The Objectives of Sexual Harassment Law, with Application to 1998's Ellerth, Oncale, and Faragher decisions January 30, 1999 Eric Rasmusen ABSTRACT Imposing liability on a company for sexual harassment by supervisors cannot be justified as promoting equality between the sexes, protection of workers, or protection of the owners of the company. Such liability might be justified to prevent breach of contract or behavior offensive to the general public-- a "civility code". The recent Supreme Court ruling in Oncale that same-sex harassment is illegal can be justified on these grounds. The ruling in Ellerth and Faragher concerning employer liability for sexual harassment by supervisors contrary to the employer's interest is less satisfactory because the Court's rule will encourage litigation and defensive bureaucratic complexity. Professor of Business Economics and Public Policy and Sanjay Subheadar Faculty Fellow, Indiana University, Kelley School of Business, BU 456, 1309 E 10th Street, Bloomington, Indiana, 47405-1701. Office: (812) 855-9219. Fax: 812-855-3354. Email: Erasmuse@indiana.edu; Erasmuse@Juno.com; Erasmusen@Yahoo.com (for attachments). Web: Php.indiana.edu~erasmuse. Copies of this paper can be found at Php.indiana.edu~erasmuse/@Articles/Unpublished/harass.pdf. I thank Michael Alexeev, Robert Heidt, Vic Khanna, J. Mark Ramseyer, Larry Ribstein, an anonymous referee, and participants in vigorous discussions at Indiana University's Poynter Center for Ethics and the Law and Economics Lunch for helpful comments. I. INTRODUCTION This article will look at the objectives of the law of sexual harassment, with special attention to the issues raised by the 1998 Supreme Court decisions in Ellerth v. Burlington,1 Oncale v. Sundowner Offshore Services, Inc,2 and Faragher v. City of Boca Raton.3 I will suggest that the case law has little relation to the goal of preventing discrimination against women; that it has no basis in efficient employment law; that it can be better justified by its effect on parties uninvolved in the employment relation; and that it makes sense to apply traditional principles of agency law rather than strict liability or lack of positive efforts by the principal to control intentional torts by agents. The law of sexual harassment is relatively recent.4 The main line of federal cases comes to us via the Civil Rights Act of 1964 and the 1986 case of Meritor Savings Bank v. Vinson.5 In 1998, three major cases were decided by the U.S. Supreme Court: Ellerth, Faragher, and Oncale. Meritor distinguishes the exchange of employment favors for sexual favors, or "quid pro quo" sexual harassment, and rude behavior or language in the workplace that has a sexual element, or "hostile environment" sexual harassment. The first category might more accurately be termed "attempted solicitation of prostitution", and the second, "lewd language and behavior." Neither requires the intent to offend or continuous action that are key parts of harassment in the usual sense of the word.6 Neither is special to the workplace. When a man offers money to a woman on a city street for sex, that is solicitation of prostitution, a definite quid pro quo offer; if he makes lewd remarks to a group of ladies in the local mall he is certainly creating a hostile environment.7. This article will examine why and how law might be useful in reducing the amount of both kinds of workplace sexual harassment, with special discussion of the U.S. Supreme Court's 1998 decisions. II. AN OVERVIEW OF THE LAW Title VII of the Civil Rights Act of 1964 provides that it is "an unlawful employment practice for an employer . . . to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual's race, color, religion, sex, or national origin."8 Under Title VII, someone who believes he has been discriminated against first goes to the United States Equal Employment Opportunities Commission (the EEOC), which decides whether to pursue litigation itself. If it decides not to, the person who filed the claim may sue on his own behalf in federal district court.9 Until 1991, only lost wages could be recovered under a Title VII claim, but in that year new legislation allowed recovery of compensatory and punitive damages up to a limit varying from $50,000 to $300,000 depending on the size of the company.10 In addition, the court may award attorney's fees.11 What constitutes actionable harm has evolved over the years. The most important case up until 1998 that dealt specifically with sexual harassment was Meritor. In that case the plaintiff bank employee claimed that her supervisor had created a hostile environment for her at her workplace by repeatedly demanding sexual favors, demands with which she complied, and committing a variety of other harassing and indecent acts. She did not report his behavior to anyone higher up in the bank. Some time after the problem had ceased and when she was no longer employed there, she sued both the bank and the supervisor for compensatory and punitive damages. Meritor established that not just quid pro quo but also hostile environment sexual harassment, involving intangible harm, was actionable. The Court also held that employers might be liable for sexual harassment by their supervisors even if they are not informed of the harassment, and that "agency principles" should decide whether the employer is liable for the misdeeds of his agent. Oncale, Ellerth, and Faragher attempted to clarify the ambiguities left by Meritor. In Oncale, the plaintiff sued his employer, two supervisors, and a co-worker from his eight-man offshore oil rig crew for offensive homosexual behavior, including physical assault. After his complaints to the employer, Sundowner, were ignored, he quit, and sued, alleging that he was discriminated against because of his sex.12. The U.S. Supreme Court held that Oncale was not barred from his suit merely because both he and the defendants were male, though he would have to show that their behavior was "discrimination on the basis of sex". In Ellerth, plaintiff sued her employer, Burlington, because her supervisor had made offensive remarks and threats to her, although the threats had not been carried out. She never complained to Burlington, or used its sexual harassment grievance procedure. She quit Burlington, and then sued saying that Burlington had forced her constructive discharge. In Faragher, plaintiff sued her employer, the City of Boca Raton, and supervisors alleging that the supervisors had made offensive remarks and threats, although again the threats were not carried out. She was unaware of the city's sexual harassment policy, and did not notify the City of the supervisors' behavior, though she did complain informally to another supervisor. Ellerth and Faragher arose because Meritor had left it unclear when the employer would be liable for harassment by supervisors if the victimized employee did not notify the employer. In the Seventh Circuit's en banc opinion in Ellerth, Judge Posner, in discussing the use of the Restatement of Agency to determine when employers are liable for the torts of the supervisors they hire, wrote that It is time that we threw away the crutch of the Restatement and, recognizing the differences between workplace sexual harassment and the actual subject matter of that antiquated screed, ask as an original matter--as I think the Supreme Court invited us to do in Meritor when it understated that the common law of agency might not be fully transferable to sexual harassment--what the best regime of liability would be for these cases.13 The rule the U.S. Supreme Court applied in both Ellerth and Faragher is as follows. An employer is subject to vicarious liability to a victimized employee for an actionable hostile environment created by a supervisor with immediate (or successively higher) authority over the employee. When no tangible employment action is taken, a defending employer may raise an affirmative defense to liability or damages, subject to proof by a preponderance of the evidence, see Fed. Rule Civ. Proc. 8(c). The defense comprises two necessary elements: (a) that the employer exercised reasonable care to prevent and correct promptly any sexually