mechanism – division and subdivision of risk – is employed by p. 384the insurance industry, not only for the purpose of being able to sign ‘freak’ policies like the one in the Loch Ness case but in an effort to protect against the residual risk inherent in ordinary portfolios including a large number of homogeneous risks. The classic tool for the purpose of subdivision is re-insurance, that is, the transfer of some part of the residual risk inherent in a particular pool to another insurer. An alternative strategy chosen in the Loch Ness example would be co-insurance. In the world of liability insurance, it is not uncommon that the insurance industry of a local market pulls together in order to cover a particular class of risk, with the German ‘Pharma-Pool’ providing an example. The Pharma-Pool aggregates the capacity of the German insurance industry in order to cover liability risks associated with the marketing of drugs.

13.3 Compensation of victims as a purpose of liability insurance?

Lawyers and politicians tend to believe that liability insurance protects victims. For example, it is thought that in the area of motor accidents victims must not be burdened with the risk of being injured by a driver lacking the financial means to pay compensation. The political aim of guaranteeing that compensation be paid was the dominant motive behind the European convention that made liability insurance for motor cars compulsory (European Convention on Compulsory Insurance against Civil Liability with Regard to Motor Vehicles, signed 20 April 1959).

In contrast, most of the economic literature rejects the objective of victim compensation (Shavell, 2004, pp. 267 ff). The tandem of tort law and liability insurance is said to be a tool too clumsy and too costly to funnel compensation to victims. The better alternative would be first-party accident insurance, which comes at a much lower price in terms of administrative costs. Thus, victim compensation should not be counted among the purposes of liability insurance.

The economic argument may well be called into question (the discussion is summarized in Dewees, Duff and Trebilcock, 1996, pp. 6 ff). On the one hand, the administrative costs of first-party insurance are not negligible. Under current market conditions, it requires a whole set of policies in order to protect oneself from financial losses caused by bodily injury and damage to property of any kind. Personal disability insurance, which guarantees the level of income prior to the event, is expensive even for modest income levels. This is true even under the current institutional framework, where first-party insurance does little more than fill the interstices between tort law and third-party insurance on the one hand and social security benefits on the other. In a world where tortfeasors did not have to compensate victims, premiums would have to be much higher. One reason why first-party insurance for lost income is so expensive is the moral hazard inherent p. 385in a policy that offers full protection. Such a policy promises the current living standard without the need to labour, and this is a very attractive offer for many. As a consequence, such a system would attract many false positives, and it would require resources of considerable magnitude to isolate those from the group of the truly needy without creating too many false negatives.

On the other side of the balance, the abolition of tort law and liability insurance would leave potential injurers without a private law-type incentive to take care and to adjust their activity levels to the efficient mark. As it would be undesirable to leave things at that, policymakers would have to prop up alternative instruments to influence the behaviour of potential injurers. In particular, one would have to resort to administrative sanctions and criminal punishment in order to achieve the level of deterrence formerly achieved with the help of tort law. However, the procedural guarantees which must be observed before levying a fine or imposing punishment are even more demanding than the safeguards of the civil justice system that have to be overcome for a tort claim to succeed. Therefore, there might be less to gain in terms of administrative costs than is commonly thought by disconnecting the deterrence function from the compensation function. It might even be the case that the sum of administrative costs of the substitute systems of deterrence – administrative and criminal sanctions – and of the substitute systems of compensation – health insurance, disability insurance, property insurance – would be greater than the sum of the administrative costs of the tort system and third-party insurance. Empirical research suggests that the combination of tort law and liability insurance might be preferable in certain areas, for example products liability, and the conjunction of public law sanctions with a layer of first-party insurance in others (Dewees, Duff and Trebilcock, 1996, pp. 412 ff, 427 ff).

When comparing the current system of tort law and liability insurance to other institutional options, it must be taken into account that the costs of the civil justice system in the US are much higher than in any other country. In the special field of tort law and liability insurance, European jurisdictions like France, England and Germany have developed systems of lump-sum compensation, not of victims themselves but of ‘collateral sources’ (tiers payeurs), that is, third parties who compensate victims in their capacity as employer or health insurer. The most important institutions here are social insurance carriers, which pay for the immediate needs of the victim in terms of health care and financial assistance. After this has been done, they then turn to the liability insurers in order to recoup their costs. As both the social insurance carriers and the liability insurers are repeat players, these parties have devised schemes that allow for the efficient processing of claims and for their disposal on the basis of statistically p. 386informed estimates (for a comparative account, cf. Wagner, 2003, pp. 306 ff, 2006, pp. 1039 f). Such systems work smoothly and consume but few resources for their own administration.

In sum, even from an economic point of view, the tandem of tort law and liability insurance beats the alternative of distinct systems, one for the purpose of deterrence and another for the compensation goal. The joint administrative costs of distinct institutions will be higher, not lower, than the administrative costs of tort law and liability insurance, which achieve deterrence and compensation in one step.

