Compensating for Unforeseeable Damages in Torts

Transcription

1 Compensating for Unforeseeable Damages in Torts Jeong-Yoo Kim Kyung Hee University November 6, 2007 Abstract The doctrine regarding unforeseeable damages in a contract was established in the well known case of Hadley vs. Baendale. According to the judgement, a plaintiff cannot be compensated for unforeseeable damages in an incomplete contract unless he informs the defendant of the possible unforeseen contingency beforehand. In this paper, I etend the argument to the tort case in which it is hardly possible to communicate between a plaintiff and a defendant before an accident occurs. I argue that in the case of the sequential bilateral accident, the social optimum may be attained under the negligence rule by conditioning the defendant s negligence standard and the award of unforeseeable damages on the care level actually taken by the plaintiff. JEL Classification Code: K13 Key Words: negligence, tort, unforeseeable damages 1 Introduction The scope of liability in torts and contracts is, in principle, limited to damages reasonably foreseeable by the defendant. 1 If the defendant could not reasonably have foreseen that mailing address: Department of Economics, Kyung Hee University, 1 Hoegidong, Dongdaemunku, Seoul , Korea, (Tel) , ( ) jyookim@khu.ac.kr 1 An eception is damages for fraudulent misrepresentation when all damages suffered by the defendant are claimable. Also, for intentional torts like assault and battery, the defendant can recover totally unforeseeable 1

2 someone might be hurt by his actions, he is usually not liable. 2 The doctrine regarding unforeseeable damages in a contract was established in the well known case of Hadley vs. Baendale. In Hadley a miller entered into a contract with a carrier to transport a broken shaft. Since the miller had no spare shaft, he had to shut down operations until a new one was delivered. However, delivery was delayed for some reason, and the miller sued for the profit loss from the increased shut-down period. In this case, the court held that the buyer did not inform the seller fully of the possible contingency incurring him huge losses in case of a contract breach. Since the seller could not reasonably anticipate the consequential losses to the buyer, the buyer was not entitled to be compensated for the unforeseeable damages. If the seller had informed the buyer of the potential contingency, the parties might have contracted for full damage insurance, 3 and the outcome would have been changed. As a result, this case established a consequential damage rule whereby a plaintiff cannot be compensated for unforeseeable damages in an incomplete contract unless he informed the defendant of the possible unforeseen contingency in advance. Ayres and Gertner (1989) argues that such a consequential damage rule can induce the high-damage plaintiff to contract around the default rule by penalizing him for withholding information that would allow the seller to take an efficient level of precaution. 4 In tort cases, however, the potential plaintiff and the potential defendant are not supposed to be acquainted with each other. Usually. they are totally strangers to each other and have little chance to echange or transmit information before an accident. Accordingly, unlike in contract case, it seems reasonable not to allow unforeseeable damages in no case, because it does no good to allow unforeseeable damages in order to give a proper incentive not to impose costs on others if the defendant cannot know of the possibility itself. This argument is correct only to the etent that the plaintiff and the defendant take care simultaneously. If their actions are taken sequentially 5 so that the uninformed defendant can observe the damages (Eggshell skull rule). 2 This is known as remoteness. 3 Ayres and Gertner (1989), p For more formal analysis of this default rule, see Bebchuk and Shavell (1991). 5 A typical eample of a sequential care-taking tort case is New York Central Railroad Company vs. Thompson. In this case, a woman who caught her foot in railroad tracks was injured by a train that failed to brake in time. 2

