Docket No. 99-7550 August Term, 1999

After a jury trial in the United States District Court for the Southern District of New York (Robert P. Patterson, Jr., J.), plaintiff-appellee Debra Ciraolo was awarded $19,645 in compensatory damages and $5,000,000 in punitive damages against defendant-appellant the City of New York (the "City"), in her suit under 42 U.S.C. § 1983 for, inter alia, an unlawful strip search of her person. The City now appeals the award of punitive damages, and, for the following reasons, we reverse.

* In January 1997, following a complaint by her neighbor, with whom she was apparently having a legal dispute, Debra Ciraolo was arrested for aggravated harassment in the second degree, a misdemeanor. Ciraolo was taken to the police station and then to Central Booking, where she was subjected to a strip and body cavity search by two female Corrections Department employees. Ciraolo was ordered to strip naked and made to bend down and cough while she was visually inspected. After spending the night in jail, she was released on her own recognizance; the charges against her were subsequently dismissed. Ciraolo was traumatized by the entire experience, and particularly by the humiliation of the strip search. Diagnosed with post-traumatic stress disorder, she entered therapy and began taking antidepressants.

The search of Ciraolo was not an isolated incident. Rather, it was in accordance with an established City policy of strip-searching all arrestees, including misdemeanants, whether or not there was reasonable suspicion that the arrestee possessed contraband. In July 1996, the City's Correction Department had adopted "guidelines [for] the acceptance of all Police Cases for the Manhattan Court Division," providing that "[a]ll police prisoners received shall be strip search[ed] by the officer assigned to the search post." In October 1996, the Executive Officer of the Manhattan Detention Complex implemented the guidelines by sending a memo to all personnel ordering that, "[e]ffective immediately, all female police prisoners arriving at this facility . . . be strip searched." Accordingly, when Ciraolo was arrested in early 1997, she was strip-searched as a matter of course, despite the conceded lack of any reasonable suspicion for the intrusive and demeaning search.

Ciraolo brought suit under § 1983 against the City, the police department, and the individual police officers involved in her arrest and search, claiming that she had been falsely arrested, that the police had employed excessive force during the arrest, and that her strip search violated the Fourth Amendment. She also alleged battery, in violation of state law.

The district court found that the City's policy of strip-searching all arrestees regardless of the existence of reasonable suspicion violated the Fourth Amendment, in contravention of the clearly established law of this Circuit that "the Fourth Amendment precludes prison officials from performing strip/body cavity searches of arrestees charged with misdemeanors or other minor offenses unless the officials have a reasonable suspicion that the arrestee is concealing weapons or other contraband." Weber v. Dell, 804 F.2d 796, 802 (2d Cir. 1986). Because the parties had stipulated to the existence of the City's policy of strip-searching all arrestees, and because there was no evidence that Ciraolo was believed to be concealing contraband, the court instructed the jury that the City was liable for any injuries the jury found to have been proximately caused by the strip search.

After soliciting letter briefs from the parties on the issue of punitive damages, the district court concluded that, in this case, punitive damages were available against the City because the strip search was pursuant to an official City policy that was contrary to the settled law of the Circuit. Accordingly, over the City's objection, the court charged the jury that punitive damages could be awarded against the City if they found that the City had acted maliciously or wantonly. In doing so, the court told the jury to consider whether compensatory damages would be adequate to punish the City and to deter future unlawful conduct.

The jury found in favor of the defendants on Ciraolo's claims of unlawful arrest, excessive force, and battery. It concluded, however, that the City had acted in wanton disregard of Ciraolo's rights when she was strip-searched, and awarded her $5,000,000 in punitive damages, as well as $19,645 in compensatory damages.