harassing behavior, and (b) that the plaintiff employee unreasonably failed to take advantage of any preventive or corrective opportunities provided by the employer or to avoid harm otherwise. . . . No affirmative defense is available, however, when the supervisor's harassment culminates in a tangible employment action, such as discharge, demotion, or undesirable reassignment.14 Thus, if a supervisor engages in quid pro quo harassment, the employer is strictly liable. If the supervisor creates a hostile environment, the regime is strict liability with a limited defense of employee negligence. The careful employer can escape liability if the victimized employee did not take reasonable measures to report or avert the harm; the careless employer is liable even if the employee was equally careless. III. THE OBJECTIVES OF SEXUAL HARASSMENT LAW. A. Sex Discrimination The law has taken curious turns in reaching its present attitude towards sexual harassment. Judges are stretching Title VII to cover wrongs that are not within the original meaning of sex discrimination, in a way reminiscent of how in Blackstone's day legal fictions were used to wedge desirable policies into existing laws. Blackstone notes how the writ of Trespas was used to bring torts to the King's Bench, but the trespassing allegation was always then dropped, and the tort pursued.15 Just so, we now hang sexual harassment onto the writ of Title VII. Blackstone's Gothic castle of the law has its advantages, but we would do well in the present day to develop a principled basis for sexual harassment law. Sexual harassment law is, on its face, a way to fight sex discrimination, and in some cases sex discrimination indeed is involved. A clear example would be if an employer encouraged g male workers to engage in aggressive sexual banter and horseplay to drive female workers to quit. This was not the problem in Oncale, Ellerth, or Faragher, however. Rather, the connection with Title VII and sex discrimination in those cases is that a certain employee is subject to offensive behavior to which a member of the opposite sex would not have been subject.16 Under this logic, if a bisexual supervisor terrorized both male and female employees with demands for sex, the law should hold that behavior harmless.17 Even in these cases, however, the employee's sex is not the only, or even the primary, reason for victimization. Usually the harasser only harasses beautiful women. If a man supervises 100 men and 100 women, and harasses the one woman with red hair, it is peculiar to say that he is discriminating on the basis of sex rather than hair color. The motive is sexual desire for an individual, not an attitude towards a class of people. This kind of victimization would continue even if supervisors were bisexual, or if we were a race of intelligent flatworms having only one sex but propagating sexually.18 Although there may be good reason to object to a practice that allocates benefits to attractive, immoral people rather than than unattractive moral people,19this is not discrimination on the basis of sex.20Moreover, even to the extent that harassment discriminates on the basis of sex, it is not clear that it discriminates against women as a class. If a male supervisor only gives promotions in return for sexual favors from female employees, male employee are at a clear disadvantage. To be sure, this is not the case if the offer of promotion for sex would cause unhappiness to most women, or if supervisors use their power as a stick rather than a carrot, threatening demotion rather than promising promotion to the attractive employees. The supervisor's incentives, however, give reason to believe that he will use the carrot more often than the stick. First, as will be discussed further below, employment is voluntary, and if an employer wants the employee to provide more of anything, whether it be hours of work, unpleasant conditions, or sexual favors, the employer must give up something to obtain the extra employee effort. This is most obvious if the harassment takes the form of offering to hire someone only if they will provide sexual favors; an employer cannot threaten someone who is not yet his employee.21 Second, sexual harassment is most often a concern when the employer has hired a supervisor to oversee the employee, and the supervisor takes advantage of his position to harass the employee. This, too, will be discussed in more detail below, but it has one implication that is relevant immediately: the supervisor has more reason to try to obtain sexual favors with promises than with threats. Ordinarily, the employer who hires both supervisor and employee loses if the supervisor reduces the efficiency of the company by promoting employees in return for personal favors Such promotions puts the wrong people in the wrong positions and also reduces all employees' incentives to work hard. In order to avoid punishment, the supervisor will try to hide his quest for sexual favors. If he demotes a woman for not acceding to his wishes, questions will be asked about why, and she has a strong incentive to bring the reason to the employer's attention. If the supervisor fails to promote someone who would otherwise have been promoted, the anomaly is equally apparent. If, however, he promotes a woman who is unqualified, she will not complain to anyone, and his misbehavior may well go unnoticed. Moreover, if a woman undeserving of promotion rejects his offer, and he fails to promote her, she has no case to make against him. The company will not care that he has not promoted her, and he can deny that he ever made the offer. In light of these advantages of the carrot over the stick, we should expect that quid pro quo sexual harassment helps women more than it hurts them. Since the focus of attention has been on the sexual harassment of women, this suggests that the public is worried about some harm other than sex discrimination.22 Hostile environment harassment also may not clearly be sex discrimination if the harasser persecutes both male and female employees in different ways. It may be discrimination where a workplace is made inhospitable to women precisely because they are women, in the hopes of driving them out or for the pleasure of tormenting them, and where men are not subject to torment. But, as discussed in Section C below, this may involve a different problem from discrimination that takes the form of hiring women in order to use company resources to gain sexual favors. Whether Title VII is correctly applied to the evils of sexual harassment or not, it is still useful to ask what other objectives sexual harassment law might be addressing. As discussed in the following sections, these include employee protection, employer protection, and public morality. B. Efficient Labor Law: Protecting Employees. Alternatively, sexual harassment law might be useful as a way to protect employees who are hurt by supervisors. The problem is that, like other working conditions, sexual harassment can be negotiated between employer and employee. If the harm is internalized in the employment relation, there is no market failure, and no need for the government to mandate a particular punishment scheme. Indeed, government-imposed liability might even lead to worse consequences for both parties. Let us start by examining the effect on the employee of sexual harassment law when the employer, or an agent acting pursuant to the employer's instruction, harasses the employee, as distinguished from harassment by a supervisor acting solely in his own interest. First, consider quid pro quo harassment. A pay raise in exchange for sexual favors is simply payment for services and no less efficient than when the employer gives a pay raise to the employee who tells the best jokes. It is the employer's money, and he is not bound to use it as a reward for hard work or intelligence. Similarly, if the employer wants to fire an employee for not providing sexual favors, that is no different from firing the employee for not agreeing to work harder for the same pay. As long as the firing does not violate the employment contract-- an obvious exception, which has long given the victim recourse to