3 informed plaintiff s action of care, 6 the defendant may be able to infer the true state of the world. Thus, the plaintiff may be able to signal the unanticipated occurrence of the contingency via his action. Although a defendant has little chance to be directly informed of the plaintiff s private information, he may have a good chance to infer the information. The plaintiff s action of care in this case plays a comparable role to preannouncement or disclosure of information in the case of a contract. Based on the insight, I consider the damage rule that allows unforeseeable damages only when the plaintiff takes more precaution than usual, and that this rule, which is an analog to the Hadley rule and referred to as conditionally limited liability rule, may give both the plaintiff and the defendant incentives to take the socially optimal level of care if the legal authority sets the defendant s negligence standard conditioning on the plaintiffs care level under simple negligence. The intuition goes as follows. It is relatively easy to implement the socially optimal care level of the defendant, because it is possible simply by adopting the dual due care level which is equal to the efficient care level in each state of the world. Then, since the plaintiff must bear all the accident costs in each period, it is not surprising for him to choose the socially optimal level of care. In this equilibrium, the true type of the plaintiff (true state) is perfectly revealed by his care level viz. his signal. The concern that remains is whether or not one type of plaintiff can be benefited from pretending to be the other type. This may or may not be the case, depending on whether the distortion cost eceeds the benefit from the imitation. If so, the social optimum can be attained; otherwise, some efficiency loss due to distortion will be inevitable. Depending on the size of the signaling cost, the equilibrium can be either a separating one involving costly signaling or a pooling one in which no useful information is conveyed. Signaling by a plaintiff under special circumstances is often observed. For eample, a driver may use hand gestures or the sign baby on board. In those cases, a stricter due care level of the injurer is usually required. I also consider two alternative rules, one of which is to deny unforeseeable damages under any circumstances and the other of which is to allow any full damages. I show that neither of the two rules, which will be referred as limited liability rule and unlimited liability 6 If the plaintiff s care level is not observable to the defendant, the situation is identical to the case of simultaneous moves. 3

4 rule respectively, 7 can never achieve the social optimum as far as the accident-preventing technologies of the victim and the injurer are symmetric. It is well documented that in a bilateral accident, the strict liability rule gives the victim no incentive to take care. 8 However, it is shown in this paper that the claim is valid only when the court awards the full damage. If there is the possibility that the damage award can be limited, the victim who cannot recover all his losses will take some care. The paper is organized as follows. In Section 2, I set up a simple basic model. The social optimum that can serve as the benchmark will be described in Section 3. The analysis of the basic model will be provided in Section 4. I discuss some of the issues that are not addressed in the basic model in Section 5. Conclusion will follow in Section 6. 2 Model An accident can occur by an unlawful act of a potential injurer (defendant, or D) against a potential victim (plaintiff, or P ). The accident incurs P foreseeable losses and possibly unforeseeable losses as well. 9 The size of foreseeable losses is common knowledge, while the size of the total losses is private information of P. I assume that both of the injurer and the victim are risk-neutral. I also assume that the accident is bilateral, in other words, the probability of an accident depends on the victim s care level as well as the injurer s care level. For a liability rule, I consider the simple negligence rule. The following notation will be used throughout the paper. = the care level of a victim ( 0) y = the care level of an injurer (y 0) y = the legal standard for the care level of an injurer p(, y) = the probability of an accident (p, p y < 0, p, p yy > 0) 7 I borrow the two terms from Bebchuk and Shavell (1991), although the meaning of the limited liability rule is slightly different. 8 See Shavell (1980) for eample. 9 The corresponding dichotomy in civil law is ordinary losses and special losses respectively. Special losses occur under unforeseeable special circumstances. 4

5 L e = foreseeable losses L u = unforeseeable losses α = the probability of the unforeseen contingency (0 < α < 1) The injurer s true total losses will be denoted by w. Then, w is either L e with probability 1 α and L = L e + L u with probability α. I will call the victim with w = L type h and the victim with w = L e type l. 3 Social Optimum Given that w is realized, I can define the (e post) social cost as follows; C(, y) = + y + p(, y)w. This is minimized at the efficient values s (w) and y s (w) satisfying C = 1 + p ( s (w), y s (w))w = 0, (1) C y = 1 + p y( s (w), y s (w))w = 0. (2) Equation (1) and (2) have the usual interpretation that the marginal benefits from an increase in care equals the marginal cost of greater care to each party. I assume that C(, y) is strictly conve and thus that the solution to (1) and (2) is unique. Since s and y s are interwoven so as to affect the marginal social cost of each other, it is not trivial to see the change in s and y s with respect to a change in w. Thus, I introduce some taonomy. The two care levels and y will be called substitutable if p y > 0, complementary if p y < 0 and independent if p y = 0. Then, I have the following proposition. Proposition 1 If and y are complementary or independent, both of the socially optimal care levels, s and y s, increase with w. If and y are substitutable, both of them cannot decrease with w. Proof. Total differentiation of (1) and (2) yields d s dw = p yp y p p yy, (3) w H 5