On appeal, the City does not contest the district court's holding that the strip-search policy was clearly unconstitutional under this Circuit's decision in Weber. Nor does it challenge the award of compensatory damages to Ciraolo. Rather, it contests only the award of punitive damages, arguing that the award was foreclosed by the Supreme Court's decision in City of Newport v. Fact Concerts, Inc., 453 U.S. 247 (1981), which held that, ordinarily, municipalities are immune from punitive damages under § 1983. In her turn, Ciraolo argues that Newport, in a footnote, permitted municipalities to be charged punitive damages in cases where "the taxpayers are directly responsible for perpetrating an outrageous abuse of constitutional rights," id. at 267 n.29, and that her case falls within this exception. We are, of course, bound by Newport. And we believe that the only fair reading of that decision, and of the footnote on which Ciraolo relies, mandates a conclusion that punitive damages are not available to her.

In Newport, a musical concert promoter, Fact Concerts, Inc., brought suit against the City of Newport, Rhode Island, under § 1983 for a violation of its First Amendment rights. Fact Concerts had entered into a contract with Newport to present a jazz festival. Shortly before the festival, Fact Concerts hired the group Blood, Sweat and Tears to replace an act that had dropped out. See id. at 250. Under the impression that Blood, Sweat and Tears was a "rock" group and that it might attract "a rowdy and undesirable audience," id., Newport tried to prevent the group from performing; eventually, the City Council voted to cancel the contract, a decision reported in the local media, see id. at 250-52. Although Fact Concerts obtained an injunction from a state court allowing the festival to proceed, ticket sales were off substantially due to the adverse publicity. See id. at 252. At trial, Fact Concerts won $200,000 in punitive damages against Newport, as well as other damages against the individual officials. See id. at 253.

The Supreme Court reversed the award of punitive damages. See id. at 271. In doing so, the Court applied "a two-part approach" to interpreting § 1983. Id. at 259. It considered, first, the extent of municipal immunity at common law and the legislative history relevant to § 1983, and, second, the policies behind punitive damages and their compatibility with the purpose of § 1983. See id. at 258-59.

In its historical inquiry, the Court reviewed nineteenth-century tort cases, and concluded that by 1871, when § 1983 was enacted, "municipal immunity from punitive damages was well established at common law." Id. at 263. Proceeding on the premise that Congress would have been fully aware of this common-law immunity and would have been explicit had it intended to abolish the immunity, the Court turned to the legislative history of § 1983 to see if any such intent could be divined. See id. at 263-64. Because, in fact, there was almost no legislative history surrounding the passage of § 1983 itself, the majority looked to the debate that took place over the proposed Sherman amendment to the Civil Rights Act that included § 1983. The Sherman amendment would have made municipalities liable for compensatory damages for injuries caused by mob violence. See id. at 264. Noting that, as one supporter said on the floor of the House, the amendment was not intended "as a punishment for the county," but as "mutual insurance" to "indemnify the injured party," the Court concluded that "a Congress having no intention of permitting punitive awards against municipalities in the explicit context of the Sherman amendment would [not] have meant to expose municipal bodies to such novel liability sub silentio" (and more broadly) under § 1983. Id. at 265.

The Court then turned to the question whether the two major policy arguments for punitive damages - deterrence and retribution - would be advanced by assessing punitive damages against cities. It concluded that punitive damages against cities were not justified by a policy of deterrence because (1) it was unclear that municipal officials would be deterred by the prospect of damages borne by the taxpayers; (2) voters would, nonetheless, be likely to vote wrongdoing officials out of office absent punitive damages, both because they had done wrong and because of the possibility of compensatory damages; (3) punitive damages assessed directly against the offending officials would be a more effective means of deterrence; and (4) punitive damages could "create a serious risk to the financial integrity" of cities. Id. at 270; see id. at 268-70.

Nor, the Court opined, was the goal of retribution furthered by punitive damages, since such damages would punish only the innocent taxpayers. "[P]unitive damages imposed on a municipality are in effect a windfall to a fully compensated plaintiff, and are likely accompanied by an increase in taxes or a reduction of public services for the citizens footing the bill. Neither reason nor justice suggests that such retribution should be visited upon the shoulders of blameless or unknowing taxpayers." Id. at 267. Accordingly, the Court held that "a municipality is immune from punitive damages under 42 U.S.C. § 1983." Id. at 271.

