Ivadelle and Theodore Johnson Professor of Law and Business
University of Southern California Law School
© Copyright 1999 Jennifer Arlen
This chapter surveys the literature on tort damages, including damages for
serious physical injury and death. Section A presents the basic analysis of tort
damages for injuries to replaceable commodities, focusing on the relationship
between liability rules and optimal damage rules. Section B discusses damages
for physical injury and death, examining, among other issues, the deterrence
and insurance values of life. Section C examines special topics, such as optimal
liability when corporations are defendants, whether recovery should depend on
defendants’ wealth, economic and nonpecuniary losses, and the calculation of
damages. An exhaustive bibliography is attached at the end.
JEL classification: K13
Keywords: Tort Damages, Pain and Suffering, Tort Compensation, Wrongful
Death, Physical Injury, Nonpecuniary Losses, Value of Life
Efficient damages awards are critical to the optimal functioning of the tort
system. A rich literature now exists on the subject of efficient damage awards.
This analysis has shown that no one damage rule is optimal in every situation,
and that, contrary to common wisdom, damage awards need not necessarily
fully compensate victims. Rather, the optimal damage award depends on: (1)
the nature of the injury (harm to replaceable versus irreplaceable commodities);
(2) the relationship of the parties and type of risk (unilateral, bilateral or market
risks); (3) the liability rule (strict liability or negligence); (4) whether liability
is individual or vicarious and (5) any existing imperfections, such as
information costs, uncertainty, and judgment-proof problems. This analysis
reveals that current rules governing recovery for replaceable losses may be
optimal, but those governing recovery for death and physical injury generally
A. Replaceable Goods: Basic Analysis
This section considers efficient tort damages for purely replaceable
commodities. A commodity is replaceable if its owner believes that equivalent
commodities are available on the market (Cook and Graham, 1977, pp.
144-146). A person can be fully compensated for the loss of a replaceable
commodity if he is paid the market price of the good since he can use this
money to buy a perfect substitute. Items such as cars, clothing and money
generally are replaceable.
2. Efficiency Criteria and Initial Endowments
Although the essential goal of economics is to maximize social welfare, law
and economics scholars employ a number of different efficiency criteria to
determine when this goal is satisfied. Some employ the Pareto criterion: a state
of the world is Pareto efficient if no one’s utility can be increased without
decreasing someone else’s utility. Others employ social utility maximization:
under this criterion a tort rule is efficient if it maximizes the total utility of
society, even if some people are made worse off by the change. Finally, others
employ the Kaldor-Hicks criterion under which a change is an improvement if
those who benefit from the change could compensate the losers monetarily for
their losses and still come out ahead. A state of the world is Kaldor-Hicks
efficient when there is no way to increase anyone’s utility in such a way that the
winners could compensate the losers.
Under any efficiency criterion, tort liability must minimize the joint costs
of accidents and accident prevention. This in turn implies that tort liability and
damage rules must induce optimal care-taking (defined as the efforts
individuals undertake to avoid an accident given that they are engaged in the
risky activity) and induce optimal activity levels (defined as the frequency with
which individuals engage in the risky activity). In addition, a tort regime
should induce optimal risk-spreading by any risk-averse parties by ensuring
that they are optimally insured against risk, either through the insurance system
or the tort system, whichever is most effective.
The Pareto criterion imposes an additional requirement: to be Pareto
efficient injurers’ risk-taking must make no one worse off than they would be
otherwise, in the status quo ante. This requires that damage awards ensure that
potential victims are left no worse off than they would have been ‘otherwise’.
What that status quo ante is depends on the initial allocation of entitlements
(see discussion below).
3. Optimal Deterrence Damages Under Strict Liability
Under the traditional economic model of accidents, the efficient tort damage
rule for injury to replaceable commodities is simple: fully compensate victims
for their losses. Specifically, victims should be awarded the market price of the
Under a strict liability regime, this damage rule will optimally deter
wrongdoing, ensure efficient risk-spreading by victims, and - depending on the
liability rule and initial endowments - may satisfy the requirements for Pareto
efficiency that no one be left no worse off than he would be otherwise. Risk
averse injurers will not optimally spread their risk, however, unless they are
able to purchase first-party insurance (Shavell, 1980, 1987).
In the standard economic model of accidents, injurers impose risk on
victims who are ‘strangers’ to them (that is, not in a market relationship with
them) (Brown, 1973). The injurer can reduce the risk imposed on the victim
both by taking ‘care’ in the manner in which he undertakes his activity and also
by controlling the frequency with which he engages in the risky activity
(Shavell, 1980). For simplicity, this section will assume that all parties are risk
neutral (see Section 6, discussing risk-spreading). Initially, we will take the
social goal to be maximizing total social welfare which is the benefit of the
injurers’ activity minus total accident costs (defined as cost of care plus
expected accident losses). Injurers’ behaviour is efficient when they employ the
level of care and engage in the activity at the frequency that maximizes social
welfare (Shavell, 1987, p. 21).
Consider now what damage rule induces optimal deterrence when injurers
are held strictly liable for any harms they cause. Under a strict liability rule,
injurers will take the level of care that minimizes the joint social costs of
care-taking and accidents if the expected cost to them of risk-taking - for
example, their expected liability - equals the cost to society of their risk-taking.
This implies that damages should equal the cost to society of the accident,
which implies victims shall recover their losses (Brown, 1973; Shavell, 1980).
This damage rule also generally will induce injurers to engage in the optimal
amount of the activity because the injurers’ cost of engaging in another unit of
the activity - the resulting increase in his expect liability - will equal the cost
to society of his decision - the resulting increase in expected accident costs
(Shavell, 1980). Yet this rule will not induce optimal activity levels if each
injurer’s probability of being in an accident depends on the number of people
engaged in the activity. In this case, even under strict liability, injurers will
over-engage in the activity because they will not take into account the effect of
their actions on other injurers’ probabilities of being in an accident (Hindley
and Bishop, 1983, pp. 60-61).
When potential victims’ care levels and activity levels affect expected
accident costs, tort liability rules also must induce victims to take optimal care
and engage in the efficient amount of the activity. Strict liability with full
compensation damage awards may undermine this effort, however. When
victims are fully compensated for their injuries, a pure strict liability rule will
not induce victims to take either efficient care or efficient activity levels because
victims do not bear the cost of any injuries they suffer; thus they have no reason
to spend resources to avoid the harm (Shavell, 1980). Victims will take due
care, however, when strict liability is supplemented with the defense of
contributory negligence, because in this situation victims bear the full cost of
any failure to take due care. Nevertheless, they still will not engage in the
efficient amount of the activity because a victim who has taken due care will be
fully compensated for any injuries he suffers (Ibid.).
Under the Pareto criterion victims often must be fully compensated for their
losses in order to ensure that the injurers’ risk-taking leaves potential victims
no worse off than they would be otherwise. If victims possess the initial
entitlement to the property in question, then they must be fully compensated in
order to ensure that they are not made any worse off (see Section 3 which
discusses the entitlement issue).
4. Optimal Deterrence Damages Under Negligence Liability
Under a negligence rule, full ex post compensation of purely pecuniary losses
is an optimal damage rule, although it is not the minimum optimal rule. Full
compensation for property losses will induce injurers to take optimal care under
a negligence rule, provided courts correctly set the standard of care (Brown,
1973; Shavell, 1980). However, a lesser award also may suffice. Cooter has
shown that negligence liability can induce injurers to take due care using
less-than-full-compensation damage awards because an injurer’s expected
liability increases dramatically from zero to the expected damage award if he
reduces his care-taking from due care to less than due care. This dramatic
increase provides a strong incentive to take due care (Cooter, 1984). This result
does not hold, however, if the application of ‘but for’ causation effectively
eliminates the discontinuity in the injurer’s expected liability function. In this
case, injurers will not take due care unless damages fully compensate victims
(Grady, 1983; Kahan, 1989). In all events, under a perfectly functioning
negligence rule, injurers will take due care if damages fully compensated
victims forcing injurers to bear the full costs of their risk-taking (Shavell, 1980;
Cooter, 1984). Negligence liability will not induce injurers to engage in
efficient activity levels, however, because non-negligent injurers are not liable
and thus do not bear the social costs of their activity level choices. Their
activity levels therefore will be too high (Shavell, 1980).
Negligence liability will induce victims to take efficient care and engage in
efficient activity levels, however. Under this regime, injurers will take due care
and thus will not be liable for the risks they impose. Victims, therefore, will
bear the full cost of their injuries and will have the requisite incentive to take
due care and reduce their activities to the optimal level (Shavell, 1980).
Negligence liability thus will optimally deter to the extent it can, when
damage awards fully compensate victims for their losses. Yet the question
remains whether this regime is Pareto efficient. Whether this regime satisfies
the Pareto criterion depends on the initial allocation of entitlements.
Negligence liability implicitly grants to the injurer the right to impose
reasonable risks on the victim, yet entitles the victim to freedom from harms
from unreasonable risks. When this describes the victim’s initial endowment,
then forcing injurers to fully compensate victims for injuries resulting from
unreasonable risks satisfies the requirement for Pareto efficiency that victims
be no worse off than they would be in the status quo ante. Negligence liability
is not Pareto efficient, however, if the victim’s initial endowment is to be free
from any risk of harm created by the injurer. In this situation, victims are made
worse off by a rule that allows injurers to impose reasonable risks on victims
without compensating them and only forces injurers to fully compensate victims
for any unreasonable risks they impose (Arlen, 1985; see Calabresi and
Melamed, 1972, discussing the intersection of entitlements and liability rules)).
5. Comparison with Existing Law
The preceding analysis suggests that in the case of injury to replaceable
commodities, damage awards generally should fully compensate victims for
their losses. Thus tort law should force injurers to pay for any damage they
cause. This implies that the ‘thin skull plaintiff rule’ - which holds injurers
liable for a victim’s injuries even if the victim is unusually sensitive - is
efficient because it forces the injurer to take into account all expected costs of
his care-taking and activity levels. By contrast, the use of the proximate cause
doctrine to eliminate liability for foreseeable but excessive losses is not efficient
(Shavell, 1987, pp. 128-131).
To induce efficient behavior, however, ‘full compensation’ should be based
on the optimal loss caused by the injurer. After an accident occurs victims often
can take steps to mitigate the harm - for example, by seeking medical treatment
for an injury. To encourage optimal mitigation, victims’ recovery should be
limited to optimally mitigated losses plus mitigation costs (Shavell, 1987, pp.
144-146; Wittman, 1981). Recovery also should be reduced - or set off - by
those amounts the victim would have spent anyway had the loss not occurred
(Shavell, 1987, p. 140).
6. Efficient Risk Spreading
The preceding analysis assumes that injurers and victims are risk neutral - that
they care only about the expected value of losses and not their magnitude. This
assumption may accurately describes many corporations. Thus, injurers in
products liability cases may be risk neutral (but see Arlen, 1992b, pp. 419-420,
discussing when corporations should be treated as being risk averse).