contract law for remedies-- the firing is not unfair or inefficient. Suppose a clerk can command a wage of $40,000 per year in the labor market, and can produce clerical services worth $30,000 to the employer. The employer could choose not to hire her, which is the efficient outcome if he cares only about maximizing profits. If her sexual favors, or talent for laughing at unfunny jokes, however, is worth 10,000 dollars or more to the employer but has opportunity costs of less than that to the employee, she may choose to work for the unfunny and lascivious employer. More likely, she will demand a pay premium for these activities, and we may observe her choosing to work for him at a wage of $45,000 instead. In any case, the terms of employment can be negotiated between worker and employer, who have a far better idea than the government of how much they value these things. Hostile environment harassment can be analyzed in the same way. Many workplace environments are unpleasant for a variety of reasons. Often the unpleasantness is unavoidable because of the external environment, such as staffing the Alaska post offices or because of other employees' obnoxiousness.23 But nobody is forced to work in Alaska or with the obnoxious. Employers who wish to maintain unpleasant environments therefore must pay a wage premium. Both employer and employee will rationally make the decision of how to trade off money against working in or trying to improve the unpleasant working conditions. Now consider an employer who is known for making crude jokes and offensive remarks to women and prefers to hire women. The market wage for a secretary is $20,000. He, however, is willing to pay up to $10,000 extra to be able to be offensive to a woman. If he offers a wage of $28,000, the job will be taken and kept by some woman for whom the extra $8,000 is worth the unpleasantness. Both employer and secretary are better off than if he did not hire her. Even if the secretary would not say that she enjoys her work, or likes it that her boss is offensive, she does like the package as a whole. If the law allowed secretaries to sue offensive employers, the employer would behave himself. Knowing he would have to behave himself, however, he would only offer the market wage of $20,000. The secretary would prefer $30,000 and the offensive remarks, but the law prevents her from getting that deal. The law ends up hurting the very person it was intended to protect.24 It is not uncommon for regulation to hurt the people it purports to protect. High standards for housing quality, for example, notoriously result in the disappearance of cheap housing, and, indeed, could not do otherwise.25 The justification for such regulation has to be found outside of the benefit to the supposed beneficiaries, those who are prevented from buying low-cost, low-quality housing. In the case of housing, one such external benefit might be that fewer cheap and unattractive buildings are built, making the city more attractive for richer people. We will see that a similar argument can be made for harassment laws. Assuming a labor market where all employees are paid the market wage, any loss from sexual harassment will be borne entirely by the employer, not the employee. Any employer who tries to offer a worse deal than the market wage will find himself with no employees. An employer who offers the usual wage with a hostile environment will not retain any workers. If we do not assume a competitive labor market, the argument still holds, but with a variant. Suppose that for some reason only one employer exists for whom a woman wishes to work. Since she is not a slave, however, she does still have the option not to work at all. This means that if the job is unattractive enough, she will still refuse it. In the extreme, if the job requires her to commit suicide in a painful way, she will reject it even if the alternative is unemployment with starvation. More realistically, when people say that a woman has no alternative but to work for a particular employer, they mean that switching would require a large pay cut, or would require her to take an entirely different kind of job. Suppose that Anne is a computer programmer generating $100,000 per year in profits for her employer, and that because her skills are highly specialized to the software her employer uses, her next best alternative job is to work in a fast food restaurant earning $15,000 per year. Suppose also that except for the pay, she is indifferent between the two jobs. In a competitive labor market, she would earn $100,000, but since her skills are firm-specific, the employer can get away with paying her much less. If we assume that she has no bargaining power, in fact, the employer will pay her a mere $15,001 for her computer programming-- just enough to keep her from quitting. But now suppose that the employer wants to subject her to obnoxious jokes, so obnoxious that an accurate estimate of the damage to her would be $1,000. He cannot continue to pay her just $15,001, or she will quit. He must pay her $16,001 to compensate for the unpleasant working conditions. More generally, lack of bargaining power does not alter the fact that an employer who wishes to provide worse working conditions must provide a higher salary. The employer will use his strong bargaining power to reduce the salary as much as possible in any case, but he will not be able to keep the salary as low if he wants to provide poor working conditions as well. Harassment may, of course, be a breach of contract. Indeed, if sexual harassment is unusual, its absence may be considered an implicit term in employment contracts. Harassment may, then, give rise to a claim for breach of contract. Suppose, for example, that Jane agrees to work for Richard with the understanding that the job has normal working conditions and will last for at least two years. She spends $5,000 to move from Chicago to Seattle for the job. Richard harasses her enough that she finds the salary inadequate and quits. If Jane is not compensated for her $5,000, wealth will not be maximized. She may have had offers in Chicago that were almost as lucrative as the Seattle job, and which she would have accepted had Richard been clear about the terms of employment. The $5,000 she spent is a social waste. But this is no different from other types of breach, such as the employee's failure to work as hard as promised or the employer's failure to provide a job as attractive as promised. Apart from her remedy for breach, the employee may simply be able to leave and look for another job. If an employee and employer only care about the short term, then any disappointment in the conditions of work will have limited effect. The wage can be renegotiated for the next day, at a rate appropriate to the unpleasantness of the work. But this is a problem if workers cannot easily switch jobs once they discover that their employment environment is not as pleasant as they expected. 