The Court did, however, indicate that in some rare cases, the goal of retribution might be furthered by punitive damages against a municipality. In a footnote, the majority mused, "It is perhaps possible to imagine an extreme situation where the taxpayers are directly responsible for perpetrating an outrageous abuse of constitutional rights. Nothing of that kind is presented by this case. Moreover, such an occurrence is sufficiently unlikely that we need not anticipate it here." Id. at 267 n.29.

Ciraolo relies on footnote 29 for her claim that punitive damages are warranted in her case. She reads the footnote as creating an "outrageous abuse" exception to the general municipal immunity established by Newport, and contends that the City's strip-search policy is sufficiently outrageous to fall within that exception. We believe that Ciraolo misapprehends footnote 29.

Footnote 29 must be read as part of the Court's discussion of the retributive function of punitive damages. As such, it elaborates on the Court's conclusion that it is unjust to punish the taxpayers for tortious conduct by municipal officials, because the taxpayers are not normally directly responsible for that conduct. The footnote suggests that, in the rare case where the taxpayers are directly responsible for an outrageous violation of a plaintiff's constitutional rights, it may well be just for the taxpayers to bear the burden of punitive damages. To the extent that footnote 29 creates an exception to Newport's general rule against punitive damages, therefore, it is not an exception for particularly outrageous abuses, as Ciraolo would have it, but rather an exception for outrageous abuses for which the taxpayers are directly responsible.

The footnote does not tell us when, in the Court's judgment, taxpayers could be held to be directly responsible for a municipal policy. Although it could be argued that, to the extent that they are also voters who play a part in choosing municipal officials, taxpayers are always responsible for municipal policies, such responsibility is clearly too indirect to give rise to liability for punitive damages under the logic of Newport. Footnote 29 seems, instead, to contemplate a much more immediate connection between the taxpayers' behavior and the unconstitutional municipal policy, perhaps - for example - as close a link as a referendum in which the taxpayers directly adopted the invalid policy.

We are, moreover, aware of no case in which a court of appeals has upheld an award of punitive damages against a municipality premised on footnote 29. Indeed, very few cases have even discussed the footnote, although the First and Fifth Circuits have offered considered analyses of it. See Webster v. City of Houston, 689 F.2d 1220 (5th Cir. 1982), rev'd on other grounds on reh'g, 735 F.2d 838 (5th Cir. 1984) (en banc); Heritage Homes of Attleboro, Inc. v. Seekonk Water Dist., 670 F.2d 1 (1st Cir. 1982). In Heritage Homes, the First Circuit appeared to acknowledge that the footnote might apply in a case where the taxpayers voted in favor of an unconstitutional policy, but nevertheless declined to allow punitive damages where residents of the Seekonk Water District had engaged in "blatantly raci[st] discussions" before voting to exclude from access to water a housing development owned by developers willing to sell houses to black families. Id. at 2. The court reasoned that only a tiny minority of Water District members had actually voted to exclude Heritage Homes, and that "[a]bsent widespread knowledgeable participation by taxpayers . . . to award punitive damages against the Water District would not serve the purposes of punishment or deterrence. . . . The actions of a small claque of voters would burden several thousand non-participants, many of whom presumably were unaware of the entire controversy." Id.