Individuals, however, generally are risk averse: most individuals would prefer
the certain loss of $10 to a 10 percent chance of losing $100, even though the
expected value of these losses is the same. A risk-averse individual facing a
potential loss, thus, would be better off if he could purchase actuarially fair
insurance that fully compensates him for the loss (see generally Shavell, 1987,
6.1 Insurance is Available
If both injurers and victims are risk averse and both parties can buy actuarially
fair first- and third-party insurance, then people will optimally spread risk by
purchasing insurance; tort liability rules need only be concerned with optimal
deterrence (Shavell, 1987, pp. 210-212). Indeed, in this situation tort liability
probably should not be used as an insurance mechanism: the administrative
costs of tort liability far exceed the administrative costs of either first- or
third-party insurance (Weiler, 1993b, pp. 926-927). Tort liability should be
imposed to optimally deter risk-taking, however. In this situation, full
compensation damages are optimal. This conclusion holds even if there is
moral hazard in the sale of liability insurance because insurers cannot observe
the injurers’ precautions (Shavell, 1982; Shavell, 1987, pp. 211-212).
6.2. Insurance is not Available
By contrast, if insurance is not available, then tort rules should not only induce
optimal deterrence but also optimal risk-spreading by risk-averse parties. In the
case of replaceable losses, the optimal amount of insurance is full compensation
(Shavell, 1987, Ch. 8; see also Chapter 5700 on Insurance Regulation). Thus,
victims’ risk-spreading will be optimal if they are fully compensated for any
losses they may suffer. This implies that a full compensation damage rule will
ensure that victims optimally spread risks when the injurers’ activity is
governed by a rule of strict liability (Shavell, 1982; Shavell, 1987, pp.
Victims’ risk-spreading will not be optimal under a negligence regime,
however, because a perfectly functioning negligence regime will induce injurers
to take due care, in which case victims will not be compensated for their losses
(Shavell, 1982; Shavell, 1987, pp. 210-227). If the regime does not function
perfectly, however, and injurers are deemed to be negligent (for example,
because of court error or principals’ inability to induce their agents to take due
care), then a full compensation damage rule will induce efficient risk-spreading
by victims (Shavell, 1982).
If insurance is not available and injurers also are risk averse, then the very
damage and liability rules that yield efficient risk-spreading by victims will
produce inefficient risk-spreading by injurers, and vice versa. Strict liability
forces injurers to bear any injuries they cause; by contrast, under a negligence
rule injurers’ risk-spreading is efficient because they can avoid all losses by
taking due care (Shavell, 1987, pp. 208-209).
Moreover, the presence of risk-averse injurers who cannot purchase
insurance also alters optimal damage rules. To induce due care and optimal
activity levels under a strict liability rule when injurers are risk averse and
uninsured, injurers must be liable for an amount that is less than the victims’
losses. This is because the injurer will view both the loss itself and the risk as
a cost. Yet, if damage awards do not fully compensate victims, then victims’
risk-spreading will not be optimal. A full compensation damage rule will
induce injurers to take due care when they are subject to a negligence liability
rule, however, because they can avoid all risk by taking due care. As previously
explained, however, neither injurers’ activity levels nor victims’ risk-spreading
will be optimal under such a regime (Shavell, 1982; Shavell, 1987, pp.
7. Accidents in a Market Setting
When victims and injurers are in a consensual or market relationship and
victims are fully informed about any risks imposed on them by injurers, market
forces will ensure that injurers take optimal care and engage in the optimal
level of the activity. This is because victims will treat the risk of harm as a cost
of buying the product (or service), thus giving the potential injurer the incentive
to minimize the joint cost of accident losses and his care-taking. Activity levels
will be efficient because victims will only buy the commodity if the benefit to
them exceeds the total cost, including expected accident costs (Polinsky, 1980;
Shavell, 1980; Shavell, 1987, Ch. 3; Spence, 1977). In this situation, therefore,
the only function for the tort system, if any, is to induce optimal risk-spreading
in those situations where individuals cannot purchase insurance. As explained
above, when it is appropriate to use the tort system to provide insurance to
victims, then damages should fully compensate victims for their losses (see
Epstein, 1985 and Chapter 5140 on Products Liability for a complete discussion
of the use of the tort system to provide insurance to victims).
Victims are not necessarily fully informed about risks, however. If victims
do not accurately estimate the risk of loss, then market forces alone may not
induce optimal care-taking and activity levels (Spence, 1977; see A. Schwartz,
1988; Shavell, 1987, Ch. 3). Consumers may fail to accurately estimate risks
as a result of either imperfect information or cognitive ‘imperfections’ that
cause people’s attitudes towards risk to deviate from that predicted by expected
utility theory. For example, evidence suggests that consumers are likely to
underestimate hidden product risks; they also tend to underestimate known but
large risks, such as the risk of dying of cancer, and risks that are substantially
within the individual’s control (Viscusi, 1991, pp. 64-65).
When consumers underestimate risks, injurers will take insufficient care
and engage in too much of the activity, unless they are liable for the injuries
they cause or unless the problem can be corrected with warranties (Polinsky and
Rogerson, 1983; Shavell, 1980; Shavell, 1987, pp. 52-53; Spence, 1977; see
Schwartz and Wilde, 1985). Producers will not necessarily respond to markets
with imperfectly informed consumers by offering full warranties, and the
availability of warranties will not necessarily induce optimal product quality
(Cooper and Ross, 1985; Grossman, 1981; Polinsky and Rogerson, 1983; Salop,
1977; Schwartz and Wilde, 1985; see Priest, 1981).
Tort liability may be able to induce optimal care-taking and activity levels
in those circumstances when market forces alone cannot do so. When
consumers underestimate risks, injurers can be induced to take due care and
engage in optimal activity levels if they are held strictly liable for consumers
losses and subject to a full compensation damage rule. When injurers bear all
expected accident costs, they will directly bear the full cost to society of their
products. Thus, in their effort to minimize their own costs, they also will select
the care levels that minimize social costs. Product prices thus will equal the
social cost of producing the goods, provided product markets are competitive.
Consumers, therefore, will purchase the optimal amount of the product, leading
to optimal activity levels (Shavell, 1987, pp. 67-68; see Polinsky and Rogerson,
1983, modifying this conclusion when producers have market power). (For a
more complete discussion of this topic, see Chapter 5140.)
Of course, consumers do not necessarily underestimate all risks. Indeed,
evidence suggests consumers overestimate small but known risks that have
received substantial publicity, as well as risks that are vivid or dramatic (such
as the risk of dying in a plane crash) (Viscusi, 1991, pp. 64-65; see Arlen,
1998). In this situation, market focus may cause injurers to take excessive care
and engage in too little of the activity. Tort liability will only exacerbate this
problem (Viscusi, 1991, pp. 64-65; see Arlen, 1998; A. Schwartz, 1988).
8. Automobile Accidents and other Bilateral Risks
The preceding analysis assumes that individuals are either injurers and victims
but not both - in other words that accidents result from ‘unilateral risk’
activities. Many common accidents, however, result from ‘bilateral risk’
activities - situations where both potential parties to the accident risk being
injured should an accident occur. For example, automobile collisions, the
largest single source of tort cases, are bilateral risk accidents.
In the bilateral risk context, a full compensation damage award will induce
optimal care-taking by both parties to the accident if the activity is governed by
a duty-based liability rule: such as negligence (with or without contributory
negligence) or strict liability with contributory negligence. Each party will take
due care because he is best off taking due care whether or not the other takes
due care (Arlen, 1990a, 1990b, 1992a, compare with Diamond, 1974, p. 117,
concluding that pure negligence is not efficient).
In fact, duty-based liability rules can induce efficient care-taking employing
a less-than-full-compensation damage rule. People will take efficient care
provided that damage awards are sufficiently large that each is better off taking
due care and getting the resulting reduction in liability than he is not taking
due care and paying the additional cost. The damage award that induces
efficient care-taking is less than the amount that fully compensates victims for
their losses (Arlen, 1990a, 1990b, 1992a).
When people are risk averse, the three duty-based liability rules will not
induce optimal risk-spreading, however, even when combined with a full
compensation damage award. Under these regimes, each party will take due
care. Thus, under negligence (with and without contributory negligence) each
party will take due care and thus will bear his own losses. Accordingly,
negligence is consistent with efficient risk spreading only if parties can
purchase first-party insurance. Under strict liability with contributory
negligence each party will take due care and thus will bear the other parties’
losses but not his own. Thus this regime is consistent with efficient
risk-spreading only if parties can purchase third-party insurance (Arlen,
Regardless of the damage rule, however, pure strict liability will not induce
efficient care-taking. To see why, assume both people suffer the same injury
whenever an accident occurs. In this case, each person’s expected damages
equals his expected recovery. Thus, pure strict liability is equivalent to a no
liability regime: neither is efficient because neither risk-imposer bears the cost
to the other person of his risk-taking (Arlen, 1990a, 1990b, 1992a).
9. Factors Affecting Liability
Of course, even in the standard ‘unilateral risk’ case, determining the optimal
damage award may be more complicated than it might at first appear. This part
considers several factors that may affect the magnitude of liability, focusing on
the traditional model of accidents in which parties to the accident are strangers
and are either potential victims or injurers, but not both.
9.1 Litigation Costs
The standard analysis of tort law assumes that there are no litigation costs. Yet
litigation costs are substantial. This raises the issue of whether damages should
equal the victims’ actual losses plus their litigation costs.
(a) Pareto Criterion Under the Pareto criterion, damage rules must include the
victims’ litigation costs because a damage rule that limits compensation to
harm suffered without including the victims’ litigation costs does not leave the
victim no worse off than he would be otherwise, at least in those cases where
the victim originally possessed the entitlement the injured property.
(b) Strict Liability Under strict liability injurers should compensate victims for
their litigation costs in some circumstances but not in others. According to
Polinsky and Rubinfeld (1988) optimal deterrence requires that injurers
compensate victims for their litigation costs. Injurers must compensate victims
for two reasons. First, if victims bear their own litigation costs, some victims
with good claims will not find it worthwhile to sue because litigation costs
exceed their expected recovery. Thus, injurers will not bear the full expected
cost of the harms they impose because not all victims will sue. Second, even if
all victims sue, injurers will not bear the full social cost of their actions
because, properly defined, the social cost of the injurers’ risk-taking is the sum
of the victims’ losses and the cost of litigation. Thus, under a rule of strict
liability with compensatory damages injurers will not take efficient care or
engage in efficient activity levels. Litigation costs also will be excessive
(Polinsky and Rubinfeld, 1988).
Yet, Polinsky and Rubinfeld argue, deterrence is not the only objective of
the tort system. A full analysis reveals that in some circumstances victims
should not be compensated for their litigation costs. In addition to inducing due
care, an optimal regime must minimize litigation costs. Whether compensatory
damages should be adjusted upwards or downwards depends on the effect of
changes in damages on the injurers’ incentives to take care and the victims’
incentives to sue. Specifically, if the injurers’ care-taking affects not only the
probability of an accident but the magnitude of the harm, it may be optimal to
enable the victim to recover some but not all of his litigation costs: enough so
that it is worthwhile for the victim to sue when an injurer takes less than
optimal care and the victim’s harm is large, but not enough so that it is
worthwhile for a victim to sue if the injurer takes optimal care, thereby
reducing the amount of harm suffered (and thus the amount of recover)
(Polinsky and Rubinfeld, 1988). This would induce optimal care-taking and
minimize on litigation costs by effectively transforming the strict liability rule
into a negligence rule. As with any negligence rule, such a regime would not
induce injurers to undertake the optimal amount of the activity.