26 Thus, if sexual harassment is a problem, it is not because of the actions themselves, but because the employer harasses the employee without providing bargained-for compensation. Employers make employees do many unpleasant things, but that is why they call it "work" and pay people for it. It is not the unusual job, but the typical one, which is unpleasant enough that people would not do it unless they were paid. Even if the employer does provide adequate compensation, however, so that both employer and employee were satisfied, it may be that many people outside the relationship would still object to both quid pro quo and hostile environment harassment. Even understanding that the employee freely consented, and even believing that the employee personally derives greater satisfaction from the extra pay than dissatisfaction from the extra duties, these observers would wish to prohibit the relationship. It is important to understand, though, that protection of the employee is a weak basis for sexual harassment law, inconsistent with rationality of employers and employees even when labor markets are imperfect. C. Efficient Agency Law: Protecting the Employer We have so far limited ourselves to situations in which the harassment is by or at the direction of the employer. What if the harassment is by a supervisor who is acting against the employer's interest? Do we need laws to protect the employer? Law is indeed necessary here, but to allow the employer and employee alike to obtain redress from the supervisor, not to allow the employee to sue the employer. . We certainly cannot protect the employer against the misdeeds of the supervisor by making the employer liable for those misdeeds. There is reason, however, to make the supervisor liable for harm, since he might otherwise obtain the benefit from harassment without bearing the cost. Assume the same hypothetical as in the previous section. A clerk can provide clerical services worth $20,000 to the employer of the company, but she is hired by the supervisor for $28,000 because she agrees to provide sexual services to him. Both the clerk and the supervisor are happy with this arrangement, but we now must ask whether it is still efficient, given that the employer pays the bill. It may be that the employer agreed to pay the supervisor $95,000 per year for supervisory services worth an expected $100,000. In that case, the employer would not have hired this supervisor if he had foreseen that the supervisor would expend an extra $8,000 for sexual perks. On the other hand, the employer might have agreed to pay the supervisor just $70,000 and to give him a fund of $25,000 with which to pay above-market wages for employees if he wishes to for any reason. In that case, the sexual (or other) perks are part of the arrangement and the outcome is efficient.27 Whether the supervisor asks for straight cash or for some nonpecuniary work benefit that is costly to the employer should not matter if both parties agree. The employer does not need sexual harassment law to protect it from authorized supervisor behavior, even if that behavior, considered in isolation, hurts the employer. The problem comes with supervisor actions that are costly to and unauthorized by the employer. If, unknown to the employer, the supervisor harasses employees, making the environment unpleasant and requiring the employees to be paid more or to be lost to the company, this is inefficient. It is just as if the supervisor were to demand monetary kickbacks from the employees, on top of his authorized salary. Agency law and criminal law may both be helpful to employers whose supervisors misbehave. Agency law allows the employer to sue the supervisor for damages. 28 Criminal law allows the employer to bring a complaint to the prosecutor, who may then prosecute and imprison the supervisor for particularly egregious kinds of misbehavior. If the sexual harassment rises to the level of rape, assault, or indecent exposure, the supervisor would be subject to criminal penalties. With regard to sexual harassment in particular, I do not know of any cases of employers suing supervisors, but it would be no different in principle from suing for other kinds of harm. If, as in Ellerth and Faragher, a supervisor harasses an employee, the employer as well as the employee can lose from lost employee productivity, under-promotion of the qualified, overpromotion of the unqualified, and recruiting and training costs to replace employees who quit or are fired. Assume, for example, that Supervisor Sam harasses employee Emily every day with hostile sexual comments. After six months, Emily quits. The employer must pay $2,000 to train a new employee, and Emily incurs $1,000 in costs to find a new job. Who should bear these costs? Punishing supervisor Sam-- the only party who benefits from the harm if losses lie where they fall-- is the obvious and efficient solution. Sanctions against supervisors are especially effective compared to sanctions against agents generally, because supervisors have more to lose. By virtue of their higher positions, they are less likely to be judgment-proof and more likely to care about being fired , about social sanctions, and about criminal penalties. All of these remedies will be discussed below in a later section. Protection of the employer does not justify making the employer himself liable for unauthorized harassment by the supervisor. The following argument can be made for why such liability could benefit the employer, but after stating the argument, I will show why it is fallacious: "The employer is hurt by the supervisor's sexual harassment, just as the employee is. Especially in the case of quid pro quo sexual harassment, however, the employer may not be able to discover the supervisor's bad behavior except at great cost. The employer would therefore like the employees to have an incentive to bring the supervisor's behavior to the company's attention. Employer liability provides this incentive. If the employer can be sued for the supervisor's sexual harassment, this will help the employer in two ways. First, it will give the employer more incentive to monitor the supervisor. Second, the employees will help the company monitor the supervisor, by bringing suit. Moreover, with this extra monitoring, the amount of sexual harassment will decline, and so the potential liability of the employer may in the end not result in many lawsuits anyway." This argument is flawed. First, the employer does not need legal liability to give the employees incentive to bring harassment to its attention. The employer can voluntarily offer rewards equivalent to the potential damages from lawsuits-- and employer and employee alike can avoid trial costs and legal fees, a substantial benefit. Second, the company can also voluntarily increase the amount of monitoring it does even if it has no legal liability, and with such liability the amount of monitoring will be inefficiently great. There is such a thing as too much employer time, money, and attention being spent combatting sexual harassment. One way to see this last point is by analogy to bribery of purchasing agents. Suppose that Supplier A might bribe an employer's purchasing agent to buy inferior products, at a cost of $10,000 to the employer. The employer opposes this, of course, and would devote up to $10,000 in extra monitoring to prevent this, spending the money on more elaborate accounting, extra paperwork, redundant approvals, and suchlike. Moreover, the employer would like it if other suppliers competing with Supplier A would help inform the employer of the bribery, and might offer implicit or explicit rewards for such information. If a proposal is made to now let other suppliers sue the employer for $6,000 when the purchasing agent unjustly awards a contract to Supplier A, will this increase efficiency and benefit the employer? No. The frequency of bribery will indeed fall. The employer will discover it more often, because it will be sued by suppliers more often. The employer will also increase its expenditures to combat bribery, spending up to $16,000 to prevent it. But spending that much to fight bribery is inefficient; one ought not to spend $16,000 to prevent $10,000 in damage, and despite the fact that damage falls, the overall result is undesirable. In the same way, increased employer liability for sexual harassment will induce the employer to take more precautions, and will reduce the amount of sexual harassment, but this is not necessarily desirable, because both lawsuits and extra precautions have extra costs. If they did not, optimal policy would be clear: impose legal liability for sexual harassment, but make the damage award equal to the entire value of the company, perhaps adding jail terms for all shareholders. This reductio ad absurdum would surely reduce the amount of sexual harassment, as employers strained to devote every resource to preventing it, and the amount of sexual harassment would become negligible. But this is clearly undesirable. It is true that the public might wish to reduce the amount of sexual harassment even if that reduced the profit of the employer as well as the satisfaction of the employee. In the same way, in fact, the public might wish to reduce the amount of bribery more than the employer does, simply to increase society's rewards to virtue and reduce its rewards for sin. But that is a different