In Webster, the Fifth Circuit addressed the potential applicability of footnote 29 in a § 1983 suit brought by the parents of Randy Webster, a seventeen-year-old boy shot and killed by Houston police after a car chase -- while he was unarmed and attempting to surrender. The police had attempted to cover up the shooting by planting a gun in the boy's hand, and it came out at trial that the Houston police had an unwritten custom of carrying "throw-downs," guns or knives to be planted near suspects who had been shot in dubious circumstances. After trial, the jury awarded the Websters, inter alia, $200,000 in punitive damages to be assessed on the city. See Webster, 689 F.2d at 1223. Reversing this award of punitive damages, the court stated:

The plight of Randy Webster, however reprehensible, however tragic, does not rise to the level of outrageous conduct to which Justice Blackmun referred [in footnote 29]. . . . [Section 1983 was] enacted in a time of frightening violence, to ensure the most basic constitutional rights of citizens in southern states. If the members of Congress who drafted that Act did not intend to establish a rule of punitive damages, we believe that it would take a far more serious violation than that we confront to ground punitive damages against Houston.

Id. at 1229. The Webster majority, in contrast to the First Circuit, thus focused on the outrageousness of the particular conduct at issue, rather than the extent to which taxpayers could be held directly responsible for it.

In a powerfully written concurrence, Judge Goldberg disagreed with the majority's interpretation of footnote 29. In his view, "rather than creating an exception, this footnote precludes any exception." Id. at 1231 (Goldberg, J., concurring). He commented that "[i]t might be possible to read footnote 29 as allowing an exception where the electorate of a political unit votes affirmatively in favor of outrageous conduct," although "[e]ven in that situation there would be some difficulty in assessing damages where a minority of the electorate voted against the outrageous conduct." Id. at 1231 n.1. Nevertheless, he took the thrust of the footnote to be essentially a bolstering of Justice Blackmun's argument that allowing punitive damages against a municipality would not further their traditional retributive purpose. See id. at 1231. Although he believed that the Supreme Court's holding in Newport swept too broadly, and was both unwarranted and unwise in a case like Randy Webster's, see id. at 1231-32, he therefore concurred in the judgment.

We are inclined, like the First Circuit, to the view that any exception envisioned in footnote 29 must be an exception for cases in which the taxpayers are directly responsible - in their role as voters - for the adoption of an unconstitutional municipal policy, rather than an exception for especially outrageous abuses of constitutional rights. If, for example, a town adopted, by a unanimous vote, a referendum establishing an unconstitutional rule, we can see no reason in the policies discussed in Newport why punitive damages ought not to be awarded.1 But such a case is not before us today. (And so nothing we here say about the applicability of footnote 29 to a referendum can be binding on any future court presented with the question.) Rather, we are here faced with the much more common situation in which an unconstitutional policy has been adopted by municipal officials without any clear endorsement of the policy by the electorate. In such circumstances, while we emphatically deplore the City's conduct in adopting a policy that this Circuit had earlier clearly held unconstitutional, the taxpayers themselves cannot be held to be responsible for the policy under the reasoning of Newport.

The district court, while not relying explicitly on footnote 29, concluded that Newport did not preclude the award of punitive damages to Ciraolo because the Newport cause of action stemmed from one improper action by Newport officials, while in Ciraolo's case the City had adopted a policy contrary to the clearly established law of this Circuit. We sympathize with the district court's desire to allow the jury to award punitive damages in this case. Like it, we are seriously troubled by the City's adoption of a policy we had previously held to be unconstitutional. Nevertheless, we conclude that the district court misread Newport, and that the distinction drawn by the court has little force in this context.

Municipal liability under § 1983 occurs, if at all, at the level of policy-making, and cannot be premised on a theory of respondeat superior. See Monell v. Department of Soc. Servs., 436 U.S. 658, 694 (1978) ("[A] local government may not be sued under § 1983 for an injury inflicted solely by its employees or agents. Instead, it is when execution of a government's policy or custom . . . inflicts the injury that the government as an entity is responsible under § 1983."). As a result, the fact that Ciraolo's strip search occurred because of an established City policy in no way distinguishes her case from other § 1983 cases in which municipal liability is established, and cannot suffice to place this case within the seemingly narrow exception carved out by footnote 29.