13.4 Economic problems of liability insurance

13.4.1 Asymmetric information and imperfect insurance

If insurance simply allowed risk-averse individuals to transfer the risk of becoming liable towards third parties on to insurers against a premium that reflected the actuarial value of the particular risk in question plus administrative costs, the solution would be first-best (Shavell, 2004, p. 262, 2000, 2 B; Graf von der Schulenburg, 2005, pp. 285 f). Individuals would be relieved of risk, and insurance carriers would diversify it away and subdivide the remainder. The insured would still do everything efficiency requires in order to reduce the combined costs of precautions and of residual damages to their optimal level, that is, they would take cost-effective precautions and they would engage in a dangerous activity only if the utility generated outweighed either the sum of the costs of precautions and of residual damages or the insurance premium. An insurance contract which rates the risk at its true value and which leaves the incentives of the insured unimpaired is a perfect insurance contract.

In reality, however, insurance contracts – like any other human institution – are imperfect. The source of imperfection is asymmetry of information between the insurer and the insured. Inefficiencies caused by an asymmetric distribution of information between the parties are not peculiar to insurance contracts but represent a general problem in principal/ agent relationships (Arrow, 1985, pp. 37 ff; cf. also Kotowitz, 1987, pp. 207 ff; Graf von der Schulenburg, 2005, pp. 282 ff; Hellwig, 1988, pp. 1065 ff). In the case of insurance, it is not confined to the stage of contract formation and the initial calculation of the premium but continues to haunt the insurance relationship during the currency of the contract.

At the stage of contract formation, the insurer is confronted with the task of calculating the correct premium. In order to do so, the insurer would have to be able to fully identify the factors that bear on the value of the risk represented by a particular insured individual. In reality, the insurer does not have access to the information needed because the p. 387relevant facts are within the sphere of the person insured (hidden characteristics). In life insurance, it is impossible to predict with accuracy the date of death of a particular person. In liability insurance, it is equally impossible to fully appreciate the types of damage the insured will cause during the currency of the insurance contract, to identify potential grounds for liability, to calculate damage levels and to estimate the likelihood that suit will be brought and either settled or litigated successfully. Even if it were feasible to anticipate likely scenarios and to estimate outcomes, it will not be worthwhile doing so in many cases. The effort required perfectly to risk-rate policies comes at a high price in terms of transaction costs. Every dollar or euro spent by insurance companies for the purpose of risk-rating must be earned back in the particular market. Where average premiums are modest, there is very little to be gained by costly efforts at risk-rating and, at the same time, it is unlikely that the insurer will be able to recoup administrative costs through attracting the lower premiums he can offer.

Even where the risk has been rated correctly, the insurer must not sit back and watch things develop during the currency of the contract. Rather, risk evaluation is an on-going job that requires close monitoring of the insured. In most lines of insurance, the event insured against is not fully out of the control of the insured herself. Without doubt, liability insurance is one of the cases in which the insured exerts considerable influence on the proceeds the insurer has to pay under the policy. After all, the insured controls both the level of care and the level of activity. Where the principal is unable perfectly to observe the actions of the agent (hidden action) or where the agent has private information inaccessible to the principal (hidden information), it is impossible for the parties to write an efficient contract (supra, p. 386). A perfect contract would require that the insurer knew everything the insured knew and in addition could observe the actions and omissions of the insured. Only by monitoring the behaviour of the insured ex post, after the contract has been signed, can the insurer learn about changes in risk profile and adjust the premium accordingly. In the example of motor liability insurance, the insurer would have to sit next to the driver and record the amount she drives and the care and foresight that she employs in any given situation in order to adjust the premium continuously. Needless to say, it is impossible to do this, and even if it were possible, an insurer would be ill advised to do so because costs would be prohibitive.

To sum up, real-world insurance contracts are never perfect. It is either impossible or too costly to rate the risk accurately up front, to calculate an actuarially fair insurance premium, and then to monitor the behaviour of the insured constantly in order to make necessary adjustments in due course. The imperfection of insurance contracts gives rise to a tandem of p. 388problems that every insurance market has to cope with: adverse selection and moral hazard.

13.4.2 Adverse selection

Adverse selection can occur in any market with asymmetric information concerning the quality of goods, services or risks (of the pathbreaking Akerlof, 1970; cf. also Rothschild and Stiglitz, 1976, pp. 629–49; Wilson, 1977, pp. 167–207; Spence, 1977, pp. 427, 447; for an overview cf. Wilson, 1987, pp. 32–4; Hellwig, 1988, pp. 1065 ff). In the special case of insurance, it denotes the fact that in a world where insurance premiums are set at average levels and thus do not accurately reflect the risk represented by each individual insured, insurers attract a disproportionately number of bad risks which in turn might cause the market to unravel (Borch, 1990, pp. 319 ff; Graf von der Schulenburg, 2005, pp. 297 ff; Zweifel and Eisen, 2003, pp. 320 ff). The cause of adverse selection is the above-described asymmetric distribution of information between the parties: the insured has more information about the amount and probability of loss than the insurer.