argument from protecting the employer, and must be dealt with in a later section. E. Sexual Harassment Law as a Civility Code. The employer, supervisor, and employees are not the only people interested in sexual harassment. People care about what other people are doing even when the behavior is consensual. Many would rather not live in a society in which John sells heroin to Mary, or Mary sells her body to John, or John provides Mary with hugs in exchange for being allowed to beat her. This kind of conduct may increase the joint wealth of the immediate parties but if it offends other people, it imposes a negative "mental externality." This is the economic justification for social regulation of many kinds. The idea of economic efficiency takes tastes as given-- they are simply preferences about the state of the world. A person's tastes, including his moral beliefs about other people's behavior, enter into wealth maximization just as surely as do his consumption preferences. A person's strongest preferences are ordinarily about his own behavior rather than those of other people. Richard cares more about the flavor of ice cream that he eats himself than about the flavor that David eats. To the extent that Richard cares about what David does, however, as measured in his willingness to pay to affect what David does, economic efficiency treats Richard's preferences about David's actions the same as David's.29 The presence of externalities does not necessarily imply that regulation will increase efficiency-- the magnitude of the preferences determines that-- but it disrupts the argument for the efficiency of laissez faire. Although an externality justification for government intervention may be based on tastes, that does not necessarily excuse every government regulation. Consider, for example, such government interventions as insider trading laws and occupational licensing. These cannot be justified on the basis of mental externalities, because nobody seriously pretends that the public would be bothered by the fact that some people trade stocks using private information unless that somehow affected prices and hurt other people, or that some people are barbers without having passed certifying examinations although they are just as competent as other barbers.30 Such regulations as those must stand or fall on other arguments. Nor is the mental externality argument a positive description of policy making such as one finds in public choice theories. Its justification is normative: that to maximize societal wealth, or to achieve efficiency, people's tastes regarding the behavior of other people need to be taken into account. Public choice theory and mental externalities interrelate to the extent that political decisions will tend to efficiency, since people will vote their tastes. On the other hand, the greatest beneficiaries of Title VII are plaintiffs' attorneys, who would expend considerable political effort to retain or expand the current law of sexual harassment, and the imperfections of the political system may give them undue weight. Normatively, however, a given person really thinks that efficiency is a social objective, the presence of mental externalities should make him ask whether regulation might not be justified. The idea of externalities applies easily to sexual harassment. Most of us find sexual harassment objectionable even if the parties involved are consenting. Even someone with no direct connection to the company is at last mildly displeased on hearing that someone has been promoted not for merit, but for sleeping with the boss. This might be because the observer feels sympathy for other employees who were not promoted, or it might be simply that the observer thinks the behavior is immoral. Both reasons are quite independent of the satisfaction of the boss and the promoted employee and of the financial well-being of the company. Moreover, the sexual element in the offense plays a part in the moral outrage, even if the dollar impact of the offense is the same as for some behavior not involving sex. People dislike bribery of purchasing agents, but they do not become as worked up about it as they do about quid pro quo sexual harassment. People seem to care very little about other people being in safe but unpleasant working environments with unpleasant bosses or unfriendly co-employees, but someone else working in an environment of sexual harassment arouses more sympathy. We need not explain here why sexual behavior arouses strong feelings, but if it does, then an externality is present that requires regulation if the efficient outcome-- considering all people in society, not just the employer and employee-- is to be attained. Most people would rather live in a society in which it does not occur, and would be willing to pay something to obtain that society. This is true even if they personally would like to engage in sexual harassment, or would be willing to accept harassment if paid enough. As with burglary or the income tax, my ideal society is one in which I am allowed to do anything I like, while everyone else is restricted. I may, however, prefer to prohibit everyone, including myself, from engaging in the bad behavior, rather than allowing everyone to engage in it. Thus, while individual citizens might find criminal or civil laws against sexual harassment annoying or constricting in their own lives, they might nonetheless rationally support such laws for society as a whole. This reasoning does, of course, make sexual harassment law into a civility code, exactly what Justice Scalia said it was not in Oncale: "Respondents and their amici contend that recognizing liability for same-sex harassment will transform Title VII into a general civility code for the American workplace. But that risk is no greater for same-sex than for opposite sex harassment, and is adequately met by careful attention to the requirements of the statute.''31 It is, to be sure, difficult to extract a civility code from the intent or plain language of Title VII, but from an efficiency point of view, civility is a legitimate motive for a law of sexual harassment. Just as public nudity is outlawed because it is displeasing to many people, so is sexual harassment. The reasoning is, of course, independent of whether sexual harassment occurs in the workplace or outside of it. Indeed, if offensiveness to the public is the grounds for punishing harassment, it ought to be punished more if it occurs in a public space rather than in the workplace, and we should see judges extending Title VII to courthouse squares and shopping malls. But we already have statutes prohibiting prostitution and its solicitation, which is analytically very similar to some forms of quid pro quo sexual harassment-- the employer's offering a promotion for sexual favors is identical to offering cash for sexual favors. We also have civility codes suitable for hostile environment harassment-- ,flashing, for example, has been long been banned outside the workplace, and anti-stalking laws are a example of a more recent innovations in civility regulation.32 More generally, the police have vaguely worded statutes at their disposal for the kind of behavior that sexual harassment law covers in the workplace.33 The civility code idea explains sexual harassment law better than other rationales. Even if sexual harassment is a victimless crime as far as employer, employee, and supervisor are concerned, the public loses from it. The civility code idea can explain the 1991 amendments, which allowed compensatory and punitive damages, but only up to a cap of from $50,000 to $300,000, depending on the firm.34 Limiting the compensatory damages and allowing punitive damages make the legal damages more like a fine in an amount that reflects the harm to the public. 