Accordingly, and despite what might be the salutary effects of punitive damages in a case such as this, we are constrained by the Supreme Court's holding in Newport to reverse the award of punitive damages.2

* * *

The award of punitive damages against the City is REVERSED and the case is REMANDED to the district court for entry of judgment in accordance with this opinion.

Although the result the court reaches today is compelled by Supreme Court precedent, I respectfully suggest that the policies behind punitive damages and the purpose of 1983 would be better furthered by a different outcome. I write separately to explain why I believe this to be so. Cf. Abbate v. United States, 359 U.S. 187, 196, 3 L. Ed. 2d 729, 79 S. Ct. 666 (1959) (Brennan, J., concurring in his own majority opinion).

* The purpose of 1983 is "not only to provide compensation to the victims of past abuses, but to serve as a deterrent against future constitutional deprivations, as well." Owen v. City of Independence, 445 U.S. 622, 651, 63 L. Ed. 2d 673, 100 S. Ct. 1398 (1980). Indeed, the Court in Newport recognized that punitive damages serve the dual purpose of deterrence and retribution. See Newport, 453 U.S. 247 at 266-67, 69 L. Ed. 2d 616, 101 S. Ct. 2748. Nevertheless, and as discussed above, the Court concluded that "the deterrence rationale of 1983 does not justify making punitive damages available against municipalities," id. at 268, because (1) it was unclear that municipal officials would be deterred by the prospect of damages borne by the taxpayers; (2) voters would still be likely to vote wrongdoing officials out of office absent punitive damages, both because they had done wrong and because of the possibility of compensatory damages; (3) punitive damages assessed directly against the offending officials would be a more effective means of deterrence; and (4) punitive damages could "create a serious risk to the financial integrity" of cities. Id. at 270; see id. at 268-70.

The Court's analysis, however, neglected at least one aspect of the deterrence function of punitive damages - an aspect underscored by this case. Punitive damages can ensure that a wrongdoer bears all the costs of its actions, and is thus appropriately deterred from causing harm, in those categories of cases in which compensatory damages alone result in systematic underassessment of costs, and hence in systematic underdeterrence.

It is easy to show why this is so. A rational actor1 will undertake an activity when the benefits of doing so exceed the costs. In doing so, it will make some sort of formal or informal, spoken or unspoken, cost-benefit analysis, based on the information it possesses, to determine if a particular activity is worth its price. Such an analysis cannot be even roughly accurate unless approximately all the costs of the activity are borne by the actor. When the perceived benefits of an activity accrue to the actor, but some significant part of the costs is borne by others, the cost-benefit analysis will necessarily be distorted. In such a case, the actor will have an incentive to undertake activities whose social costs exceed their social benefits. In other words, the actor will not be adequately deterred from undesirable activities. And society will suffer.2

One goal of the tort system, therefore, is to ensure that actors bear the costs of their activities. See, e.g., Calabresi, supra note 4. And that is also a goal of 1983, which, as the Supreme Court has noted, was designed to deter constitutional torts by making the tortfeasors liable in damages to their victims. See Newport, 453 U.S. at 258; Owen, 445 U.S. at 651.

In some circumstances, compensatory damages alone will be enough to promote an adequate cost-benefit analysis. In other cases, however, compensatory damages will not come close to equaling all the costs properly attributable to the activity. Costs may not be sufficiently reflected in compensatory damages for several reasons, most of which go to the fact that not all injured parties are in fact compensated by the responsible injurer. For example, a victim may not realize that she has been harmed by a particular actor's conduct, or may not be able to identify the person or entity who has injured her. See, e.g., Polinsky & Shavell, supra note 4, at 888. Where the injurer makes active efforts to conceal the harm, this problem is of course exacerbated. Moreover, even if a victim is aware of her injury and is able to identify its cause, she may not bring suit. A person will be unlikely to sue if the costs of doing so - including the time, effort, and stress associated with bringing a lawsuit - outweigh the compensation she can expect to receive. A victim is especially unlikely to sue, therefore, in cases where the probable compensatory damages are relatively low. See id. As a result, a harm that affects many people, but each only to a limited degree, will generally be given inadequate weight if only compensatory damages are assessed.