Imagine an insurance market whose demand side is comprised of an even distribution of two types of individuals, the good risks (G) and the bad risks (B). The good risks will incur a liability of 100,000 with a probability of 5 percent, the bad risks will incur the same liability with a probability of 10 percent. The fair premium for the G-type insured is 5,000 whereas for the B-type insured it is 10,000. These are the premiums that will be charged as long as the insurer is able to discriminate between G and B, charging each type its fair premium. If, however, discrimination does not occur, the insurer will allocate the average premium of 7,500 to every insured. Given such a high premium, the Gs will think twice before buying insurance and do so only if their risk-aversion is so intense that they are willing to cover a risk worth 5,000 with a premium of 7,500. Now assume that 10 percent of G drop out of the market because they are willing to live with the risk and save 7,500. Now the risk pool is made up of 52.63 percent (50/95) bad risks and 47.37 percent (45/95) good risks, and the average premium that will be charged is 7,631.58. With a premium this high, the Gs remaining within the pool will ask themselves whether the policy is worth this amount. Again, some proportion of G – let us assume another 10 percent of the original population – will reach the conclusion that a price of 7,631.58 is too much for a cover worth only 5,000 to them. Now the risk pool is 55.56 percent (50/90) B-type and 44.44 percent (40/90) G-type, and the average premium amounts to 7,777.78, which will cause another fraction of the remaining Gs to drop out. This process may continue until all the Gs have left the pool. Once the last G is gone, the premium will be 10,000, which is fair for the Bs. Whether they remain within the pool depends on their degree of p. 389risk-aversion and their ability to pay a high premium, set at a level which reflects the costs of accidents which the bad risks cause.

As demonstrated by the above example, adverse selection destroys part of the demand for insurance in the sense that risk-averse individuals who would be willing to buy cover for a fair premium refrain from doing so if they are charged the average premium. The fact that this demand cannot be satisfied in the market is a deadweight loss within the social balance sheet.

Adverse selection is not only a cause of social welfare losses but may also jeopardize the performance of insurance companies. In order to demonstrate this effect, it is enough to modify the example in such a way that there are two insurers active in the market, a perfect insurer P and an imperfect insurer X. Assuming that P is able to discriminate between good risks and bad risks, he will demand a premium of 10,000 from Bs and a premium of only 5,000 from Gs. His imperfect competitor X, in turn, is not able to do this but charges the average premium of 7,500 from both Bs and Gs. The consequences are obvious: the good risks can save 2,500 by signing up with P instead of X or from switching from X to P. Eventually, all Gs will be customers of P and pay the low premium of 5,000 whereas the Bs will remain with X and pay the high premium of 10,000. However, until the two types have separated perfectly, X will constantly lose money. As X is not able to discriminate between Gs and Bs, the only thing he can do is adjust the premium upward after he has learnt that his pool has attracted more Bs than Gs. Losses incurred in the previous period of insurance remain with X.

13.4.3 Moral hazard Effect: destruction of incentives generated by tort law

Whereas adverse selection is a problem at the stage of contract formation and price calculation, the phenomenon called moral hazard occurs after the insurance contract has begun. The term was developed in marine insurance where it was used as a demarcation concept for the purpose of defining the event insured against (Borch, 1990, p. 325; a different historical account is provided by Baker, 1996, pp. 246 ff). In contrast to the ‘physical hazard’ represented by the high seas, the ‘moral hazard’ was created by the ship owner and crew themselves. In short, it may be said that moral hazard is a risk of an endogenous nature in the sense that the fact that an insurance contract exists changes the incentives of the insured to take care and thus the probability of loss as well as the magnitude of damages (Arrow, 1971, p. 142; Shavell, 1979 p. 541; Zweifel and Eisen, 2003, pp. 295 ff; Endres and Schwarze, 1991, pp. 10 ff; Baker, 1996, 239, who is, however, p. 390deeply sceptical about the conventional economic reading of moral hazard expounded above). Where the insured and her servants have no influence on the probability and magnitude of loss, moral hazard cannot occur. In the case of liability insurance, however, the reverse is true, as both the magnitude and the probability of harm are subject to the decisions and actions of the insured.

Moral hazard is best illustrated in the stylized case where the insurance contract runs for one period only, the risk is rated up front, and the premium paid in advance. Within such a framework, the insured has no incentive to avoid harm. The costs of safety measures, be they of a pecuniary or a non-pecuniary nature, would fall upon the insured, whereas the gains accompanying such measures in terms of reduced harm would accrue to the insurer.

Imagine that the insured faces a 10 percent chance of harm in the region of 100,000. The probability of harm could be reduced to 5 percent by implementing a safety measure that costs 4,000. Because the costs of the precaution (4,000) are smaller than the value of the harm prevented (5,000), the safety measure should be implemented. However, the insured gains nothing if she spends 4,000 on something that benefits a third party to the tune of 5,000. Rather, it is in her self-interest to remain idle, do nothing and let the insurer internalize any losses that might materialize during the currency of the contract.

The resulting loss of any incentive to the insured to prevent harm through the exercise of care amounts to a disaster. In the above example, moral hazard frustrates the purpose of tort law of generating incentives for careful behaviour. If moral hazard were left to itself, the preventive function of tort law would lie in ruins. Given the pervasiveness of liability insurance and the high administrative costs consumed by the tort system, the investment seems to be in vain. Fortunately, this conclusion is premature because it overlooks the causes of moral hazard and thus does not arrive at the instruments for remedy. Source: informational asymmetry

Adverse selection and moral hazard are fruits of the same tree. They both owe their existence to asymmetries in information (Arrow, 1963, p. 941; Pauly, 1968, p. 531; Hellwig, 1988, p. 1072 ff). In a perfect world, the insurer would be able to monitor every action and omission of the insured in order to re-calculate the premium anytime the insured deviated from the course of action prescribed by efficiency (see Section 13.4). If such constant adaptations were feasible, then the self-interest of the insured would coincide with the self-interest of the insurer.