6 dy s dw = p p y p y p, (4) w H where H = p p yy p 2 y. Since strict conveity of C(, y) implies that H > 0, it follows that ds, dys dw dw imply that > 0 if p y 0. Suppose that ds dw, dys dw < 0 when p y > 0. Equations (3) and (4) p y p y < p p yy, p p y < p y p. Multiplying the two inequalities yields p p yy < p 2 y, which is a contradiction to H > 0. If the injurer s care and the victim s care are substitutable and one (for eample, y) is superior to the other in preventing the accident in the sense that p y > p, it can be socially efficient for the victim to increase his care and for the injurer to reduce his care when w is increased. However, if accident-preventing technologies of the two parties are symmetric in the sense that p (a, b) = p y (b, a) for all (a, b), it is socially optimal that both parties increase their care levels when w is increased. Throughout the paper, I will assume that accident-preventing technologies of the injurer and the victim are symmetric. 4 Sequential Game under Incomplete Information In this section, I analyze the sequential game in which the informed P first takes care and the uninformed D reacts by choosing y, based on his observation of. 10 is essential; otherwise, D action cannot be made contingent on P s action y. Observability of Following the spirit of most literature on Hadley, 11 I consider several damages rules and eamine which rule can possibly lead to the first-best outcome. 10 In the common law tradition, the doctrine of last clear chance is usually applied. Under this doctrine, an increased duty of care than normally required is imposed on the last mover who is in a better position of avoiding the accident. See Grady (1988) for an analysis of a sequential game that put a special focus on the doctrine of last clear chance. 11 This literature includes Ayres and Gertner (1989), Bebchuk and Shavell (1991), Maskin (2006), Posner (2006) among others. 6

7 4.1 Rule of Denying Unforeseeable Damages (Rule 1) I first consider the rule whereby the unforeseeable damages are denied in any case. This rule is an analog of the limited liability rule or the foreseeable damages rule (in incomplete contracts) that makes D liable only for the P s normal loss in case provision fails. It is clear that this rule cannot achieve the first-best outcome, insofar as the due care of D is uniform in each state, since it is not possible to induce D to take its respective effective care in each state. To elaborate, given the due care of the injurer y = y s (L e ), the injurer would be eempted from liability by choosing y = y and then P would be epected to bear all the accident costs. A low type of P will clearly choose s (L e ) since he believes that D will choose the due care of y if P s choice of care is a separating one. In the case that w = L, D might possibly not meet the due care y. But even if P chooses his first-best care s (L), there is no possibility at all that D will choose y s (L). The reason is quite straightforward. Since s (L) > s (L e ), it follows that p( s (L), y) < p( s (L e ), y). If p y 0, D may choose y y = y s (L e ) but not y > y s (L). If p y > 0, y = y must minimize the epected cost of D. Therefore, D chooses y s (L) in neither case. Now, suppose that the legal authority e post applies the dual care of D, y(l e ) and y(l), which are equal to y(l e ) = y s (L e ) and y(l) = y s (L), when it turns out that w = L e, L respectively. Then, the situation is more complicated. Since a negligent D who can infer that w = L from compensates L e which is less than L, it may be too costly to meet the due care y s (L). Thus, it is not clear that he will choose y(l). As far as my primary interest is in the implementability of the first-best outcome, it suffices to consider the case that he meets his due care by choosing y(l). Then, in a separating equilibrium, a high-type P should bear the accident loss L, thereby choosing min + p(, y s (L))L. However, he would always prefer mimicking a low type, that is, taking a lower care to induce D to fail to meet the due care. Thus, the first-best outcome cannot be attained in this case, too. Theorem 1 Rule 1 cannot implement the first-best outcome, whether the due care is unitary of dual. 7

8 Proof. The proof is immediate from the above discussions. 4.2 Rule of Awarding Full Damages (Rule 2) The second rule I consider is to make the negligent D liable for the full loss, whatever it turns out to be. This rule is an analog of the unlimited liability rule 12 or the full damages rule 13 in contracts. Suppose that the due care level of D is dual, viz. y(l e ) and y(l). Again, assume that y(l e ) = y s (L e ) and y(l) = y s (L). The defendant, who cannot tell the type of P e ante, may choose his e post first-best care level by inferring the true type from his observation of in a separating equilibrium. Consider the following separating strategy pair yielding the first-best outcome; the low-type P chooses s (L e ), the high-type P chooses s (L) and the defendant meets the legal standard of his care. It is not difficult to see that this cannot be an equilibrium. Similarly, the high-type P always has the incentive to mimic the lowtype by reducing his care level in order to induce the injurer to take less precaution, y(l e ), consequently failing to meet his due care standard. Theorem 2 Rule 2 cannot implement the first-best outcome, whether the due care is unitary of dual. Proof. The proof is immediate from the above discussions. 4.3 Rule of Awarding Unforeseeable Damages Contingently (Rule 3) Now, I consider a rule whereby the due care of D and the damages awarded at court both depend on the actual care level of P, not on the e post realized value of w. This rule most resembles, what Ayres and Gertner call, the penalty default rule that denies the plaintiff the unforeseeable damage unless he informed the defendant of the possibility of the contingency in advance. I will call this the conditional limited liability rule. 12 Bebchuk and Shavell (1991). 13 Ayres and Gertner (1989). 8