Victims should recover less than their compensatory damages, Polinsky and
Rubinfeld argue, in those situations where injurers’ care-taking is not
productive, and thus optimal care is to take none at all. In this case, a strict
liability regime is optimal if it reduces the victims’ recovery to ensure they will
not sue (Polinsky and Rubinfeld, 1988). Again, this will not produce optimal
activity levels, however.
(c) Negligence Liability Under a negligence rule, a damage rule limiting
defendants’ liability to victims’ losses will either result in optimal care or too
little care. Thus, when needed, any adjustment to compensatory damages
should always be positive (Polinsky and Rubinfeld, 1988). Under a negligence
rule, injurers will invariably take due care if victims’ litigation costs are not so
high as to discourage suit. This results in first-best efficiency because once
injurers take due care, victims have no reason to sue. If, however, litigation
costs are high enough to discourage many suits, injurers will not take due care.
This also will lead to excessive litigation costs if some victims nevertheless find
it worthwhile to sue. Increasing damages to ensure that victims sue will induce
optimal care-taking and minimize litigation costs, since once injurers take due
care victims no longer have a reason to sue (Polinsky and Rubinfeld, 1988, pp.
(d) Fees Imposed on Plaintiffs and Defendants The previous analysis assumes
that the state cannot impose fees on either plaintiffs or defendants. Shavell
(1996, 1997) shows that once one allows for such fees, optimal damages
imposed on defendants equals the harm to the plaintiff plus the plaintiff’s
litigation costs plus the state’s litigation costs. This ensures that each injurer’s
total expected costs equals the total cost to society of her risk-taking activities.
In order to ensure that plaintiffs have the optimal incentive to sue, the state can
impose fees for bringing a suit when plaintiffs would otherwise have excessive
incentives to sue or subsidies when plaintiff’s incentives are inadequate.
Shavell argues that a plaintiff’s incentives are excessive if the social benefit of
the suit - measured by its deterrence impact, not the plaintiff’s injury - are less
than the costs of bringing the suit (Shavell, 1997, pp. 575-581, 584-585).
Shavell explains that this system is superior to one in which plaintiffs must
pay the state’s litigation costs. He notes that in this situation one should not
always force plaintiffs to pay the state’s costs because often a plaintiff’s suit
provides an external benefit - deterrence - that exceeds his private benefit.
Thus, even when the fee is not imposed plaintiff may not have sufficient
incentives to sue since he does not reap the full gains from his suit (Shavell,
1996, p. 587, 1997). Similarly, Shavell rejects a simple loser-pays fee shifting
arrangement noting that under such a regime victims with good suits will
always sue. Yet whether the victim will prevail does not determine whether,
from society’s standpoint, the suit should be brought. This, he argues, depends
on the deterrence impact of the suit.
Shavell’s analysis raises an interesting question about how to measure the
deterrence value of a suit. Shavell argues that deterrence value should not be
measured ex post. Rather liability will not have a deterrence value if, ex ante,
the injurer does not expect to cause this type of accident (Shavell, 1987, pp.
129-130). Such a suit would have deterrence value, however, if the finding of
liability - separate from the occurrence of the accident - would be likely to
educate other potential injurers (or victims) about the risk (see Arlen, 1993, p.
1124, n. 110; Rubin, 1993, p. 50).
9.2 Uncertainty and Court Error
Optimal damage awards also are affected by whether the legal process is
plagued by error.
(a) Court Error in Determining Negligence The possibility of court error in
determining whether the injurer is liable under a negligence rule also may
affect the optimal damage rule. Specifically, if courts are likely to err in
determining whether the injurer took due care, negligence liability rule with
full compensation damages will not necessarily induce injurers to take due care.
Even when courts are likely to err in either direction - to find a non-negligent
injurer liable and a negligent injurer not liable - under plausible conditions,
care-taking will be excessive if damage awards fully compensate victims for
their losses. When courts err, the cost to the injurer in deciding to take due care
rather than excessive care is significant - the resulting increased likelihood of
being found negligent. This will exceed the additional cost to her of taking
excessive care, under plausible assumptions. In contrast, the possibility of error
in her favor will not induce her to take less than due care. Doing so only saves
her the additional cost of taking due care but subjects her to a significant risk
of being found liable when she otherwise would not have been. Thus, even
when courts are as likely to err in the injurer’s favor and against her, the cost
of an error going against the injurer will exceed the benefit of an erroneous
finding in her favor. Thus injurers will respond to uncertainty by taking
excessive care - provided that there is a positive probability of the court
underestimating injurers’ care levels and the distribution of errors is not too
dispersed (Calfee and Craswell, 1984; Craswell and Calfee, 1986; Shavell,
1987, pp. 79-83, 93-97; compare Diamond, 1974).
This problem can be addressed in a couple of ways. The state can retain
negligence liability but implement a less-than-full compensation damage
awards (Calfee and Craswell, 1984) or it can switch to a strict liability regime
(if courts can correctly determine damage awards) (Cooter, 1984).
(b) Role of Causation Rules Uncertainty in the determination of negligence will
not induce excessive care-taking if on average courts get the due care standard
right and causation rules function perfectly, thus eliminating the discontinuity
in the injurer’s expected liability. Causation rules may eliminate this
discontinuity because a negligent injurer is liable only if his negligence caused
the harm. Thus, if courts err in determining due care but not in assessing
causation, an injurer falsely judged to be negligent still will not be liable
because the court will recognize that it is not more likely then not that the
accident would not have occurred but for the injurer’s negligence (Grady, 1983;
Kahan, 1989, pp. 437-439).
Should causation rules eliminate the ‘discontinuity’ produced by negligence
liability, then the risk of court error in determining whether an injurer is
negligent generally will result in injurers taking less than due care - not
excessive care - because uncertainty over care-taking may lead courts to hold,
incorrectly, that negligent injurers took due care, whereas correct application
of causation rules will ensure that non-negligent injurers incorrectly found to
be negligent will escape liability (Kahan, 1989, pp. 437-439). In this situation,
inducing efficient care may require super-compensatory damage awards.
(c) Error in Determining Plaintiffs’ Losses Courts also may err in assessing
plaintiffs’ losses. Cooter has argued that when courts are subject to this type of
error the solution is to employ a negligence liability rule. He argues that
injurers’ risk-taking is less sensitive to errors in determining the damage award
under a negligence regime than under strict liability because under a negligence
rule the injurer’s expected liability increases dramatically if she fails to take due
care. Cooter argues that this implies that negligence is the superior rule when
courts cannot accurately determine damages (Cooter, 1984).
Shavell argues that possible court error in the calculation of losses does not
necessarily imply that strict liability will be inefficient. Strict liability will
provide correct incentives if courts’ estimates of losses are correct on average
(Shavell, 1987, pp. 131-132).
Moreover, even when courts err on average, Kaplow and Shavell (1996)
argue that negligence is not superior to strict liability. Kaplow and Shavell
show that, in order to determine due care, courts must be able to calculate
victims’ losses. Thus, if courts cannot accurately determine damage awards
under a strict liability rule, they also will be unable to determine due care under
a negligence rule, and injurers’ care-taking under a negligence regime will not
be optimal. It has been suggested that due care might nevertheless be easier to
determine because courts can rely on custom and other sources of information
on due care. Yet in order to determine whether the custom is efficient courts
still will need to be able to estimate harm (Kaplow and Shavell, 1996; compare
with Arlen and Kraakman, 1997, discussing strict versus duty-based vicarious
Thus, any difficulty the court faces in assessing damages should afflict
negligence liability as well as strict liability (Kaplow and Shavell, 1996). The
choice between the regimes, therefore, will likely depend on relative
administrative costs. Kaplow and Shavell argue that since determining due care
requires more information than determining damages, the risk of error and
administrative costs are higher under a negligence regime and thus strict
liability is superior. Nevertheless, negligence may produce lower administrative
costs because fewer suits will be brought (see generally Shavell, 1987, pp.
9.3 Injurers’ Ability to Escape Liability for Losses
Other imperfections also can affect optimal damage awards. For example, if
injurers are not necessarily liable for their wrongs - for example, because torts
go undetected - full compensation damage awards will not optimally deter
Empirical evidence suggests that injurers are not liable for many of the torts
they cause. For example, empirical evidence suggests that a tiny portion (less
than 6 percent) of victims of medical negligence will obtain compensation
(Danzon, 1985, pp. 22-25; Weiler et al. 1993a, pp. 74-75). Indeed, even the
probability of detecting the source of a substantial oil spill is only 60 percent
When wrongdoers may escape liability, optimal deterrence of a risk-neutral
tortfeasor appears to require a damage award equal to the victim’s loss divided
by the injurer’s probability of detection (specifically, the injurer’s probability
of being formal liable) (h/p) because this rule ensures that an injurer’s expected
liability equals the social cost of the harm (Becker, 1968; Cooter, 1989b;
Landes and Posner, 1981; Polinsky and Shavell, 1998; see Arlen, 1994). This
could be implemented using a regime of compensatory damages coupled with
punitive damages or compensatory tort liability coupled with
government-imposed civil or criminal penalties.
This damage rule is not necessarily optimal, however. As previously
explained, under a negligence regime expected damages need not equal
victims’ losses if the injurer’s expected liability function increases substantially
(and discontinuously) if he takes less than due care. Moreover, even under strict
liability reasons exist why the damage multiplier should not necessarily equal
one over the probability of detection (see, for example, Becker, 1968, p. 178;
Craswell, 1996; Karpoff and Lott, 1993; Kaplow, 1992). (See Chapter 3700 on
Punitive Damages for a more detailed discussion of this issue.) Finally, this rule
must be modified for corporate defendants held liable for torts (or crimes)
committed by their agents (Arlen, 1994; Arlen and Kraakman, 1997; see Part
C for a discussion on vicarious liability).
9.4 Intentional Torts
Optimal compensation for intentional torts also may exceed the victim’s actual
losses even if injurers are invariably found liable. Intentional torts generally are
those which require the injurer to expend resources in order to commit. It
would appear that holding injurers fully and strictly liable for such torts would
optimally deter them. But, as Landes and Posner initially observed, this
assumes that there are ‘optimal’ intentional torts. Many intentional torts
involve low transactions costs; thus, market transactions are possible. All else
equal, market transactions are preferable to the involuntary taking of another’s
property, particularly once one considers victims’ costs of avoiding non-market
takings. To force such transactions into the market, expected damages should
exceed the injurer’s gain from the transaction. This may exceed the victims’
losses (Landes and Posner, 1987, pp. 160-163; see Haddock, McChesney and
Spiegel, 1990; see also the following articles on benefit-based damages:
Epstein, 1994; Levmore, 1994; Polinsky and Shavell, 1994; Wittman, 1984,
9.5 Supercompensatory Damage Awards and Decoupling
Although super compensatory damage awards are necessary to solve some
problems, they may create others. For example, excess compensation may
reduce victims’ incentives to take care, unless a victim’s right to recovery is
governed by contributory negligence. Excess compensation also may preclude
victims from optimally spreading losses, because the marginal utility of wealth
of victims who receive excess compensation will be less than their pre-accident
marginal utility of wealth (Polinsky and Che, 1991; see Arlen, 1990b).