Imagine that a person faced with an expected harm of 0.1 × 100,000 is offered an insurance premium of 10,000, and that, by incurring costs p. 391of 4,000, she could bring the probability of harm down to 5 percent. In deciding whether to incur these costs, the insured will be aware that failure to take the precaution will result in the insurance premium remaining at 10,000, whereas taking the safety measure will reduce the expected costs, and thus the insurance premium, to 5,000. As the sum of the costs of the safety measure (4,000) and the insurance premium (5,000) is smaller than the insurance premium the insurer would charge if the precaution were not taken (10,000), it is in the self-interest of the insured to take the precaution and be rewarded with a modest insurance premium.

In the real world, insurers are never able perfectly to monitor those they insures. On the other hand, it would be an error to think that monitoring does not take place. Where machines, engines, other technical installations or buildings present risks of liability, it is a common policy for insurers not only thoroughly to screen and evaluate these structures ex ante for the purpose of avoiding adverse selection, but also to monitor the state of affairs during the currency of the insurance contract. Such monitoring is impossible to implement with respect to the day-to-day behaviour of human beings, and even were it possible to do so, it would be too costly. It is only worthwhile to incur such costs where the stakes are sufficiently large, that is, where the maximum loss is considerable and the likelihood of harm is non-negligible. Remedy: partial insurance

Where observation of the insured is either impossible or prohibitively costly, one has to look for other instruments capable of restoring the incentives of tort law to take due care. As it turns out, there is a variety of such tools. They all rest on the same idea: the insurance cover is partly removed and the insured exposed to personal liability. To the extent that this happens, the insured will behave as if she was not protected, that is, exercise due care (Endres and Schwarze, 1991, pp. 13 ff).

Partial insurance in its simplest form is represented by caps and deductibles. Both clauses work to the effect that the obligation of the insurer is limited, and the flipside is that the protection of the insured is only partial. If she becomes liable for a loss that exceeds the cap, she has to cope with the costs herself. Thus, she has an incentive to take precautions where the marginal costs of such precautions are less than the marginal loss prevented.

What a cap does at the top end of a potential damage claim, the deductible does at the bottom. With a deductible, a certain sum stipulated in the insurance contract remains with the insured, with the insurer picking up the loss only to the extent that it exceeds the deductible. Such a clause is indispensable in addition to a cap, because, in the ordinary course of p. 392business, large losses are the exception and minor ones the rule. In cases of minor losses, the ratio between the administrative costs incurred by the insurer and the benefits conferred by insurance on risk-averse insured individuals is particularly unattractive.

The question arises whether the insured will accept a deductible, given that an insurance contract lacking such clause will protect them completely from risk. The answer is that rational insured persons themselves have an interest in including a deductible because the savings in terms of reduced premiums outweigh the sum of costs of precautions and of residual losses (Shavell, 2004, p. 263).

To demonstrate, imagine an insured person that stands to become liable in the amount of 100,000 with a probability of 15 percent. Without a deductible, the fair insurance premium would be 15,000. The alternative would be a contract providing for a deductible of 20,000 such that the fair premium would be 12,000. Now assume that there is a safety measure available to the insured that would cost 1,000 and would bring down the probability of harm to 5 percent. Under full cover, the insured has no incentive to take the precaution. With the deductible, the insured will compare the cost of precaution (1,000) to the savings in terms of a reduction in personal liability (10 percent of 20,000 = 2,000) and take the precaution. In doing so, the probability of harm declines to 5 percent, with expected losses on the part of the insurer falling to 4,000 (5 percent of 80,000). The fair premium under the contract including the deductible would thus be 4,000, and the total costs to be borne by the insured would sum up to 6,000 (4,000 for the premium + 1,000 for the costs of precaution + 1,000 expected value of the deductible). The latter amount is far below the premium level of a contract providing full cover, without a deductible (15,000). Activity levels

The fact that liability insurance may impair incentives to take care is generally accepted in the literature. In contrast, the effects of liability insurance on activity levels have not been studied thoroughly. This imbalance has no substantive basis because activity levels are every bit as important for the efficient functioning of tort law as care levels.

Suppose that the insured is strictly liable for harm caused, that there is a 1 percent chance of causing harm to a third party to the value of 500,000 associated with the activity in question, and that the insured has to determine whether to engage in the activity or to refrain from it. A risk-neutral actor would eschew liability insurance and would thus be forced by tort law to consider whether the benefits associated with the activity in question are sufficient to compensate for the expected losses. As expected losses are worth 5,000, she would only engage in the activity if the profits from the p. 393activity – benefits minus costs – were greater than 5,000. If they were not, it would be in the interest of both the insured and of society that the activity should not be carried out.

Liability insurance disrupts the alignment of social interest and self-interest in much the same way as it does incentives to take care. Here, however, roles are reversed. Liability insurance creates a setting in which the benefits of a particular course of action accrue to the insured, whereas the costs are for the insurer to take up. In the above example, imagine that the net benefit from the dangerous activity is not 6,000 but only 4,000. In this case, the insured should not engage in the activity. However, because the expected losses in terms of damage payments worth 5,000 have to be borne by the insurer, the insured looks at the prospect of foregoing an activity that would benefit her to the tune of 4,000. It is apparent that the insured has an incentive to reap the benefits instead of foregoing them to the benefit of the insurer.