35 The civility justification also explains why sexual behavior, not sexual discrimination, is the heart of sexual harassment law, something odd in view of the doctrine's formal legal justification.36 As I discussed above, if discrimination were really the concern, men should be bringing the quid pro quo lawsuits, and cases like Oncale would arouse no public sympathy. That sexual behavior has captured so much attention suggests that what offends the public is not the discrimination, but the sex. Thus, Mr. Oncale wins his case against his offensive co-workers not because they dislike men or because they discriminated against him as a man, but because they were disgusting and rude. III. Designing a Policy to Appropriately Deter Sexual Harassment A. Categories of Penalties Five ways to deal with harassment are criminal punishment, property-rule protection, liability-rule protection, social norms, and self-help.37 I will briefly describe each, providing an example of how the law might use the method to deal with an employer who persistently subjects a female employee to unwanted touching. (1)Criminal punishment is characterized by public prosecution in the courts and the possibility of either monetary or non-monetary punishment, with any fines going to the government. Example: the county prosecutor prosecutes the employer for battery, asking the court to jail the employer. (2) Property-rule protection or injunctive relief is characterized by private suit in the courts to obtain the remedy of a judicial command to cease the objectionable behavior, on pain of imprisonment. Example: the woman is given the statutory right to ask a civil court for an injunction requiring the employer to cease the touching. If he persists, she can ask the court to rule him in contempt. (3) Liability-rule protection or money damages is characterized by private suit in the courts to obtain the remedy of money damages for the victim. Example: the employee is given the statutory right to ask a civil court for money damages equal to the harm inflicted upon her. (4) Social norms are a private remedy in which the courts are not involved, the victim does not obtain compensation, and the perpetrator's punishment is inflicted by those people with whom he interacts either economically or socially. Example: the employee tells everyone she knows about the unwanted touching, and the employer finds that he and his wife are no longer invited to other people's houses for dinner. (5) Self-help is a private remedy in which the punishment is an act, otherwise illegal, carried out by the victim with the acquiescence of the courts. The employee slaps the employer after one incident of harassment, and quits in violation of her contract, but is allowed to raise the employer's harassment as a defense when he tries to use criminal prosecution and civil suit for money damages against her. Each of these has its own advantages and disadvantages. Criminal punishment is prosecuted by the public prosecutor, and the prosecutorial discretion this introduces is good if it prevents punishment of acts that are not harmful even if they are technical violation of the law, but bad if it Section II argued that the three best rationales for laws against sexual harassment are that harassment violates implicit terms of the employment contract, that if done by a supervisor without authorization by the employer it is similar to theft, and that it offends the sensibilities of the general public. The first rationale is essentially breach of contract. Breach of contract affects only the parties to the contract, and does not have externalities on the outside world. Thus, the victim can be trusted to undertake efficient actions to initiate corrective action if the remedy is money damages or an injunction to cease the harassment. Injunctions to the employer to cease persistent minor irritation, however, are difficult to enforce, and while an injunction to change employment status-- for example, to grant a promotion-- is simple enough, it is hard for a court to monitor retaliation by the employer in such a complex relationship as employment. Money damages therefore seem appropriate if resort is to be had to courts. These compensate the victim, unlike criminal penalties and do not require public resources for prosecution. Judgment-proofness of the tortfeasor is a common drawback of money damages. Judgement-proofnesss reduces the deterrence effect of civil liability, although for deterrence it may be enough that the defendant is reduced to penury. More important, the defendant's inability to pay removes the plaintiffs' incentive to sue. 38 The problem is less severe for sexual harassment torts than for many other torts, however. Employers usually have ample resources, and even supervisors are likely to be wealthier than the average person, not just because of a higher salary but because they are likely to be older than other employees. On the other side of the balance sheet, the size of the sexual harassment judgement is smaller than in torts that create incapacitating harm to the victim. Even a simple automobile accident may result in medical damage of hundreds of thousands of dollars, so that an uninsured defendant is judgement-proof. A hostile work environment, on the other hand, is less likely to exhaust a supervisor's ability to pay damages. Not only is the damage less than from death or crippling, but because it depends on personal interaction between harasser and victim it cannot be so great as when an employee misuses company resources to dump toxic waste or otherwise damage thousands of citizens or customers. Lawsuits are expensive, however, and even for ordinary breach of contract in which damages are legally available. the most important incentives are often social norms or self-help. A seller who delivers shoddy goods will often find that the hurt to his reputation is far more of a disincentive than the threat of an action for breach of contract. He may also face retaliation from the buyer, to whom the Uniform Commercial Code explicitly grants the self-help rights of anticipatory repudiation if breach seems likely and to sale of certain of the seller's goods if breach has occurred.39 The same disincentives can be effective for use against sexual harassment. An employer's reputation is hurt if he treats his employees contrary to their expectations, and if he has no resort against a harassed employee who suddenly quits, he will likely incur high costs to harassment.40 These remedies have the advantage of much lower transaction costs, since lawyers and courts do not have to be involved-- not just because of the cost of paying legal and court staff, but because the reputations of innocent parties can suffer when evidence is subpoena'd and made public in litigation41 They have the disadvantage that risks to reputation fail to deter employers who are near bankruptcy or for other reasons have short time horizons, that in some cases no effective self help is available, and that they will usually fail to compensate the employee for the harm done. The second rationale, that harassment by the supervisor harms the employer, is also a matter between two private parties, and so the benefits and costs of the various corrective policies are similar. The Restatement of Agency (Second) lays out the common law remedies in Section 399, "Remedies of Principal". "A principal whose agent has violated or threatens to violate his duties has an appropriate remedy for such violation. Such remedy may be: (a) an action on the contract of service; (b) an action for losses and for the misuse of property; (c) an action in equity to enforce the provisions of an express trust undertaken by the agent; (d) an action for restitution, either at law or in equity; (e) an action for an accounting; (f) an action for an injunction; (g) set-off or counterclaim; (h) causing the agent to be made party to an action brought by a third person against the principal; (i) self-help; (j) discharge; or (k) refusal to pay compensation or rescission of the contract of employment." Thus, the employer can seek injunctive or monetary relief from a harassing supervisor, which have the same advantages and disadvantages as when the employee seeks such relief from a harassing employer.42 Just as the employee can punish the employer by hurting his reputation and by quitting suddenly, so the employer can punish the supervisor by hurting his reputation and by discharging him suddenly. Since the employer is a larger entity than the supervisor, these punishments are likely to be larger relative to his wealth