Excess compensation creates problems less often than might at first appear,
however. As Rubinfeld observes, whether victim compensation is excessive
depends on a victim’s compensation net of attorney’s fees and litigation costs,
not the gross compensation paid. Litigation costs and attorneys’ fees are
sufficiently high that defendants’ liability can exceed victims’ losses by a
considerable amount without victims being over-compensated (Rubinfeld, 1984,
Nevertheless, when injurers may escape liability, the optimal deterrence
award net of other costs often will exceed victims’ losses. This creates two
potential problems. First, if victims can influence the probability or magnitude
of the accident, super-compensatory awards may cause victims to engage in
very excessive activity levels (victims will take due care, however, provided
they are subject to a contributory negligence defense). In addition, victims’
risk-spreading will not be optimal because the optimal amount of insurance is
full compensation, not excessive compensation. One potential solution to the
problem of excessive victim compensation is to decouple injurers’ liability from
victims’ compensation, with victims being fully compensated for their losses
(including litigation costs) and any liability in excess of full compensation
being paid to the state (Polinsky and Che, 1991; see Danzon, 1984; Spence,
1977). This possibility is discussed later in this entry.
B. Damages for Death and Serious Physical Injury
Economic analysis of damages for death and serious permanent physical injury
differs substantially from the standard analysis of property damage because
these injuries include harm to irreplaceable commodities. A commodity is
irreplaceable if its owner does not perceive any equivalent commodities
available on the market (Cook and Graham, 1977, pp. 144-146). Loss of this
commodity results in a ‘nonpecuniary loss’. Health is generally considered an
Under current law, victims are not compensated for their total losses arising
from an accident that results in death or serious permanent injury. Rather,
victims are compensated for the pecuniary losses associated with the injury plus
an award for the ‘pain and suffering’ occasioned by the loss. In some cases,
victims also may receive punitive damages.
This section first examines evidence on the actual operation of the tort
system: how much are victims compensated and how much do defendants pay?
It then addresses the issue of what are optimal awards for death and physical
10. Actual Awards for Death and Physical Injury
In a 1991 study, Hensler et al. concluded that liability payments total $15.7
billion annually. Three-quarters of this amount went to victims of automobile
accidents, even though this group constituted only 59 percent of the pool of
victims who received compensation. Compensation net of legal fees and
property damage totaled $12.9 billion (Hensler et al., 1991, pp. 100-101).
This study found that the average compensation paid to victims was
$11,173. The average net compensation for personal economic cost - net of
damages for pain and suffering, property damage and attorneys’ fees - was
$5,432. The average liability payment for automobile accidents was $14,444.
In both situations, the distribution of payments was highly skewed, with most
victims receiving substantially less than the average. The median payment for
all cases was only $2,500; the median for automobile accidents was $4,000
(Hensler et al. 1991, pp. 102-103).
Evidence suggests that victims of nonfatal accidents generally are not
compensated for all the losses they suffer. Hensler et al. found that nonfatal
accidents impose direct and work-loss costs of $175.9 billion annually. Victims
generally bear 38 percent of the monetary loss directly; they only recover 62
percent from other sources (including the tort system and insurance). The study
found that the recovery rate for direct costs is 75 percent, but for work loss is
only 34 percent. Tort liability payments, net of attorneys’ fees, comprise only
7 percent of the total compensation victims receive (Hensler et al. 1991, pp.
102-107). Similarly, Viscusi’s study of product liability claims paid in 1977
found that the average bodily injury loss, $27,446 (1990 dollars) exceeded the
average bodily injury payment of $19,990 (1990 dollars) (Viscusi, 1991, p.
102). Other studies also conclude that seriously injured victims are likely to be
under-compensated (Galanter, 1996, pp. 1117-1119).
This finding that seriously injured victims are under-compensated holds
even though most victims receive awards for pain and suffering. Viscusi found
that 62.5 percent of the cases with positive awards for bodily injury included
awards for pain and suffering. There was considerable variation in the amount
paid, however. The mean value of pain and suffering awards over all cases
where such damages were awarded is $200,000 (1990 dollars) in cases
involving brain damage and paraplegia. However, in 14 of the 18 injury
categories for which pain and suffering was awarded, mean awards are below
$60,000. On average, pain and suffering awards constitute two-thirds of the
total compensation awarded in those cases where pain and suffering damages
are awarded (Viscusi, 1991, pp. 103-107).
Leebron (1989) analyzed cases involving pain and suffering before death
and found that the mean award was $83,700 and the median award was
$28,500 (in 1987 dollars). In general, pain and suffering awards averaged about
23 percent of the total award and almost 76 percent of the wrongful death
amount. Leebron also found considerable variation in the awards, however, not
only in absolute terms but in the per minute payment for suffering. Differences
in the manner of death explains some, but not all, of this variation.
The previous discussion of whether damages under-compensate victims
compares victims’ compensation with actual ex post losses suffered. It also is
instructive to compare compensation to the value the individuals place on their
life and health, as determined by the amount that workers require to
compensate them for additional risk of death or injury. Examining workers’
preferences, Viscusi concluded that their behavior yields an ex ante
compensation demand value of life of $6.4 million. There is considerable
variation, however. Workers in high-risk jobs appear to place an implicit value
on their lives of $1 million or less, whereas workers in very low-risk jobs may
value their lives at $10 million or more (Viscusi, 1991, p. 108). Valuations of
nonfatal injuries yield estimates of $12,000 (1984 dollars) to $50,000 (1987
dollars) per injury. Approximately half of this amount is for noneconomic loss
(Viscusi, 1991, pp. 109-110).
11. What is Optimal Full Compensation for Death or Injury?
The preceding discussion suggests that actual damage awards, therefore, do not
fully compensate victims of serious injuries for their losses. This raises the
issues: what is full compensation for death or permanent injury and should the
tort system try to fully compensate victims of these injuries?
The answer to the first question depends on what efficiency criterion is
employed. Under the Pareto criterion, tort liability must ensure that the
injurers’ risk-taking leaves no one worse off than he would be otherwise. This
criterion would seem to require that victims be fully compensated for their
losses. But whether, and to what extent, victims must be compensated depends
on the initial allocation of entitlements.
This part discusses the damage rule that satisfies the Pareto criterion’s
requirement that no one be made worse off than otherwise. The next part
examines the tort regimes that minimize the joint costs of accidents and
11.1 Injurers are Entitled to Impose Risks at Will
Society may grant to potential injurers an entitlement to impose certain risks
on others. In this case, the Pareto optimal solution to the accident problem is
determined - as in Shavell (1987, pp. 247-254) - by maximizing the utility of
the potential victim subject to the constraint that injurer be made no worse off
than otherwise. Tort damage awards thus need not compensate victims for their
losses because, in essence, they have nothing to lose (see Shavell, 1987, pp.
11.2 Victims Possess Initial Entitlement to Their Health
Alternatively, victims may possess the right to be completely free from certain
types of risks. In this context, the Pareto-optimal solution to the accident
problem is determined by maximizing the utility of the injurer subject to the
constraint that the victim be no worse off than if the injurer does not engage in
the risky activity. This would seem to imply that tort damage awards must fully
compensate the victim for his losses by awarding his sufficient compensation
to return the victim to his pre-accident (healthy) level of utility. But this is not
necessarily the case.
(a) Ex Ante Compensation and Complete Insurance Markets The standard
accident of economic models of tort law assumes that accidents occur as the
result of injurers imposing risks on victims, without either party imposing any
risk of harm on the injurer. In this situation, fully compensating the victim
apparently requires that the award leave the injured victim no worse off than
when he was healthy (see, for example, Arlen, 1985; Friedman, 1982). A
problem arises, however, because many serious permanent injuries dramatically
effect the victim’s marginal utility of wealth. For example, a victim of a fatal
accident may not derive any utility from wealth paid to her after the accident.
Thus full ex post compensation for such injuries would seem to be impossible
(alternatively, it is sometimes said that ex post full compensation award is
infinite, see, for example, Buchanan and Faith, 1979, pp. 245-246; Friedman,
1982; Mishan, 1971, p. 693). Yet, as David Friedman observes, the reason
injurers cannot fully compensate victims for such injuries is not that life has
infinite value, but rather that money paid to a dead person has no value
(Friedman, 1982). The full compensation amount would be finite - and perhaps
quite small - if it were determined ex ante: if those who impose risk
compensated all potential victims ex ante, before any one is injured. Because
the money would be paid to healthy people who value it, only a finite amount
of compensation would be required (Friedman, 1982, p. 83; see Graham and
Implementing a system of ex ante damage awards probably is unworkable,
however, because it would require courts to calculate, in advance, the risks
imposed by an enormous variety of activities and to oversee constant cash
payments to millions of people (Friedman, 1982, p. 83). Yet, Friedman argues,
it may be possible to implement a system which awards damages only ex post
based on the ex ante value of life. Friedman notes that victims could be fully
compensated for their losses with substantially smaller awards than at present
if mechanisms exist that enable potential victims to obtain the benefit of the
compensation while healthy - when the money is valuable. If victims could
benefit from the compensation while healthy, then the proper measure of full
compensation is the amount paid to the victim once injured that fully
compensates her for her loss given that she will get the benefit of some (or all)
of this money while healthy (Friedman, 1982, pp. 83-85, 89-90; compare with
Fraser, 1984a). This solution would be possible if insurance markets were
complete, in that victims could sell insurance on themselves (Friedman, 1982).
Alternatively, the state could allow a market in unmatured tort claims, where
healthy victims could sell their potential claims to others (Cooter, 1989a; see
Shukaitis, 1987, discussing the merits of a market in matured tort claims).
Currently, however, victims cannot sell unmatured tort claims and there are
reasons to doubt the feasibility of either such a market or of ‘reverse insurance’
(Arlen, 1990b, p. 61, n. 90). Yet even so tort damage awards need not
necessarily fully compensate victims ex post in order to ensure that potential
victims are no worse off than otherwise.
(b) Bilateral Risk Accidents First, most accidents result from bilateral risk - not
unilateral risk - activities. Bilateral risk activities are those where both parties
to the accident risk being injured should an accident occur. Car crashes are an
example of a bilateral risk accident.
In many cases, each potential victim of a bilateral risk accident risks injury
only if he also is engaging in a risky activity. Each person thus can avoid being
subject to the risk in question - thereby ensuring his initial level of utility - by
refraining from imposing risk on others. In turn, each assumes a risk of injury
to himself in return for the ability to impose risk on others. In this situation,
tort liability rules are Pareto efficient if they maximize the utility to individuals
of engaging in the risky activity by minimizing the joints costs of accidents and
accident prevention. There is no additional requirement that injured victims be
compensated for their losses. The system compensates each person ex ante for
any risk imposed on him by allowing him to engage in a risky activity,
imposing risks on others which they would otherwise be entitled to be free
from. Moreover, individuals would not want damage awards to be higher than
that necessary to achieve optimal deterrence and risk spreading because this
would require each person to expend greater resources while healthy in return
for the right to receive additional compensation if injured, which necessarily
is worth less to the person if risk spreading is optimal (Arlen, 1990b, pp.