In contrast to the issue of incentives to take efficient care, there is very little insurers can do to protect or reinstate incentives to engage in efficient activities only. In theory, the remedy is obvious; the insurer would ‘only’ have to monitor the insured and to charge a premium that is proportionate to risk. For example, insurers could make the premium contingent on the chosen activity level, such as, kilometres driven, the amount of exhaust fumes emitted or the amount of products marketed. In some markets, insurers have introduced such measures, if only haphazardly, and with little accuracy.

13.5 Economic virtues of liability insurance

The previous section addressed the fact that liability insurance may impair the incentives generated by the law of torts. Now the analysis turns to the way in which liability insurance can create incentives to take care and curb activity levels over and above those created by the mere threat of liability. The potential of liability insurance actually to improve matters is caused by the fact that, in practice, liability in tort is always limited.

13.5.1 Unlimited liability on paper, limited liability in reality

In theory, fault-based liability in tort or delict is unlimited, that is, the defendant is bound to compensate losses sustained by victims in full. However, many jurisdictions operate with caps limiting liability to maximum amounts stipulated by law. Within Europe, the directive on products liability provides an instance in point as its Article 16 paragraph 1 allows member states to cap the obligation of the producer at 70 million euros (Council Directive 85/374/EEC of 25 July 1985 on the approximation of the laws, regulations and administrative provisions of the member states concerning p. 394liability for defective products, Official Journal L 210 of 07.08.1985, Article 16 paragraph 1: ‘Any Member State may provide that a producer’s total liability for damage resulting from a death or personal injury and caused by identical items with the same defect shall be limited to an amount which may not be less than 70 million ECU.’).

Regardless of whether there is a cap or whether liability is unlimited on paper, there are much tighter constraints for damage claims in practice. Very few individuals are in a financial position to pay up claims in the region of 70 million euros or above. Many firms will be overburdened by an obligation above 70 million euros too. As far as fault-based liability is concerned, legal responsibility is usually unlimited but real-world resources never are. Regardless of the size and financial strength of a corporate entity, liabilities of the order of hundreds of millions of euros would deplete corporate funds and place the enterprise into the state of insolvency.

13.5.2 The pervasiveness of limited liability

Within the economic literature, Shavell has coined the term ‘judgment proof problem’ for an analysis of the incentive effects in cases where the assets of the tortfeasor are limited (Shavell, 1986, p. 45, 2000, p. 174). Groundbreaking as Shavell’s account was, the language used may present the problem in a false light. The talk of ‘judgment proof’ injurers suggests that the problem affects a particular group within the population only. After all, most of us are not judgment proof but solvent. If this perception exists, it is wrong. In reality, 90 percent of the population is incapable of satisfying a major damage claim by a single victim, let alone the several claims of a plurality of victims. Most people do not own fungible assets subject to execution of judgments, and the fraction of personal income not set aside for household consumption and thus exempt from execution is too small to cover more than the interest due on large damage claims. As to corporate defendants, of course everything turns on the size of the company. However, experience from the US shows that large firms attract large claims, that is, they are singled out by the plaintiff bar as their favourite defendants. In this sense, the existence of a large pool of assets works to attract liability. Firms have reacted to this fact by adapting their corporate structure, that is, by forming subsidiaries with limited assets, which serve as a shield against tort liability (Lopuci, 1998). Therefore, even a large segment of corporate entities falls into the category of ‘judgment proof’ defendants.

13.5.3 Limited liability and the incentive to take care

The inability of a tortfeasor to pay valid claims is problematic for several reasons. In the case of liability in tort, the obvious consequence is that p. 395victims are not compensated. From an economic point of view, this is a point of little concern because there are alternative routes to compensation: first-party insurance and social insurance schemes. In real cases, however, it is too late to insure or the benefits are too small to satisfy victims. The incentive effects of limited liability are discomfiting as well. Asset-based limited liability, that is, the limitation of financial responsibility due to the fact that the tortfeasor holds limited assets only, has the same effect as caps imposed by law. Damages that exceed the legal cap or the value of assets available simply drop out of the calculus of potential tortfeasors. As a consequence, incentives to take care are diminished.

To illustrate, imagine that damages are 1,000,000 and that the probability of harm is 15 percent. The defendant holds assets in the amount of 700,000. There is a safety measure available that would cost 80,000 and bring the probability of harm down to 5 percent. Because the expected value of the measure is 100,000 (10 percent of 1,000,000), society has an interest that it be taken. However, the gain to the firm from exercising care is only 70,000. It will thus decline to make the investment in safety.

13.5.4 Limited liability and the incentive to insure

In the present context, the more important question is whether the firm will buy insurance coverage in an amount exceeding its 700,000 in assets. The answer is that it may well do so because it risks losing these assets in the event of a large, uninsured loss (Shavell, 2004, p. 275; Easterbrook and Fischel, 1991, p. 50). Where the insurance policy has a limit of 1,000,000, and the damages are 1,400,000, the firm will have to hand over an additional 400,000, which is a substantial fraction of its total assets. In this example, a damage claim to the amount of 1,700,000 or greater will cause the insolvency of the firm. Therefore, firms will have an incentive to cover potential damage claims even to the extent that they exceed the value of its assets.

On the other hand, the incentive to insure the ensuing harm in full is diminished when liability is limited. An agent who holds assets worth 100,000 and runs a 10 percent risk of causing harm of the order of 1,000,000 would have to pay a premium of 100,000. If the agent did so, 90 percent of the expenditure would be applied to cover losses she would not otherwise have to bear. Furthermore, the agent would have to hand over her entire assets to obtain full coverage. Even a risk-averse person would not do so.