than when the employee punishes the employer.43 The third rationale, the problem of mental externalities requires a different approach. If the public shares in the harm from sexual harassment, then civil suits for injunctions or money judgements equal to the damage inflicted will not provide adequate deterrence. The employer and employee could come to an agreement that benefited both but still caused harm to the public in excess of their benefits. The employee's decision to sue, if agreement is not reached, would be based on the employees calculation of a judgement equal to the private damage, which would be smaller than the public damage, and some suits useful for social deterrence would not be brought.44 Moreover, judgement-proofness of the harasser becomes a much more important problem if the damage is not just the unpleasantness of the work environment to one person but the offense to a large number of members of the public. There is thus a much stronger case for criminal penalties if the law's objective is to correct for mental externalities. Prosecution is at the discretion of the public prosecutor. Since his decision to bring cases depends on how they will affect his overall record with the voters rather than on the probability and amount of damages recovered, the decision to bring suit is decoupled from the penalty inflicted on the harasser. Since imprisonment is a possible penalty, the possible judgement-proofness of the harasser does not prevent his punishment. The legal process is still costly, but the prosecutor takes into account not only his own costs, but those of the government court, and directly considers the deterrent effects, not just the benefits obtainable in the particular case at bar. B. The Problem of the Harassing Supervisor Suppose criminal penalties are ruled out. . What then should be done? Should the supervisor be liable? How should the loss should be split between the employer and the employee, if the supervisor is judgment-proof? The division of losses is even more relevant to sexual harassment by fellow servants, who will have less wealth than supervisors.45 Should we let losses lie where they fall, or try to reallocate losses among the victims? A standard question in the economic analysis of tort law is which party could avoid the harm at lowest cost. When an auto strikes a pedestrian, the court must decide whether to let the loss lie on the pedestrian, to impose it entirely on the driver, or to somehow split the loss. To put the loss on the party at "fault" is meaningless without a definition of fault. One definition is to ask who could have avoided the harm most easily and to say that party, the least-cost-avoider, is at fault. Since both parties could have avoided the harm, the court must look at their relative costs of avoiding harm, not just whether they could have prevented it. The pedestrian could have avoided the accident by looking out more carefully, jumping at an appropriate moment, or by not walking at all. The driver could have avoided it by looking out more carefully, swerving, or not driving at all. Imposing the loss on whoever could have avoided it at least cost is a simple way to providing more efficient incentives than simply letting losses lie where they fall. The question is then whether the improvement in incentives is worth the cost of using the courts. This idea can and has been refined in a large literature that has grown up exploring the effects of negligence, comparative negligence, and strict liability on incentives.46 Certain problems are clear. If we impose strict liability or no liability on the driver, that entirely eliminates either the driver's or the pedestrian's incentives for care, and can actually increase the amount of harm.47 If we must impose the entire loss on one party or the other, as is traditional in the common law,48 we should impose it on the party who could have avoided the accident at least cost.49 The same principle can be applied to the harm of sexual harassment.50 In the example of Sam and Emily, the employer could perhaps have prevented the harm by taking care in hiring supervisor Sam, by watching him more carefully, or by disciplining him before employee Emily quit. Emily could perhaps have prevented the harm by behaving differently towards Sam or by notifying the employer of Sam's behavior. In determining liability according to the least-cost-avoider rule, the court would try to fashion a rule that imposes liability on whichever party could have avoided the harm at least cost. Sam himself has the least cost of avoiding harm of any of the three parties, which is why ordinarily liability for intentional torts lies squarely on the tortfeasor. But here we are assuming Sam is unreachable. The usefulness of the least-cost-avoider idea is even greater for sexual harassment than for torts generally because sexual harassment occurs in the context of an underlying contract. The aim of imposing liability on the least-cost-avoider is to maximize the sum of the surplus of all the parties involved. In the context of most torts, this has the deficiency that it does not necessarily make all parties better off, since they do not meet each other in advance and cannot make side payments to balance the inequalities created by the legal liability rule. The efficient tort rule might, for example, favor pedestrians at the expense of drivers, and all we can say is that most people occupy both roles at different times and so would favor a rule that maximized total surplus. In harassment, however, any imbalance the legal rule creates can be adjusted in the underlying contract. If the legal rule requires the employer to exert much more effort than the employee to prevent harassment, for example, bargaining between employer and prospective employee will lead to lower wages than if the legal rule favored employees. Since side payments in the form of wage adjustments will split whatever gain in total surplus is achieved by using the efficient rule, both employer and employee should favor the efficient rule. If we are faced with the two alternatives of imposing liability for the employee's $1,000 loss on the employer or letting losses lie where they fall, it is appropriate to ask who is the least-cost avoider, employee or employer.51 If the employer knew about the harassment and did nothing, as in Oncale, we may conclude that the employer should be liable for the employee's loss. Ratification of an agent's action by a principal makes him liable for damages from that action, and may even be evidence that the action was authorized in the first place. C. The Scope of Employment Rule The common law of agency makes principals liable for the torts of their agents when the agents are acting on their behalf-- in the "scope of employment".52 But when is the agent acting on behalf of the principal? Section 228 of the Restatement (Second) of Agency defines the "scope of employment" as follows: (1) Conduct of a servant is within the scope of employment if, but only if: (a) it is of the kind he is employed to perform; (b) it occurs substantially within the authorized time and space limits; (c) it is actuated, at least in part, by a purpose to serve the master, and (d) if force is intentionally used by the servant against another, the use of force is not unexpectable by the master. (2) Conduct of a servant is not within the scope of employment if it is different in kind from that authorized, far beyond the authorized time or space limits, or too little actuated by a purpose to serve the master. Of particular importance here are the requirements that the servant's conduct be "of the kind he is employed to perform" and actuated "by a purpose to serve the master." The second of these is amplified in Section 235, which uses the following illustrations: @ 235 (a) Illustration 3. P, a railroad company, employs A as a freight brakeman and instructs him to eject trespassers from the train. A permits persons to ride upon the freight train upon payment to him of a small amount. He ejects T, one of A's regular customers, for failure to pay the bribe. A's act is not within the scope of employment. @ 235 (c) Illustration 4. A is employed to eject