This analysis may extend, with some modifications, to the tort system
generally, if we view it as a general set of rules that allow us to impose certain
risks on others in return for the right of others to impose risk (perhaps different
risks) on us. In this situation, the measure of compensation afforded by the tort
system to each potential victim should include the benefit to him of imposing
risks on others under this rule. People thus might generally prefer damage
awards which do not fully compensate victims ex post if in return for not
receiving full compensation if injured (when the money may have less value)
each party has to pay less to injure others (thus reducing their expected liability
which they often must pay while healthy, when the money is more valuable).
In this case, properly defined, a ‘full compensation’ damage rule - which
ensures that no one is worse off than in the status quo ante where no risks are
imposed - would set ex post recovery at less than the amount that returns an
injured victim to his pre-accident utility (See Arlen, 1990b, 1985, p. 1136).
(c) Products Liability
Damage awards also need not fully compensate victims ex post in product
liability cases if fully informed customers can avoid product-related injuries by
not buying the product. Customers will purchase the product only if its
purchase - including the resulting risk - leaves them no worse off than they
would be otherwise. Thus, regardless of the damage award, customers are never
made worse off (Shavell, 1980). This analysis assumes, however, that
customers are fully informed. If customers underestimate the risk in question,
then they will not be as well off as they would be otherwise unless tort damages
fully compensate them for any harm they suffer (see Section 14).
11.3 Injurers are Entitled to Impose Reasonable Risks
Victims may only be entitled to be free from unreasonable risks of harm
imposed by injurers (Graham and Peirce, 1984). In this situation, under a
system of ex ante damage awards, victims will be as well off as they would be
otherwise if they are compensated only for the cost to them of any risks
imposed on them over the ‘reasonable’ level - for example, the level that results
if the injurer takes optimal care. If, by contrast, compensation is only paid ex
post, then to be as well off as they would be otherwise, victims must be paid the
full ex post compensation demand value of the risk imposed.
12. Inducing Efficient Deterrence and Risk-spreading
To minimize the joint costs of accidents and their prevention, tort rules must
optimally deter injurers’ risk-taking (and, where relevant, victims’ risk-taking)
and be consistent with optimal risk-spreading by risk-averse victims and
injurers. The present analysis examines what tort liability and damage rules
governing serious permanent injuries induce injurers to undertake efficient care
and activity levels, while being consistent with efficient risk-spreading by
victims. The question of what rules will also induce victims to undertake
optimal care and activity levels is addressed earlier in this chapter and in
Chapter 3100. The issue of how to induce injurers to efficiently spread risk is
addressed both in Chapter 5700 and in Chapter 3800.
12.1 The Potential Conflict between Deterrence and Risk-spreading
When accidents result in death or serious physical injury, it may be difficult to
induce both optimal deterrence and optimal risk-spreading. Optimal deterrence
requires that injurers bear the full social cost of their risk-taking activities,
including nonpecuniary losses. Yet were the tort system to award damages to
victims that fully compensated them, ex post, for death or serious permanent
injury, the award would preclude victims from engaging in optimal
risk-spreading, because, assuming that health is a normal good, full ex post
compensation for death or serious permanent injury exceeds the amount that
induces efficient risk-spreading.
Victims efficiently spread the risk of loss when they derive that same
marginal benefit from a unit of wealth whether healthy or injured. If health is
a normal good, then the optimal amount of insurance is less than the amount
needed to fully compensate the victim for his full loss. In other words, a
potential victim’s marginal utility of wealth will be the same, whether healthy
or injured, if he is paid an amount if injured which fails to return him to his
preaccident level of utility (Arrow, 1974; Cook and Graham, 1977; Spence,
1977). To take an extreme example, a person with no heirs would prefer to
have more wealth when alive than when dead, because money has no value to
him once he is dead. Thus, he will not purchase life insurance. And, absent
deterrence concerns, he would prefer to receive even a small amount of money
while healthy to receiving millions of dollars after his death. Thus, any
compensation paid to him after he dies would produce greater social utility if
given to someone who is healthy.
Similarly, in the case of serious permanent injuries, if health is a normal
good then individuals would not fully insure against the risk of a serious
permanent physical injury (Arrow, 1974; Cook and Graham, 1977; Spence,
1977). Thus, damage awards which fully compensate a victim for serious
physical injuries - in the sense of returning him to his preaccident level of
utility - overcompensate victims in that they exceed the efficient amount of
insurance coverage (Cook and Graham, 1977; Spence, 1977).
12.2 Does the Conflict Exist?
The preceding analysis suggests there is an apparent conflict between
deterrence and risk-spreading. In order to be efficient, a tort regime must take
this conflict into account.
Yet in order to address this conflict, we must understand exactly when it
arises, and when it does not. For example, the conflict between deterrence and
risk-spreading may be overstated in the case of negligence liability. Indeed,
under a perfectly functioning negligence rule, there is no conflict between
deterrence and risk-spreading, provided victims can purchase first-party
insurance. Under a perfectly functioning negligence rule with optimal
(deterrence-based) damage rules, injurers will take due care and thus will not
be liable. Thus, tort liability will deter without compensating, and damage
awards will not distort victims’ ability to spread risks using first-party
insurance (Arlen, 1990b; Calfee and Rubin, 1992, p. 403). Of course,
negligence liability does not eliminate the conflict between deterrence and
risk-spreading if injurers are found negligent, for example because courts err
or because principals are unable to induce agents to take due care.
In addition, even the activity is governed by a strict liability regime (or if
injurers risk liability under a negligence regime) employing deterrence-based
damage awards will not necessarily interfere with victims’ ability to optimally
spread risks. Although it is true that in order to induce optimal deterrence
injurers must bear the full cost of any harms they cause, this does not imply that
injurers’ liability must equal the amount that leaves the victim as well off after
the serious injury as before - that is, equal to the full ex post compensation
demand value of the injury. From society’s standpoint, the cost of an injurer’s
risk-taking activities is better measured by the cost of the risk he imposed on
all potential victims than by the ex post compensation value of the actual
injuries suffered by any victims. In other words, the deterrence award should
be based on the ex ante compensation demand value of the injury: on the
amount individuals would require to accept the risk the injurer imposed,
divided by the expected number of accidents (Arlen, 1990b, 1993; Viscusi,
1991, p. 108).
While this amount may be large, in the case of very serious injuries it
generally will be less than the amount needed to compensate injured victims ex
post. Ex ante compensation is based what amount must be awarded to healthy
individuals to compensate them for bearing a risk of a harm that may never
materialize. People often accept relatively less when compensated ex ante,
because they are receiving the money when they are healthy and can enjoy it
more; indeed, individuals often are willing to accept relatively small sums of
compensation for relatively small risks of quite serious injuries. The sum of all
the compensation required generally is much less than the ex post
compensation demand value of the injury. By contrast, the ex post value is
based on the amount needed to compensate a seriously injured person for his
loss. If the injury eliminates, or dramatically reduces, any utility to the victim
of money, then this amount might be astronomical. Indeed, in the case of death,
no such amount may exist - or, the amount may be infinite (Mishan, 1971) -
even though a healthy person might require a relatively small amount ex ante
to induce him to accept a risk of such loss (Arlen, 1993, pp. 1097-1098;
Danzon, 1984; Viscusi, 1991, pp. 89-91; see Geistfeld, 1995, pp. 804-810).
Thus, in order to optimally deter injurers it is not necessary that the award
fully compensates victims for their actual losses. The award must simply equal
the cost to society of the risk imposed. Should this amount equal - or, if there
is first-party insurance, should it be less than - the victim’s optimal insurance
coverage, then even under a strict liability rule, setting damage awards equal
to the amount necessary to optimally deter injurers will not preclude victims
from engaging in efficient risk spreading.
Moreover, even if the ex ante deterrence award exceeds the optimal amount
of insurance, holding injurers liable for this amount will not necessary result
in victims being over-compensated. While injurers’ behavior is based on the
gross award paid - including injurers’ litigation costs and any part of the award
that goes to victims’ litigation costs - victims’ compensation depends on the net
award. A victim’s compensation is excessive only if the net award - net of
attorneys’ fees and litigation costs - exceeds the efficient amount of insurance.
Thus, even if an injurers’ liability equals the full deterrence award, the net
award to the victim may be less than or equal to the optimal amount of
insurance (Geistfeld, 1995, pp. 800-802; Rubinfeld, 1984, pp. 553-555).
Evidence suggests that if damage awards were set equal to optimal
deterrence levels, litigation costs would substantially reduce the risk of excess
compensation. For example, currently pain and suffering awards appear to
roughly equal the standard lawyers’ contingency fee (Viscusi, 1991, p. 114).
Thus, the victim’s net award may be roughly equal to pecuniary losses.
Provided victims can purchase first-party insurance against nonpecuniary
losses, this net award will be consistent with optimal risk-spreading by victims.
Similarly, Geistfeld argues that on net victims will not be over-compensated if
victims are not compensated for all tortuously caused injuries; any
over-insurance for some injuries may offset under-insurance for others
(Geistfeld, 1995, p. 800).
Yet while for some injuries the net deterrence award may be less than or
equal to the optimal amount of insurance, for many it still will exceed the
optimal insurance award. Scholars have estimated the ex ante value of life to
be approximately $3-6 million (Viscusi, 1991, p. 108; see Danzon, 1984, p.
527, placing the value at approximately $1.4 million (1980 dollars), citing
Arthur, 1981; see generally, Kornhauser, 1990). Even net of litigation costs,
this ex ante deterrence damage award likely exceeds the efficient amount of
insurance - particularly if, as some have suggested, individuals would only
insure against pecuniary losses (Danzon, 1984, pp. 522-524; Priest, 1987, p.
1547; A. Schwartz, 1988, pp. 364-367; Leebron, 1989, p. 274; see also Viscusi,
1991, p. 110; Viscusi and Evans, 1990).
12.3 Measure of Optimal Insurance
The magnitude of the gap between the optimal deterrence award and the
optimal insurance award depends critically on what is the optimal amount of
compensation to pay to victims. A number of scholars have argued that
individuals only insurance against pecuniary losses, even when faced with a
risk of a serious nonpecuniary loss, such as is occasioned by death or serious
permanent physical injury (Danzon, 1984, pp. 522-524; Priest, 1987, p. 1547;
A. Schwartz, 1988, pp. 364-367; Leebron, 1989, p. 274).
Yet whether optimal insurance for such injuries is indeed limited to purely
pecuniary losses remains an open question. As a theoretical matter, optimal
insurance can exceed pecuniary losses: optimal insurance for injuries that
increase the victims’ marginal utility of wealth exceeds the victims’ pecuniary
losses. Individuals will insure against nonpecuniary losses if the injury
increases the marginal utility of wealth: implying that the individual derives
more utility out of a dollar spent if injured than he did if healthy. This might
occur if an injured person would be using the wealth to enable him to get
essentials, like mobility, whereas when healthy he would only be using the
money to purchase luxuries. Individuals will not insure against nonpecuniary
losses (and will not even insure against all their pecuniary losses), however, if
the injury decreases the marginal utility of wealth, as occurs with wrongful
death (Calfee and Rubin, 1992; Cooter, 1989a, pp. 388-391; Viscusi, 1991, p.