In certain circumstances, however, the incentive to buy liability insurance will be reduced even further than that. The crucial requirement for such an outcome is that there must be a considerable gap between the time when the limited liability entity makes a choice about its care level and the later point p. 396in time when the damage materializes. Practical examples of such long latency periods involve defective products and the operation of facilities for waste management and disposal. Perhaps the most striking example is a waste dump operated by a limited liability company. Assume that the operator faces a choice between a standard of high safety and another of low safety. The high standard requires safety measures in the amount of 40 per unit of waste; the low standard represents the minimum measures required by regulatory law, which cost 20 per unit of waste. If the high standard is observed, the harm caused after 20 years will be 0, whereas under the low standard, the harm will be 50 per unit of waste. Here, it would clearly be desirable for the operator to commit to the high standard, and incentives would lead the operator to do just that if the investment in the safety of the site and the damage payment were due simultaneously or within the same fiscal year. However, if there is a lapse of 20 years between the investment in safety and the payoff in terms of damages saved, matters are different. The shareholders of the operator might rationally adopt a looting strategy, that is, commit to the low standard of safety and let the company generate high profits which are paid out to shareholders over 20 years in order to let it fall into bankruptcy once the harm has materialized.

In the above example, shareholders have the chance of reaping additional profits to the tune of 20 per unit of waste by charging customers a price of 40, reflecting the costs of the high standard of safety, but then adopting the low standard which involves costs of only 20 per unit of waste. The prospect of pocketing large extra profits over a long period of time may well outweigh any losses incurred by the eventual demise of the corporation after the damage has materialized.

Within such a scenario, the shareholders have no incentive to insure their company against the future liabilities looming beyond the horizon. Doing so would diminish present profits to the benefit of tort victims whose claims would be left uncompensated without insurance.

In the context of corporate torts, a way of improving the situation by exposing shareholders of a limited liability company to unlimited liability vis-à-vis non-contractual creditors has been suggested (Hansmann and Kraakman, 1991, p. 1879). An alternative remedy would be to make directors liable for the torts committed by the enterprise they are running (Kraakmann, 1984, pp. 867 ff). It is difficult to decide which of these strategies would be more effective (cf. Leebron, 1991, pp. 1574 ff; Wagner, 2004, pp. 1054 ff). Although it may be doubted that shareholders of a public corporation can do much to monitor the behaviour of its agents (Shavell, 1987, p. 176), the only decision that counts in the present context is the one to take out insurance or not. In this respect, both shareholders and directors are in a position to exert considerable influence on corporate p. 397decisions. On the other hand, it is obvious that exposing shareholders or directors to personal liability for torts of the corporation would destroy some significant part of the benefits limited liability was invented for. Therefore, one might think of a third solution which would focus not on liability regimes but on the law of insolvency. It is suggested that tort and other non-voluntary creditors of an insolvent corporation should be accorded a kind of ‘superpriority’ with regard to the distribution of assets (Painter, 1984, pp. 1080 ff).

13.5.5 Limited liability and activity levels

Of course, the same reasoning that was just applied with regard to safety measures also applies to the level of activity. Again, the efficient mark will be missed if the agent can reap the benefits from the activity at the time he pursues it and postpone its costs in terms of damages into the future. In cases of long periods of latency tortfeasors will not do what efficiency requires, that is, compare the utility derived from the activity with total costs – including the costs of future liabilities. Rather, the agent will compare utility with present-day costs in order to engage in it whenever utility exceeds just these present-day costs. Therefore, there will be too much of the latently dangerous activity, and too much damage caused.

13.5.6 Compulsory insurance

There exist several legal instruments to counteract looting strategies and to ensure that future liabilities are anticipated in present-day decisions. One such instrument is a legal obligation to take out liability insurance (Jost, 1996, pp. 263 ff; Polborn, 1998, p. 141; Skogh, 2000, pp. 529 f). If liability insurance is mandatory, limited liability companies have no choice over whether to protect against future liabilities. As they are liable to pay premiums for the coverage of future tort claims today, they are forced to internalize the anticipated amount of damage payments that are likely to become due after the harm has materialized.

Against this reasoning one may object that it does not improve matters if insurers are brought into the game since the agent still lacks incentives to take care or to curb its level of activity because she has shifted the risk of liability onto the insurer (Shavell, 2000, p. 176). This argument is correct in pointing out the danger of moral hazard in scenarios with long latency periods from the time when the agent decides whether to engage in the activity and what level of safety to observe. Important as this aspect is, insurers have shown themselves to be aware of the problem, and have resorted to effective countermeasures (see Section 13.6.3). In the special area of environmental liability insurance, insurers have gone to great lengths to see to it that each risk is comprehensively analysed ex ante, and p. 398to monitor those insured closely lest they are confronted with a tail of long-term liabilities (Wagner, 2007, pp. 96 f).

Regrettably, some of the instruments employed by insurers to limit their own exposure have resulted in rolling back the risk to the limited liability company, as is the case with claims-made policies lacking an extension period. Under such a policy, the firm remains essentially uninsured against long-term liabilities and thus may still have an incentive to play the looting strategy.