In addition, some victims will not sue even if the damages they could expect to receive would exceed the costs of suing. Victims will differ greatly in their knowledge of and access to the legal process, and those who are relatively poor and unsophisticated, as a practical matter, are frequently unable to bring suit to redress their injuries even if those injuries are grave. A harm that disproportionately affects such victims, therefore, is also particularly likely not to be accurately reflected in compensatory damages.

For these and other like reasons, compensatory damages are, in wide categories of cases, an inaccurate measure of the true harm caused by an activity, and, as a result, making an injurer bear only such damages does not provide adequate deterrence against socially harmful acts. In such circumstances, additional damages, assessed in the cases that are brought, may be an appropriate way of making the injurer bear all the costs associated with its activities.

This idea is far from new. Many years ago, in an influential article that was in part responsible for his receipt of the Nobel Memorial Prize in economics, Professor Becker pointed out that charging a thief the cost of what he had stolen would not adequately deter theft unless the thief was caught every time. Since thieves will not always be caught, they must be penalized by more than the cost of the items stolen on the occasions on which they are caught. This "multiplier" is essential to render theft unprofitable and properly to deter it. See Becker, supra note 4. More recently, scholars have recognized that punitive damages can serve the same function in tort law. See, e.g., Thomas C. Galligan, Jr., Augmented Awards: The Efficient Evolution of Punitive Damages, 51 La. L. Rev. 3 (1990); Polinsky & Shavell, supra note 4.

Professors Polinsky and Shavell, for example, have argued that punitive damages should be assessed whenever a tortfeasor has a significant likelihood of escaping liability, and have even suggested a formula (simpler to state than to apply) for calculating such damages: total damages should equal the amount of loss in a particular case, multiplied by the inverse of the probability that the injurer will be found liable.3 See id. at 889. Thus, if the injurer causes harm of $ 10,000, but will be found liable only one-fifth of the time, total damages - according to Polinsky and Shavell - should equal $ 50,000 ($ 10,000 in compensatory damages and $ 40,000 in punitive damages). See id. Use of such a multiplier, they argue, will cause damages to equal the total harm caused, forcing the injurer to take into account all the costs of its activity and thus decreasing the likelihood that socially harmful activities will persist.4 See id.

Such a conception of punitive damages, again, is not new, and it has been recognized by courts as well as scholars. Indeed, the Supreme Court, considering punitive damages in a quite different context from that in Newport, acknowledged that in determining such punitive damages, it is proper to consider the extent to which the tortfeasor might otherwise escape liability. See BMW of North America, Inc. v. Gore, 517 U.S. 559, 582, 134 L. Ed. 2d 809, 116 S. Ct. 1589 (1996) (noting that "[a] higher ratio" of punitive to compensatory damages "may . . . be justified in cases in which the injury is hard to detect"); see also id. at 592-94 (Breyer, J., concurring) (suggesting that "punitive damages awards that, as a whole, would take from a wrongdoer the total cost of the harm caused" might constrain potential arbitrariness and "counsel[] more deferential review" by appellate courts). In addition, the Seventh Circuit, in a 1983 suit against an individual defendant, has explicitly endorsed the multiplier function of punitive damages. See Kemezy v. Peters, 79 F.3d 33, 35 (7th Cir. 1996) (Posner, J.) (pointing out that "when a tortious act is concealable, a judgment equal to the harm done by the act will underdeter" because the tortfeasor "will not be confronted by the full social cost of his activity"). Finally, Judge Goldberg, in his concurrence in Webster, offered a powerful critique of Newport's treatment of punitive damages and suggested that deterrence would be furthered by the assessment of punitive damages against municipalities in 1983 actions where the municipal policy at issue is one that tends toward concealment of wrongdoing. See Webster, 689 F.2d 1220 at 1235-38 (Goldberg, J., concurring).