Empirical evidence suggests that serious on-the-job injuries may reduce the
victim’s marginal utility of wealth (see Viscusi and Evans, 1990). In this case,
the net optimal deterrence award will exceed the optimal amount of
compensation because the former generally exceeds the victim’s pecuniary
losses. Yet there is some evidence to suggest that some injuries do increase the
marginal utility of wealth - or at least that individuals anticipate that it will. For
example, people currently insure against nonpecuniary losses - purchasing, for
example, uninsured motorist coverage. This suggests that such insurance may
be optimal in some circumstances (Arlen, 1990b, pp. 73-74, n. 149; Croley and
Hanson, 1995; Geistfeld, 1995, pp. 795-796).
However, even when optimal insurance includes some nonpecuniary losses,
individuals do not fully insure against nonpecuniary losses, and optimal
insurance often (indeed generally) will be less than the net optimal deterrence
value of the injury. Thus, even when optimal insurance includes insurance for
nonpecuniary losses, the optimal insurance damage award nevertheless
generally will be lower than the (net) optimal damage award, thus presenting
a conflict between risk-spreading and deterrence.
12.4 Eliminating the Conflict Between Risk-Spreading and Deterrence
Thus, in the standard model of accidents - where accidents occur between
‘injurers’ and ‘victims’ who are strangers to each other - if injurers must pay
victims an award based on the deterrence value of the risk, victims’
risk-spreading generally will be suboptimal.
(a) Decoupling The state may be able to eliminate this conflict between
deterrence and risk-spreading by decoupling the injurer’s liability from the
victim’s recovery. In other words, the state can require the injurer to pay the
victim an amount that ensures that the victim’s net recovery equals the
insurance value of the harm, and then, where necessary, impose a deterrence
surcharge on the injurer payable to the state to ensure that the injurer’s total
liability equals the ex ante cost of the risk he imposed (Danzon, 1984; Shavell,
1987, pp. 233-235; Spence, 1977).
It has been suggested that this solution may not be feasible because a
majority of cases settle out of court, enabling the injurer to avoid paying the
fine to the state (Geistfeld, 1995, pp. 799-800; Rubinfeld, 1984, pp. 553-557).
Yet Polinsky and Che argue that decoupling may enhance efficiency even if
cases settle (Polinsky and Che, 1991, pp. 566-567). Kahan and Tuckman
(1995) argue, however, that Polinsky and Che’s result does not necessarily
hold, and that the requisite levies may be ineffective if agency problems affect
the plaintiff-lawyer relationship. Yet Shavell shows that settlements will not
undermine the state’s ability to collect fines if the state refuses to enforce any
settlements that are not registered with the court, with the injurer being
required to pay a levy when he registers. This would ensure that injurers are not
tempted to settle secretly because plaintiffs on the receiving end of secret
settlements could turn around and sue the injurer. The only way for the injurer
to get the benefit of settling would be to register and pay the fine (Shavell,
(b) Bilateral Risks Even without decoupling, most standard torts resulting in
serious permanent injury or death may not produce a conflict between
deterrence and risk-spreading. Most tortuous injuries result from bilateral risk
accidents - those in which both potential parties to the accident risk being
injured. Indeed, most tort cases are automobile accidents - which are single
activity accidents where both parties to the accident are engaged in the same
activity and impose risks on each other. In this situation, the tort system
operates more effectively as a mechanism for deterring risk-taking than for
compensating victims, and tort damage awards may not interfere with victims
ability to engage in optimal risk-spreading through the purchase of first-party
Consider the most extreme type of bilateral risk accident: single activity
accidents where both parties will be injured should an accident occur and will
suffer identical losses. In this situation, neither party will receive any net
compensation under any of the standard liability rules. Under negligence
liability, each party will take due care and will not be liable. Under strict
liability with contributory negligence, each will take due care and will be liable
to the other, but because the accidents are symmetrical, damages paid will equal
damages received with the result that on net neither receives or pays any
compensation. Tort liability thus does not result in victims receiving net
compensation and thus does not interfere with each parties’ ability to employ
first-party insurance to optimally spread the risk of loss. Rather it serves a
purely deterrent function: it will induce due care if the state employs a
duty-based liability rule - that is, negligence liability or strict liability with
contributory negligence - because each party takes due care in order to avoid
net liability for the other’s losses (Arlen, 1990b; see Diamond, 1974, discussing
deterrence but not insurance).
The state also can induce optimal care-taking and promote optimal
risk-spreading when parties to bilateral risk accidents do not suffer the identical
losses. Under any negligence-based liability rule (for example, pure negligence,
negligence with contributory negligence, and strict liability with contributory
negligence) each injurer will take optimal care if damages are set such that the
injurer is better off taking due care than bearing net liability for the accident
(Arlen, 1992a; see Arlen 1990b; compare with Diamond, 1974). If damages are
based on ‘deterrence values’ - that is, on the amount necessary to induce
efficient care-taking - then under either negligence liability rule recovery also
will be consistent with efficient risk-spreading by victims, provided victims can
purchase first-party insurance. Under both negligence and negligence with
contributory negligence, each risk-taker will take due care and thus neither will
be liable. Thus, tort damages only affect potential injurers’ care levels and do
not affect the parties’ ability to spread the risk of loss efficiently. Each will do
so by purchasing first-party insurance (Arlen, 1992a).
Under strict liability with contributory negligence, each will take due care
and thus will be liable. Because each is liable for the other’s losses, the
defendant who suffered the larger injury will receive an award greater than his
liability; the other party will not receive any net compensation. Thus, the party
who faces net liability invariably can spread his risk of loss, provided that he
can purchase insurance, because there is no risk of his being over-compensated.
The party who will receive net compensation should an injury occur also may
be able to optimally spread the risk of loss, if he can purchase insurance and if
his net recovery (net of fees, costs, and the liability to the other) is less than the
amount needed to induce optimal risk-spreading. Alternatively, the state can
guarantee that risk-spreading is efficient by arbitrarily ruling that both parties
to a bilateral risk accident will pay the same damages, with the amount of the
award equaling that minimum award necessary to make each party take due
care. In this case, even under strict liability with contributory negligence,
neither receives any net compensation when each takes due care, and damage
awards do not interfere with victims’ ability to spread risks optimally by
purchasing first-party insurance (Arlen, 1992a). For a more detailed discussion
of efficient insurance, see Chapter 5700.
(c) Products Liability Products liability cases also may not involve a conflict
between deterrence and risk-spreading. If consumers are perfectly informed
about product risks, the market will induce producers to take optimal care and
produce the optimal amount of the product. Thus, damage awards need only
induce optimal risk-spreading by victims. Awards thus should be based on
‘insurance’ values, not deterrence values (A. Schwartz, 1988; Priest, 1987; see
Many argue, however, that tort liability must promote optimal deterrence
if consumers underestimate the risk of harm (see Section 14). Thus, injurers’
liability should equal the optimal deterrence award, which may exceed optimal
compensation to victims (for example, Danzon, 1984; Geistfeld, 1995, pp.
798-799; A. Schwartz, 1988; Spence, 1977). Yet, as previously discussed, a
victim often will be over-compensated - from an insurance standpoint - if his
recovery equals the injurers’ liability. As before, it may be possible to solve this
problem by decoupling the victims’ recovery from the injurers’ liability
(Polinsky and Che, 1991; Spence, 1977). (See the discussion above.)
Of course, this raises the issue of whether consumers do indeed
underestimate the risks of being injured by a product. Schwartz (1988, pp.
378-384) disputes the claim that consumers generally underestimate product
risks sufficiently to produce inefficient equilibria. However, Viscusi and others
have shown that consumers do underestimate some risks (Viscusi, 1991, pp.
134-139; see generally Sunstein, 1997, p. 1183, discussing self-serving biases).
Yet, Viscusi also has shown that consumers overestimate other product risks,
such as risks that are called to their attention (for example, by product
warnings) and risks of certain low-probability events (like cancer) (Viscusi,
1991, pp. 134-139). When consumers overestimate risks, products liability may
lead to over-deterrence.
Moreover, some scholars argue that even if victims are imperfectly
informed, damages in some products liability cases need not equal the
‘deterrence’ value of the injury. Calfee and Rubin claim that imperfect
information does not necessarily justify products liability for products, like
vaccines, which overall serve to reduce the risk of harm. If consumers
underestimate both the risk of harm from the product (for example, the risk of
side effects) and the background risk the product reduces (the risk of disease),
the consumer demand for the product may be suboptimal. In this situation,
employing deterrence-based damage awards may reduce efficiency by deterring
consumers from purchasing a product they should purchase (Calfee and Rubin,
1992; but see Geistfeld, 1995, p. 798, n. 99). Similarly, Knoll argues that stiff
penalties for products liability may have the perverse result of encouraging
firms to keep dangerous products on the market (Knoll, 1996b; see Arlen,
1994, examining the impact of strict liability on a firm’s willingness to monitor
to determine it has done something which is likely to lead to liability).
Finally, the government may be able to better solve imperfect information
problems by giving consumers information on risk rather than by imposing tort
liability. This depends in part on the consumers’ ability to rationally interpret
information about their own risk of being injured (see, for example, Sunstein,
13. Deterrence Awards: Willingness to Pay vs. Willingness to Accept
The optimal deterrence damage award for a particular injury is not the same in
every case. First, a population’s ex ante evaluation of the cost of a particular
risk will depend not only on the magnitude of the risk but also on the
background risk affecting the population, assuming, as is likely, that
preferences are nonlinear in risk (Arlen, 1985, pp. 1132-1134; Kornhauser,
1990, p. 215; see Linnerooth, 1979, pp. 57-58). In addition, it will depend
whether the optimal award is properly based on potential victims’ willingness
to pay to avoid the risk or the amount they must be compensated to accept the
Willingness to pay and willingness to accept produce very different
estimates of the deterrence value of life (Hoffman and Spitzer, 1993; Geistfeld,
1995, pp. 822-825; Knetch, 1984; Kornhauser, 1990). The willingness to
accept (or compensation demand) measure is most appropriate for accidents
between strangers governed by a strict liability rule because strict liability
implicitly grants the victim the entitlement to his health (Arlen, 1985).
Under social welfare maximization or Kaldor-Hicks efficiency, the
determination of due care and damages under a negligence rule also should be
based on the willingness to accept risk if the presumption is that people possess
the right to their health and reasonable risks are those which create benefits
which equal or exceed the amount victims would require to compensate them
for those risks (Arlen, 1985).