13.5.7 Alternative means

Mandatory liability insurance is not the only means available in order to realign the incentives of an agent having the option to reap profits now and leave future damage claims to an administrator in bankruptcy. Alternative means are minimum asset requirements (cf. Shavell, 2005, pp. 63–72), the directing of tort claims at managers or shareholders of limited liability companies, exposing managers to criminal liability, and the regulation of the dangerous activity through administrative law. This is not the place to discuss the advantages and disadvantages of these instruments in comparison to compulsory insurance (for a theoretical analysis, cf. Shavell, 2005, pp. 63–72). Suffice it to say that the alternative instruments all have their limitations and all come at a certain cost. None of these instruments forces the agent to internalize the full costs of the activity in question, including the expected value of future losses. In contrast, the premiums paid to liability insurers, if calculated accurately, make the external effects of the activity visible ex ante such that the agent has the incentive to choose the level of activity that is desirable from a social point of view (Shavell, 2000, p. 176, 2004, p. 277).

As far as the disadvantages of alternative means are concerned, minimum asset requirements restrict entry into the market for the activity in question and lock out start-up firms with little capital from competition (Shavell, 2005, pp. 63, 64). The personal liability of managers, under both tort and criminal law, comes at the price that these managers will act in a risk-averse, rather than a risk-neutral, manner, and thus fail to maximize the value of the firm. The piercing of the corporate veil in order to make shareholders personally liable removes the shield of limited liability and threatens to destroy the benefits in terms of efficient investment for which limited liability was created in the first place (Easterbrook and Fischel, 1991, pp. 41 ff). Finally, public regulation of the activity is certainly indispensable and is, in fact, common with regard to dangerous activities. On the other hand, it would be naïve to expect too much of regulation or to think that government agencies are particularly good at foreseeing the future and at anticipating damages.

13.6 p. 399The limits of insurance: insurability

13.6.1 Uninsurability as an argument in political discourse

The concept of insurability is difficult to pin down. It plays a major role in political discourse, where it is used by the insurance industry as an argument against expansions of tort law into uncharted territory, like liability for environmental harm (Wagner, 2005a, p. 99). In the US, insurability is a concern counselling in favour of tort reform, for example, by restricting the discretion of juries to award damages in exorbitant amounts. In a broad sense, insurability denotes the fact that the insurance industry is confident of being able to manage the risk because it is foreseeable and calculable.

In the US, concerns about insurability are fuelled by the indeterminacy and unpredictability of outcomes under the American tort system. Although the bulk of cases are settled in a straightforward manner, which generates outcomes that are readily foreseeable, the system suffers from a small number of blockbuster awards, mostly rendered in one of the so-called hellhole jurisdictions located in the Gulf South. In recent years, both the US Supreme Court and legislators have addressed the problem. The US Supreme Court has developed a set of principles aimed at delineating the discretion of juries by imposing upper limits on the calculation of punitive damages (State Farm Mutual Automobile Insurance Co. v. Campbell, 538 US 408, 425 (2003)). Among those, the most important constraint is that the ratio between compensatory and punitive damages must not exceed a single-digit multiplier. In addition, some state legislatures have committed themselves to tort reform and introduced damage caps, in particular with regard to punitive damages and damages for non-pecuniary losses. Empirical research on the practical effects of these changes on the law of damages confirms that they benefited insurers in the sense that they reduced losses and made them more predictable (Viscusi, 2004, pp. 9–24; Born, Viscusi and Baker, 2006).

13.6.2 Elements of insurability

Insurability is also a term of art in insurance economics. As such, insurability is a multi-dimensional concept that defines the threshold requirements for a risk to be insurable (for a more detailed account, cf. Wagner, 2007, pp. 87 ff; for an even richer array of criteria, cf. Berliner, 1982, p. 13). In order to qualify for insurance, a risk must be:

  • accidental

  • determinable and measurable

  • independent

  • non-catastrophic

p. 400Within the context of liability insurance, accidentalness requires that the harm be contingent upon a move of nature and not the result of a deliberate choice by the insured. Therefore, claims for damages that were caused intentionally by the insured are excluded from the scope of coverage of liability insurance. Of course, the intentional infliction of harm on others is only the most extreme in a continuum of scenarios in which the actions of the insured determine the probability of harm. This raises the problem of moral hazard, discussed above (see Section 13.4), but does not render the risk uninsurable. If it were otherwise, liability insurance could not exist, at least not in the area of fault-based liability.

However, liability for fault is precisely the area where liability insurance developed and in which it has thrived until today.

The requirements that the risk be determinable and measurable come closest to the common use of insurability in political discourse. Insurers charge premiums in exchange for extending cover. The premium must reflect the expected value of the liabilities for which cover is granted. Therefore, the scope of potential liabilities must be anticipated by the insurer, the expected loss estimated and the premium calculated. Where the risk is indeterminate or immeasurable, rational calculation of the premium is impossible.

In order for the Law of Large Numbers to work (see Section, the risks within an insurance pool must be independent of one another. The fact that one risk has materialized must not increase the probability that a significant portion of the other risks will materialize as well. If all car owners in the world were involved in accidents simultaneously, liability insurers would be bankrupt within an instant. The system works only because such a scenario is virtually impossible. The requirement that the risk be non-catastrophic addresses the problem that the resources of an insurer might be exhausted by a single loss of exorbitant magnitude.