Although widely accepted by economists and acknowledged by some courts, the multiplier function of punitive damages has nonetheless been applied haphazardly at best. One reason this is so is that the twin goals of deterrence and retribution are often conflated, rather than recognized as analytically distinct objectives. The term "punitive damages" itself contributes greatly to the confusion. For punitive damages, the term traditionally used for damages beyond what is needed to compensate the individual plaintiff, improperly emphasizes the retributive function of such extracompensatory damages at the expense of their multiplier-deterrent function. It also fails totally to explain the not unusual use of such damages in situations in which the injurer, though liable, was not intentionally or wantonly wrongful.5 A more appropriate name for extracompensatory damages assessed in order to avoid underdeterrence might be "socially compensatory damages." For, while traditional compensatory damages are assessed to make the individual victim whole, socially compensatory damages are, in a sense, designed to make society whole by seeking to ensure that all of the costs of harmful acts are placed on the liable actor.6

Indeed, it would not be inappropriate to disaggregate the retributive and deterrent functions of extracompensatory damages altogether and allow separate awards to further the two separate goals. In such a system, socially compensatory damages might be permitted in cases in which all the costs of a defendant's actions were not before the court, whether or not the defendant's conduct was particularly blameworthy.7 But a separate award of punitive damages would be allowed only in cases where the defendant's conduct was sufficiently reprehensible to deserve punishment apart from whatever assessment was required to compensate the individual victim or society as a whole.8

The majority in Newport raised the objection that "punitive damages . . . are in effect a windfall to a fully compensated plaintiff." Newport, 453 U.S. 247 at 267, 69 L. Ed. 2d 616, 101 S. Ct. 2748. It is likely that a feeling that such "windfalls" are not warranted - and are distributed arbitrarily among plaintiffs - lies, and properly so, behind much of the current distaste for punitive damages. Of course, socially compensatory damages will have an equivalent deterrent effect on the injurer no matter whom they are paid to. As one commentator put it, "an efficient windfall is still efficient." Galligan, supra, at 58; see also Webster, 689 F.2d at 1237 n.7 (Goldberg, J., concurring). Nevertheless, the existence of such potential windfalls may well induce undesirable behavior on the part of victims and of their lawyers. And there is no good reason why socially compensatory damages should be paid to the individual plaintiff, at least beyond a relatively small part sufficient to induce the victim to undertake the expense of pleading and proving them.

In fact, in order to achieve the goal of social compensation, as well as the goal of optimal deterrence, it would be preferable if such damages were paid into a fund that could then be applied to remedy some of the unredressed social harm stemming from the defendant's conduct. And some states have explicitly recognized this socially compensatory function by mandating that at least a portion of punitive damages awards should go not to the plaintiff, but to the state treasury or to a specified fund. See BMW, 517 U.S. at 616-18 (Ginsburg, J., dissenting) (collecting state laws providing for allocation of punitive damages to state agencies). A similar mechanism could be employed in the case of socially compensatory damages assessed against municipalities.9

As Judge Posner suggested in Kemezy, socially compensatory damages are anything but foreign to the goals of 1983. And the rationale for socially compensatory damages in the case of an individual defendant holds equally true where the defendant is a municipality. Indeed, the case before us shows especially vividly the possible utility of such socially compensatory damages. Here, the City of New York adopted a strip-search policy that it knew or clearly should have known was unconstitutional. Counsel for the City estimated at trial that about 65,000 people arrested for misdemeanors had been subjected to strip searches under the policy. Under ordinary circumstances, very few of those 65,000 victims would have been likely to sue, both because the compensatory damages they would have received would have been relatively low and because they were no doubt, in the main, relatively poor and unsophisticated.

The conditions of constitutional tort litigation for harm caused by law enforcement officials are ones in which the deterrent effect is likely to be on the low