9 Conditioning the due care and the damages on P s e ante care level as well as the e post realization of w may appear redundant, but in fact it is not the case. It plays the crucial role of transmitting the private information of P and, moreover, effectively transmitting the information by precluding the incentive of a high-type P to mimic a low-type which was inherent under both rule 1 and rule 2. I design the following form of the damages rule; the due care of D is y() = y h if h and y l if < h, and when the defendant is negligent, the court awards L if and only if h and w = L, otherwise awards L e. To see the possibility that this rule can induce the first-best outcome, let h = s (L), y h = y(l) and y l = y(l e ). As usual in sequential games, the analysis is made backwards. First, consider the decision of the injurer. Given < h, his due care is y l = y(l e ). Thus, the epected cost that he should bear is y + p(, y)l e if y < y C D (y) = l, y if y y l. Since y l is defined to minimize C D (y), the injurer will choose y = y l given < h. On the other hand, if h, the due care is y h = y(l). In this case, D can correctly infer that w = L in equilibrium, since only a high-type plaintiff will choose h. Therefore, the cost he is epected to bear is y + p(, y)l if y < y C D (y) = h, y if y y h. Again, by the definition of y h, the injurer will choose y = y h. When w = L e, the cost that P bears is + p(, y C P () = l )L e if < h, + p(, y h )L e if h. Figure 1a shows the possibility that it is optimal for the low-type victim to choose = s (L e ) < h = s (L). This is guaranteed if C P ( s (L e )) = s (L e ) + p( s (L e ), y l )L e < s (L) + p( s (L), y h )L e = C P ( s (L)). [ND] This condition can be eplained as follows. Given that D chooses y l, = s (L e ) is costminimizing to the low-type plaintiff. However, by imitating the high-type plaintiff s care to 9

10 induce a higher care level of D, P may make his total cost lower. [ND] condition implies that the second effect should not be so large. It is clear that the high-type P has no incentive to mimic the low-type. 14 Therefore, under [ND] condition, the no-distortion first-best outcome can be an equilibrium. To summarize, I have Proposition 2 Under [ND] condition, rule 3 implements the first-best care levels of the victim and the injurer. Proof. It is immediate from the above arguments. If [ND] condition is not satisfied, 15 the low-type victim has an incentive to mimic the high-type in the no-distortion outcome, so the equilibrium outcome must be distorted in some way. There are two possible equilibrium candidates; a separating equilibrium and a pooling equilibrium. A separating equilibrium can occur if y h is set high (y h > y s (L)) enough that the low-type cannot mimic. Such separation is possible due to the difference in the signalling cost across types which is measured by the difference between p(, y)l and p(, y)l e. A (distorted) separating equilibrium care level by a high-type plaintiff, h, must satisfy two incentive compatibility conditions, that is, (i) a low-type P has no incentive to choose h rather than l and (ii) a high-type P has no incentive to deviate from such a costly distorted care level. The set of equilibrium values of h is drawn in Figure 1b. A pooling equilibrium is also possible if the signaling is too costly for the high-type victim. Just as a certain type of plaintiff may have an incentive to withhold private information in a contract, a low type of the plaintiff wants to be pooled with a high type in this pooling equilibrium. 5 Discussion A. Alternative Liability Rules It is well known that in a bilateral accident the victim has no incentive to take care at all under the strict liability rule. This is true in this situation of incomplete information if the 14 Some may be concerned about the possibility that the high-type P may pretend to be the low-type by reducing the care level so as to induce the injurer to fail to meet his legal standard of care. However, as long as y depends on, this is not possible. 15 See Figure 1b for this case. 10