13.6.3 Insurability as a flexible concept

Insurability is anything but a hard concept allowing for black-and-white distinctions. It is all a matter of degree. The individual risks within a single pool may not be fully independent of one another; in most cases, the insured has some influence on the occurrence and the magnitude of losses; and measurability is usually achieved simply by extrapolating loss experience into the future on the counterfactual assumption that the future will be exactly like the past. Thus, the requirements of insurability are never met to their fullest degree. This does not cause practical problems as long as insurers can adjust the premium to reflect the added uncertainty. Where a risk is impossible to measure, the insurer might make a rough estimate and then add a sum that makes up for the remaining uncertainty. It has rightly p. 401been said, ‘a risk can be insured also when no statistics is available, and even when no theoretical analysis seems possible’ (Borch, 1990, p. 316).

A famous example of a seemingly uninsurable risk that was nonetheless insured goes back to the times when jet-powered aeroplanes were introduced into commercial air service. When the first of this new type of aircraft was taking up service, insurers set the premium for hull insurance at 8 percent of the value (Borch, 1990, pp. 315 f). Administrative costs aside, with such a premium no fewer than eight out of 100 planes could have crashed every year without the insurer losing money. This was a safe bet for the insurance industry because the flying public would never have tolerated such a high loss ratio. As this example illustrates, insurability is not only a matter of degree, it is first and foremost a matter of price. Insurability places limits on the extent to which a particular risk may be covered and it influences the premium charged. As the history of Lloyd’s but also the experience of other insurers illustrate, virtually anything can be insured, if only for ‘suitable prices’ (Arrow, 1971, p. 141).

13.7 The impact of liability insurance on tort law

It is a widespread belief that liability insurance is a major force driving the development of tort law. This is obviously true in the sense that most of the funds collected by tort victims come out of the pockets of insurance carriers. Most individuals would be unable to pay up major claims for damages brought against them (see Section If tort law functions at all as a mechanism for compensating victims it does so to a large extent because insurance is available and in fact widespread (Lewis, 2005, pp. 47 ff; Baker, 2005b, 295 ff).

Since the seminal work of Calabresi, scholars of economic analysis of law have gone further than that and developed risk-spreading into an autonomous goal of tort law in its own right (Calabresi, 1970, pp. 39 ff). Given that most individuals and even many firms are risk-averse (see Section, the spreading and diversifying of risk confers benefits on actors who would otherwise have to live with a risk they would rather like to get rid of. Insurance is a common strategy of risk-spreading. Therefore, the question arises whether liability rules should be tailored in such a way as to attach liability to the party who is in the best position to insure the loss.

A famous decision that deviated from the common restraint of judges in embracing the insurance function of tort law is that of the California Supreme Court in the product liability case of Escola v. Coca-Cola Bottling Co. of Fresno. In the words of Justice Traynor ((1944) 150 P.2d 436, 441): ‘The cost of an injury and the loss of time or health may be an overwhelming misfortune to the person injured, and a needless one, for the risk of p. 402injury can be insured by the manufacturer and distributed among the public as a cost of doing business.’ This judgment is one of the milestones in the development of strict products liability and one of the rare instances when a court has openly embraced the risk-spreading function of liability. In most European jurisdictions, strict liability is the domain of the government rather than the courts (Wagner, 2003, pp. 274 ff, 2006, pp. 1030 ff). In general, lawmakers are on the one hand concerned with victim compensation and on the other they pay much attention to issues of insurability in order to protect the tortfeasor (Faure, 2005, pp. 248 ff). Loss spreading as such plays a minor role, at best.

As far as the application and development of fault-based liability is concerned, it is lawyer’s lore that courts do take the insurance question into account even though they often pay lip-service to the principle that liability comes first and must be determined without keeping an eye on the insurance cover that may or may not be available in the case at hand (Wagner, 2005b, pp. 323 ff). It is very difficult to verify what is really going on because it is impossible to look into the minds of judges applying tort law rules to particular cases. Outside special areas like liability in equity, the evidence does not support the assumption that courts follow the policy of risk spreading via insurance. This is a healthy approach (Trebilcock, 1988, pp. 246 ff; Faure, 2005, pp. 260 ff; Wagner, 2005b, pp. 342 ff). If fault-based liability were contingent on whether the potential tortfeasor was insured or not, the incentives to buy market insurance would be distorted. In addition, one would also have to consider whether the victim enjoyed the benefit of first-party insurance for the particular loss. Liability rules would thus have to be tailored to fill up whatever interstices are left after the parties have resorted to the insurance markets. Finally, if market insurance played a role, there is no reason to decide otherwise with respect to self-insurance. Precisely because the potential to self-insure increases with wealth, the objective of risk-spreading would come close to a form of deep-pocket liability (Trebilcock, 1988, pp. 258 f).

For an overview of the state of the law and the discussion in a number of European jurisdictions cf. Wagner (2005a).

This statement is in contrast to prospect theory which established, with the help of experiments in the laboratory, that individuals are risk-averse with respect to gains but risk-preferring with respect to losses; cf. Kahneman and Tversky (1984). Therefore, the example given above may not be borne out in reality as it may be that the lottery with the 50 percent chance of a loss of the order of 2,000 may be preferred over the certain loss of 1,000. However, if the certain loss is 10 and the lottery involves a 0.05 percent chance of losing 2,000, the certain outcome will be preferred.