Willingness to pay may be the appropriate measure of damages for accidents
resulting from bilateral risks (Geistfeld, 1995, pp. 826-828). Willingness-to-pay
also may be the proper measure of damages in products liability cases because
potential victims are in fact paying for safety through higher product prices.
Yet Geistfeld argues against using this measure on the grounds that it would
be unfair: rich and poor consumers would pay the same product prices and yet
the rich would be entitled to more compensation because they would have a
higher willingness to pay (Geistfeld, 1995, pp. 805-806, n. 123). This problem
disappears, however, if injurers’ liability is based on ex ante deterrence value
of the risk - that is, on the aggregate willingness-to-pay of all consumers who
purchase the product divided by the expected number of accidents. In this case,
all victims would receive the same amount. Where this award conflicts with
risk-spreading goals, the state could decouple victims’ recovery from injurers’
liability by forcing injurers to pay a fine to the state (see the discussion above).
When implementing any damage rule, however, it is important to recognize
that actual damages awarded by courts may also be affected by how the issue
is ‘framed’. Specifically, both cognitive decision theory and experimental
evidence suggests that juries’ and judges’ determination of the value of life is
likely to depend on whether the judge or jury views tort damages as a gain to
an injured victim or as making up a loss to a previously healthy person:
whether they are told to consider the amount needed to make the victim whole
or the amount she would have to be paid to subject herself to the injury in the
first place (McCaffery, Kahneman and Spitzer, 1995).
14. Efficiency of Existing Tort Law
The present review of the literature on efficient tort damage rules for death and
injury reveals that the existing rules governing recovery for wrongful death and
physical injury are not efficient.
Under current law, recovery from wrongful death and physical injury is
based on the victim’s lost wages and medical expenses plus an award for ‘pain
and suffering’. Current measures of pain and suffering generally are designed
to measure just that - the pain of death or injury - and generally are not an
attempt to measure the lost quality of life. Moreover, and more important, pain
and suffering is victim-specific, and thus is not tied to the total cost of the risk
the injurer imposed on the entire population. Thus existing damage awards do
not equal either the ex post value of the victim’s loss or the ex ante cost of the
risk the injurer imposed (the efficient ex ante damage award). Thus, the
injurers’ total liability for wrongful death and serious physical injury is not
based on the social cost of the harms they cause. Accordingly, the tort system
is unlikely to be providing the correct incentives for injurers to take due care
and limit their activity levels (Arlen, 1985, 1993).
Indeed, actual damage awards for wrongful death or serious permanent
injury rarely exceed the efficient ex ante deterrence award of approximately
$3-6 million (Viscusi, 1991, p. 108). This suggests that the tort system may not
induce individuals to reduce the risks they impose to efficient levels. But, from
the standpoint of inducing optimal risk-spreading, the compensation victims
actually receive may be higher than is efficient - although this is unclear given
that much of the award will go to attorneys fees and other costs.
Finally, in the unilateral risk context, damage awards do not satisfy the
requirement for Pareto efficiency that injurers’ risk-taking leave victims no
worse off than they would be otherwise if victims are initially entitled to be free
from risks imposed by others.
C. Special Topics
15. Damages for Corporate Defendants
The standard analysis of optimal damages assumes that liability falls on the
actual injurer. In other words, the standard analysis assumes that the defendant
is the actual injurer, and that both people are natural persons. Yet often the
defendant is a corporation which is being held liable for wrongdoing of its
managers or other employees. Indeed, most of the famous tort cases involve
corporate defendants - for example, U.S. v. Carroll Towing Co., 159 F.2d 169
(2d Cir. 1947); The T.J. Hooper, 53F.2d 107 (S.D.N.Y. 1931); Palsgraf v.
Long Island R.R., 162 N.E. 99 (N.Y. 1928). Analysis of corporate liability for
torts of managers or other employees reveals that the standard analysis of
individual injurers does not necessarily apply to corporate liability. Holding
firms liable for their employees’ torts induces employees to take optimal care
and firms to engage in optimal activity levels if firms either can directly control
employees’ care levels or can use wages to force employees to bear the full
social cost of wrongdoing. This regime does not induce optimal care, however,
if firms cannot directly control employees’ care levels and employees cannot
pay the optimal damage award (Kornhauser, 1982; Shavell, 1987, pp. 170-175;
In this situation, optimal deterrence requires that corporate tort liability
induce firms to optimally deter their employees’ wrongdoing. To do this,
however, one must do more than simply force firms to pay for any harms
When employees are insolvent and firms cannot directly control their
employees’ care-taking, to optimally deter wrongdoing it may be necessary to
induce firms to implement measures which increase employees’ expected
liability by increasing the probability that they will be liable for their torts.
These measures - hereinafter called policing measures - include monitoring
employees, conducting self-evaluative audits of the firm’s compliance with the
law, investigating suspected wrongdoing, and reporting wrongdoing. In
addition, to optimally deter wrongdoing firms may need to undertake
preventive measures - such as screening employees - and also adjust activity
levels to reflect the full social cost of wrongdoing (Arlen and Kraakman, 1997).
The current civil corporate liability regime is best described as being a
regime of traditional strict vicarious liability, under which firms are liable for
every harm caused by their employees’ in the course of their employees’
employment and subject to a fixed damage award equal to the harm caused,
with no mitigation if the firm took steps to avoid the harm (Arlen, 1994). This
regime is not in and of itself efficient. First, if the probability of detection is less
than one, the firm will not undertake optimal prevention measures or activity
levels unless the firm’s liability equals the social cost of the harm divided by
the probability of detection (h/p) - an amount which may be many times higher
than actual harm (Arlen and Kraakman, 1997; see Kornhauser, 1982; Polinsky
and Shavell, 1993, 1998; Sykes, 1984).
Second, in order to induce optimal policing measures - specifically
measures that affect the probability that the firm will be found liable -
traditional strict liability generally should be replaced with a partial duty-based
regime (‘composite regime’). Under this regime firms would be subject to a
legal duty to monitor, audit, investigate and report optimally. In order to induce
optimal prevention measures and activity levels, the firm would be subject to
a sanction of h/p* - where p* is the optimal probability of detection - if it
satisfied its policing duties but would be subject to a much higher sanction of
h/po if it did not - where po is the probability of detection if the firm does not
undertake any policing (Arlen and Kraakman, 1997). Strict liability will not
induce firms to undertake optimal policing, unless the sanction is h/p, where
p depends on the actual probability of detection (Arlen, 1994; Arlen and
Kraakman, 1997). Full analysis of corporate tort liability also would require
examination of the impact of criminal laws and also market-based sanctions.
The present analysis reveals that one cannot simply apply the findings of the
standard economic models of torts to cases involving corporate defendants. This
suggests that we need to develop a better understanding of why corporations
commit torts and how firms respond to liability, and that we should be
circumspect about basing policing decisions governing corporate liability on
economic models of how individuals behave. This topic is discussed in more
detail in Chapter 3400 on Vicarious and Corporate Liability.
16. Economic and Nonpecuniary Losses
Even where full compensation is efficient, concern for efficiency suggests that
victims should not necessarily be compensated for their economic losses.
Economic loss usually refers to foregone profits or earnings. Scholars argue
that victims should not necessarily be compensated for economic losses because
only some economic losses are social costs. For example, assume that a firm
loses business to a competitor selling the same product at a lower price as a
result of the tortuous activity of the latter. In this case, from the perspective of
‘static’ efficiency, the ‘victim’s’ lost profits are not a measure of social loss but
are actually a measure of social gain. Similarly, when an accident halts a firm’s
production of a good, the firm’s lost profits on foregone sales are not social
costs if other firms produce perfect substitutes at the same cost (Shavell, 1987,
Nevertheless, in other cases economic losses may also be social losses. For
example, an accident that interferes with production will cause social losses if
alternative goods are manufactured at higher cost. This cost difference is the
measure of the social loss and is a proper component of damages. This amount
will be less than economic losses, however, if the price of the good does not
change. Social costs may exceed economic losses if no alternative sale is made,
because the social costs equal the lost consumer surplus which necessarily
equals or exceeds the producer’s economic loss (Bishop, 1982; Shavell, 1987,
pp. 135-140). (For additional discussions of tort liability for economic losses
see Goldberg, 1994; Landes and Posner, 1987, pp. 251-255; Rabin, 1985;
Rizzo, 1982; G. Schwartz, 1986, 1996.)
In contrast to economic losses, which often should not be included in
damage awards, nonpecuniary losses generally should be included.
Nonpecuniary loss generally refers to the loss of an ‘irreplaceable’ commodity -
a commodity which cannot be purchased on the market. Irreplaceable goods
include the sentimental value a person attaches to a family photo or other
personal possessions. Health is another irreplaceable good (see discussion
above). Nonpecuniary losses are social costs: the cost is the reduction in the
victims’ utility. In order to ensure that injurers take into account the full social
cost of their actions, injurers must be liable for both pecuniary and
nonpecuniary losses (Bishop and Sutton, 1986; Shavell, 1987, pp. 133-135).
The one exception to this is where nonpecuniary loss is small enough that the
benefit of holding injurers liable for this nonpecuniary loss is less than the
administrative cost of estimating them (Shavell, 1987, pp. 133-135). This
condition is satisfied only if the benefit of including such losses measured by
the effect of including them on all injurers who impose risks to such
commodities (whether injury results or not) exceeds the administrative cost of
calculating them in the (relatively fewer) cases where the commodity is actually
injured. Additional issues concerning recovery for nonpecuniary losses are
17. Role of Defendants’ Wealth
The standard analysis generally assumes that potential injurers are identical.
The question arises, however, what if potential injurers have different wealth?
Should this affect the damage award?
Under the Pareto efficiency criterion, defendants’ wealth should be
irrelevant to the determination of damages (see Shavell, 1987, pp. 215-227;
Miceli and Segerson, 1995). This result also holds under the total social utility
maximization criterion if each individual has the identical marginal utility of
wealth for all levels of wealth - that is, if individuals are identical and are risk
neutral (Arlen, 1992b; Abraham and Jeffries, 1989).
When social welfare is measured by maximizing total social utility, then
employing a simple model in which individuals are risk averse but are
otherwise identical it can be shown that total social utility is maximized if
wealthier people spend more on care; which can be achieved by forcing them
to pay higher damages. The reason is that in this situation defendants have
declining marginal utility of wealth. Thus, the wealthier the defendant the less
impact an expenditure of a dollar has on his welfare. Thus, optimal due care for
wealthier individuals exceeds optimal due care for poorer defendants, because,
all else equal, the marginal cost of care for a wealthier defendant is lower. It is
possible to induce greater care-taking by wealthier defendants by having
damages vary with defendants’ wealth (Arlen, 1992b).
This conclusion results from the fact that under the social welfare
maximization criterion transfers of wealth can increase social welfare - as
defined as maximizing total social utility. The fact that when the total social
utility criterion is employed it is possible to increase ‘social welfare’ through
purely distributional changes is a reason that many economists prefer other
measures such as the Pareto criterion or Kaldor-Hicks criterion (Arlen, 1990b).