11 court awards the full damages regardless of the circumstances (Rule 2). However, if the court denies the unforeseeable damages L u (Rule 1), the victim who is informed of w = L must bear the loss L u that cannot be recovered, thereby taking some positive level of care. Specifically, a high-type plaintiff will choose the care level to min + p(, y ())L u, subject to y () = arg min y y + p(, y)l e. It is part of a separating equilibrium, since the low-type plaintiff will take no care. However, it is of no significance for D to infer whether w = L e or L, because D must pay the same amount L e in the event of an accident. Also, some may wonder whether the contributory negligence rule can perform better. However, it is not difficult to see that even under the contributory negligence rule, [ND] condition is required to prevent the incentive of a low-type P to mimic a high-type requires [ND] condition. B. Asymmetric Precaution Technology In the case of asymmetric accident-preventing technology, the intuition for the outcome under Rule 1 is not affected. Now, consider Rule 1 and Rule 2. Note that s (L e ) > s (L) and y s (L e ) < y s (L) if y is a superior technology. Then, it is ambiguous whether a high-type P has an incentive to mimic a low type by choosing a higher level of care s (L e ) in the first-best outcome. P s increase in his care induces D to lower his care to y s (L e ) so as to fail to his legal standard of care. The indirect effect of shifting the accident cost to D may either dominate or be dominated by the direct effect of taking more care. Despite this modified feature, the first-best outcome cannot be an equilibrium under Rule 2, because the low-type P has an incentive to deviate by choosing a lower care level s (L) thereby inducing D to increase his precaution level, which always makes P better off. The same logic can be applied to the case that is a superior technology to y In this case, a high-type P mimics a low type by lowering his care, thereby inducing D to increase his care. 11

12 Finally, consider Rule 3. Suppose y is a superior technology. Since s (L e ) > s (L) in this case, the candidate for the optimal damages rule should be modified as follows; the due care of D is y() = y h if h and y l if > h, and the court awards L to the negligent D if and only if h and w = L. Now, consider the decision of a low-type P. If he deviates to a lower care level s (L) in order to mimic a high type, D responds by increasing his care, making P better off. This implies that [ND] condition is never satisfied, that is, the first-best outcome cannot be supported as an equilibrium. Alternatively, suppose is superior. In this case, a low-type P has no incentive to mimic a high-type, because his deviation by increasing the care to s (H) makes D lower his care, thereby increasing the accident probability. However, a high-type P may or may not have an incentive to mimic a low-type. If he reduces his care level to s (L e ) in order to mimic a low-type, it leads D to increase his care to y s (L e ). The sustainability of the first-best outcome depends on whether the latter indirect benefit eceeds or is dominated by the former direct distortion. If the direct distortion cost dominates the indirect benefit, the first-best outcome can be supported as an equilibrium, while it can otherwise. (See Figure 2 for the possibility of the first-best equilibrium.) 6 Conclusion In this paper, I proposed a new damage rule in torts by etending the argument of the penalty damages rule, what Ayres and Gertner (1989) call, in contracts, and eamine when it can be efficient to compensate the victim for unforeseeable damages. The main insight behind this proposal is that the care level by the first-moving victim can play the role of a signal for an unforeseeable contingency. I show that this signaling role enables the damage rule that awards unforeseeable damages conditionally can induce the first-best precaution of both parties, the victim and the injurer. This result is an eact analog to the result of Ayres and Gertner (1989) and Bebchuk and Shavell (1991) in the contet of torts. Considering the apparently transparent advantage of this rule, I hope that this rule will be applied to many pertinent real cases. 12

13 References [1] Ayres, I. and R. Gertner, 1989, Filling Gaps in Incomplete Contracts: An Economic Theory of Default Rule, Yale Law Journal 99, [2] Bebchuk, L. and S. Shavell, 1991, Information and the Scope of Liability for Breach of Contract: The Rule of Hadley v. Baendale, Journal of Law, Economics and Organization 7, [3] Grady, M., 1988, Common Law Control of Strategic Behavior: Railroad Sparks and the Farmer, Journal of Legal Studies 7, [4] Maskin, E., 2006, On the Rationale for Penalty Default Rules, Florida State University Law Review 33, [5] Posner, E., 2006, There Are No Penalty Default Rules in Contract Law, Florida State University Law Review 33, [6] Shavell, S., 1980, Strict Liability versus Negligence, Journal of Legal Studies 9,

14 y l L e p ), ( + L y p l ), ( + y h L e p ), ( + (L) s ˆ ˆ) ( C P )) ( ( L C s P

15 C P C P (ˆ) s ( ( L)) ˆ + p(, y l ) L e s (L) + p(, y h ) L e p (, y ) l L + h

16 L L y p s )) (, ( + (L) s ) ( e s L L L y p e s )) (, ( +