Moreover, Miceli and Segerson question the conclusion that having
defendants’ damages vary with wealth maximizes total social utility. They
agree that the efficient standard of care and damage awards vary with
defendants’ wealth when a negligence liability rule is employed and tort
liability rules are the only instrument employed to increase social utility (Miceli
and Segerson, 1995, pp. 202, 204-205). However, they conclude that under
strict liability care levels and damage awards should be independent of
defendants’ wealth (Miceli and Segerson, 1995, pp. 204-205).
Furthermore, there are additional reasons to question whether wealth
differences should affect damages, Arlen argues that we should not necessarily
base damages on defendants’ wealth - even where static analysis might suggest
that basing damages on defendants’ wealth would maximize total social utility -
because basing damages on defendants’ wealth would have negative affects on
willingness to accumulate wealth. In addition, it would entail substantial
administrative costs (Arlen, 1992b, pp. 427-429; see Polinsky and Shavell,
1998, pp. 910-913). Moreover, it is far from clear that society should design
tort rules to maximize total social utility, instead of rules that are Pareto or
Kaldor-Hicks efficient (see Arlen, 1985, 1990b).
Finally, this analysis only examines one method of achieving redistribution.
Other instruments, such as the tax system, might be superior for affecting the
redistribution of wealth potentially required for social utility maximization
(Kaplow and Shavell, 1994). Indeed, Kaplow and Shavell argue that income
taxes are superior to liability rules for redistributing wealth because while both
instruments distort people’s choice between labor and leisure, redistribution
through legal rules imposes an additional cost: it distorts the behavior the legal
rule was meant to regulate. They, therefore, conclude that legal rules should
focus on efficiency alone (Shavell, 1981; Kaplow and Shavell, 1994).
Recently, Sanchirico (1997) has challenged this conclusion that legal rules
need only focus on efficiency because equity concerns are better addressed
through the tax system. He argues that even if income taxes are optimal, tort
rules may need to take considerations other than efficiency into account if
people with different wealth levels respond in different ways to the tort system -
for example, if the wealthy are more cautious. In addition, equity considerations
may be relevant if wealth differences are explained primarily by income
differences and high income individuals tend to be more accident prone.
Moreover, complete analysis of the issue of whether taxes or legal rules should
be used to serve equity concerns would require analysis of the political economy
of legislatures versus judges - of which institution is more likely to adopt an
optimal, or second-best efficient, regime (Sanchirico, 1997).
Christine Jolls (1998) argues that the case for using legal rules to achieve
redistribution is strengthened once one takes into account well-known features
of human behavior that depart from the rational choice model. She argues that
if - as experimental research suggests - people are overly-optimistic about
uncertain events, exhibit risk-seeking behavior towards losses, and engage in
mental accounting, then redistrubtion through tort liability may distort work
incentives less than an income tax. This conclusion undermines a central
argument against using tort liability to achieve redistrubtional goals. More
research is needed to determine whether behavioral economic analysis argues
in favor of using tort liability to achieve redistribution, however.
18. Sharing Losses Among Multiple Tortfeasors
Many torts involve multiple tortfeasors. Under current law, generally joint
tortfeasors are jointly and severally liable for a plaintiff’s losses. This rule holds
each defendant potentially liable for the full amount of the plaintiff’s losses.
Apportionment rules, by contrast, apportion tort liability between defendants
in relation to their responsibility for the harm. This raises the issue of how to
apportion damages amount multiple tortsfeasors. This issue is discussed at
length in the chapter on Joint Tortfeasors (3200) and thus will only be
considered briefly in this chapter.
The standard legal regime holds multiple defendants jointly and severally
liable for any losses they cause. This implies that each defendant is potentially
liable for the full amount of the victim’s losses. Alternative regimes apportion
this liability between the defendants.
Under a negligence liability regime, joint and several liability will induce
potential injurers to take due care if they are fully solvent (Kornhauser and
Revesz, 1989; Landes and Posner, 1980; Landes and Posner, 1987, pp.
190-227; see Shavell, 1987, pp. 164-167). By contrast, if each injurer is only
liable for the amount of damage attributable to his own negligence - for
example, in a pollution case each is not liable for damage caused by the other
person’s dumping - then injurers will fail to take due care (Kornhauser and
Revesz, 1998, p. 372; 1989).
Kornhauser and Revesz argue that the result is different under strict
liability, however. They conclude that neither joint liability nor non-joint
liability will cause injurers to take due care (Kornhauser and Revesz, 1989,
1998). By contrast, Shavell concluded that under strict liability an
apportionment rule will result in injurers taking due care if they act in concert,
but not if they act independently (Shavell, 1987, pp. 164-167, 177-179). For a
discussion of apportionment rules when injurers are insolvent see Kornhauser
and Revesz (1990). For a discussion of the effect of joint and several liability
on settlement see Kornhauser and Revesz (1994a; 1994b).
19. Calculation of and Timing of Damage Awards
19.1 Accuracy in the Assessment of Damages
Litigants often devote substantial effort to establishing the level of harm. The
question is, are the costs of increased accuracy socially optimal? Kaplow and
Shavell (1996) argue that if injurers know the harms they will cause, accurate
damage awards will lead potential injurers to internalize the true cost of the
risks they impose, leading to optimal care-taking and activity levels. But if
injurers do not know the harms they will cause, then simply knowing that
damages are accurate will not induce injurers to engage in optimal behavior.
In this situation, expenditures on accuracy are wasteful, if all we are concerned
about is optimally deterring injurers’ behavior. Injurers can be optimally
deterred by basing damages on the average harm caused. Yet this conclusion
does not hold if courts’ efforts to assess harms correctly causes injurers to learn
more about the expected harms they may cause. At present, however, both
parties are likely to have excessive incentives to provide information to courts
about harm because litigants have ex post incentives to provide information
about harm if actual harm is differs from estimated harm, even if this
information has no social value (Kaplow and Shavell, 1996).
19.2 Individualized versus Scheduled Awards
Danzon (1984) argues that in products liability cases, the current system of
individualized awards should be replaced by a system of scheduled awards,
under which a victim’s recovery for a particular injury is pre-determined, based
on the average cost of such an injury. Rubinfeld (1984) suggests that scheduled
awards will not necessarily reduce administrative costs and that the variance
associated with individual awards may produce some benefits.
Arlen (1990b, 1993) and Viscusi (1991) also argue for scheduled awards
and against basing awards on victims’ actual losses. They argue that optimal
damages for physical injury should be based on the ex ante cost to society of the
risk the injurers imposed, and not on the actual victim’s ex post losses. This
proposal differs from Danzon’s in that damage awards would depend on the
magnitude and nature of the risk imposed - not just on the harm. Thus an
injurer’s liability for a given loss - for example, wrongful death - would not be
a fixed amount but rather would depend on the magnitude of the risk imposed
and the nature of the risk (say fire versus being crushed). It might well be,
however, that the victim’s recovery could be decoupled from the injurer’s
liability, so that victims who suffer a particular type of harm - say loss of an
arm - would receive the same amount. Those who would prefer to receive more
could purchase insurance.
19.3 Lump-Sum versus Periodic Damages
Compensatory damages often must cover not only past losses but also future
losses, for example future lost wages associated with a serious permanent
physical injury. The question arises whether this system of lump-sum awards
should be replaced by a system under which awards are made periodically,
contingent on the actual amount of the future loss.
Those who support periodic contingent awards argue that contingent awards
reduces the calculation problems associated with the uncertain nature of future
losses. Rea (1981) argues, however, that the case for lump-sum awards is
stronger than it might seem because periodic awards can add substantially to
accident costs. First, Rea notes that ex ante victims will never be made better
off by switching to periodic awards because a victim who would prefer periodic
awards can always transfer a lump-sum award into a periodic award by
purchasing insurance. Second, he notes that defendants may be risk averse. A
risk-averse defendant will prefer a certain lump-sum payment to uncertain
contingent payments with the same expected value. The defendant thus subject
to a contingent award will likely purchase insurance. Rea suggests that a system
under which the victim obtains a lump-sum award and purchases insurance is
preferable to one under which the defendant is subject to a periodic award and
purchases insurance (Rea, 1981; see also Schuck, 1990).
19.4 Prejudgment Interest
Because it often takes many years from the time of an injury for a successful
plaintiff to receive an enforceable judgment, courts award prejudgment interest.
A proper award of prejudgment interest is necessary both to ensure that the
plaintiff is fully compensated and that the defendant pays the full cost of the
injury. The question is, what is the appropriate rate of prejudgment interest?
Michael Knoll argues that economic and finance theory implies that when
the parties have ready access to the capital markets (as publicly traded
corporations do) prejudgment interest should be awarded at the defendant’s cost
of unsecured borrowing. A defendant who is assessed a judgment has in effect
borrowed the judgment amount from the plaintiff. As with other loans, the
plaintiff is compensated if he receives interest at the defendant’s borrowing
rate. Because judgment creditors are treated along with other unsecured
creditors in bankruptcy, awarding prejudgment interest at the defendant’s cost
of unsecured borrowing will both compensate the plaintiff and prevent the
defendant from being unjustly enriched (Knoll, 1996a). For other articles
discussing this topic see Fisher and Romaine (1990), Patell, Weil and Wolfson
Economic analysis has greatly enriched our understanding of damage rules. It
reveals that damages serve a complex and multi-faceted role: deterring
risk-takers, helping victims spread risks, and compensating them for their
losses. Economic analysis reveals how to design tort liability and damage rules
to serve these goals. This analysis can guide legislators and courts as they
design tort liability and damage rules.
Existing research suggests that at present damage awards for serious
personal injury and death generally are not sufficiently large to induce potential
injurers to take due care and engage in optimal activity levels. Yet economic
analysis also suggests that victims of physical injuries may be receiving too
much compensation. This suggests that states should consider decoupling
defendants’ liability from victims’ compensation.
Economic analysis also suggests that the fact that tort liability for certain
accidents is too low does not imply that all damages awards should be
increased. Although defendants’ liability is too low in some circumstances,
existing research suggests that it may be too high in others. A defendant’s full
costs of risk creation include litigation costs, any market penalties the
defendant bears as a result of either the harm or being held liable, and any
additional government-imposed sanctions, such as civil penalties imposed by
an administrative agency or criminal penalties. These costs must be included
in any assessment of whether tort damage rules are adequate. Moreover, more
empirical evidence is needed on the questions of what is the optimal ex ante
deterrence award for different risks, and what are potential victims’ preferences
for spreading the risk of various losses.
Economic analysis of tort damages also would benefit from additional
analysis on why accidents happen in the first place - particularly, why are
people negligent? Simple economic models predict that if damages and liability
rules are set correctly, no one will be negligent. Yet there is considerable
anecdotal evidence - and some empirical evidence - that people are indeed
negligent (see Weiler, 1993a, pp. 72-73 showing that doctors are negligent in
a substantial portion of cases). The question is why? Is it that damages are too
low? Is it that optimal care is set incorrectly? Or are there other forces at work
- perhaps informational or institutional problems - that result in the tort system
not creating adequate incentives for people to take due care. Economists have
explored some of the possible reasons: such as the risk of court error; the
possibility injurers will not be liable for their torts; and the fact that many tort
defendants are business organizations, not individuals. These and other possible
explanations warrant additional analysis.