trespassers. A boy of four seeks to enter the premises. A could easily prevent the child from so doing by calling to him. Instead, he shoots at the child and kills him. This evidence indicates that A was not actuated by an intent to perform his master's business and hence that his act was not within the scope of his employment.53 Because the agent has to be plausibly acting on behalf of the principal, application of agency law to Title VII sexual harassment has trouble for the courts. It has been pointed out that Employers should virtually never be liable for sexual harassment if respondeat superior is the agency principle of choice because such behavior is rarely, if ever, within the scope of employment under the Restatement's criteria.54 One court wrote with respect to racial harassment, "It would be the rare case where . . . harassment against a co-worker could be thought by the author of the harassment to help the employer's business."55 Another court has written that "...confining liability . . . to situations in which a supervisor acted within the scope of his authority conceivably could lead to the ludicrous result that employers would become accountable only if they explicitly require or consciously allow their supervisors to molest women employees."56 When an agent deliberately takes an action that the principal has forbidden and that cannot possibly benefit the principal exempting the principal from liability is hardly ludicrous. Few employers sacrifice profits in order to put women at the mercy of lecherous supervisors. Thus, sexual harassment is "too little actuated by a purpose to serve the master" to fall within the scope of employment as the Restatement defines it. As discussed earlier, however, the employer may be willing to allow harassment as a perk of the supervisor, something which costs the employer by reducing productivity but helps the employer by allowing him to pay supervisors lower wages. Then, the harassing supervisor is indeed acting within the scope of his authority, and harassment by the supervisor can properly be imputed to the employer.57 The employer has ratified the harassment. And indeed, the common law has made principals liable for the torts of their agents if the principal's care could easily have prevented the harm- even intentional torts outside the scope of employment. Section 317 of the Restatement (Second) of Torts says,58 A master is under a duty to exercise reasonable care so to control his servant while acting outside the scope of his employment as to prevent him from intentionally harming others or from so conducting himself as to create an unreasonable risk of bodily harm to them, if (a) the servant (i) is upon the premises in possession of the master or upon which the servant is privileged to enter only as his servant, or (ii) is using a chattel of the master, and (b) the master (i) knows or has reason to know that he has the ability to control his servant, and (ii) knows or should know of the necessity and opportunity for exercising such control.59 As Judge Posner wrote (in a railroad tort opinion), "...plaintiff may prevail in an intentional tort case by showing either that the intentional tort was committed in furtherance of the employer's objectives or that the employer was negligent in hiring, supervising, or failing to fire the employee."60 Just what negligent hiring is depends, as negligence always does, on the circumstances, and there often exist state statutes specifically delineating what employers of people with particularly great opportunities for crime must do.61 The issue is not whether the tortfeasor is working on behalf of the employer, but whether the employer has carelessly given him the opportunity to commit the tort. Such negligence could apply to a customer or a volunteer who was given access to criminal opportunities just as easily as an employee.62 In the context of sexual harassment, an employer can do little to avoid it via careful supervisor selection except to check for rape convictions and read reference letters for lurid stories.63 The main scope for negligence would come rather in learning about supervisor misbehavior and irresponsibly ignoring it, in which case negligence comes to a result very similar to the agency doctrines of apparent authority and ratification.64 The victimized employee also has an important role in preventing and correcting sexual harassment. An employee who failed to complain about the hostile environment at the time, waiting until it was too late for anyone to take corrective action, should not be compensated. Indeed, it is more reasonable to require that the employee should be liable for damages to an employer who has been hurt by harassment that the employee could easily have tried to stop as to require that the employer be liable to the employee for unauthorized harassment by a supervisor. The employer has been hurt by the necessity of finding and training a new employee, a cost that could have been avoided if the old one had complained about the supervisor instead of just quitting. Neither employer nor employee, however, has as much control over an intentional tort as the actual tortfeasor, the supervisor, and the small increases in care that would result from imposing tort liability on either of them is unlikely to be worth the transaction costs. Since we know that creation of tort liability will lead to litigation costs, we should not create it unless we think the improvement in incentives is substantial, especially in a context such as employment where the parties can create liability by contract. Thus, imposing liability for the sexual harassment of supervisors on employers is a mistake unless it will clearly improve the employer's incentive substantially, while reducing the employee's incentives no less substantially. If it is unclear which party is the least-cost-avoider, the law should let losses lie where they fall. Clearly, the supervisor's incentives are improved if he becomes personally liable for an intentional tort such as sexual harassment (unless he is judgement-proof, in which case there are no court costs either, since he will not be sued). Equally clearly, the supervisor's incentives are diminished if the employer is made liable, since litigation will be shifted away from him and towards the employer. 65 If the employer has ratified the supervisor's actions, either explicitly or by obvious acquiescence, the employer's incentives are improved if he becomes liable. Otherwise, if the employer cannot prevent the harassment, and is nonetheless made liable for it, the punishment has been put on the wrong party. More important, when a person has little influence over whether a tort occurs, to impose tort liability results in little extra care but in the considerable extra transaction costs of the legal process. D. Other suggested rules More complicated liability rules than imposing liability on employer, supervisor, or employee are possible. A recent article by Jennifer Arlen and Reinier Kraakman contains a useful classification scheme for liability rules concerned with giving the employer incentive to investigate and report the intentional torts of its agents.66 They note that strict liability may give the employer very weak incentives to investigate because it can uncover evidence against the employer.67 A liability standard would be duty-based, on the other hand, if it exempts the employer from liability if he undertakes a reasonable amount of investigation. Mixed regimes of adjusted strict liability and composite liability can ameliorate the perverse incentives of strict liability. Adjusted strict liability is the same as strict liability except that it protects the firm from the effect of the firm's investigative measures. An example is a privilege that protects the firm's internal disciplinary records from discovery. Composite liability separates liability for the basic wrong and liability for failure to investigate or prevent the wrong. An example is to have a small penalty for the basic wrong but a large additional penalty for failure to investigate the wrong. Arlen and Kraakman deals, however, with intentional torts that benefit the employer and for which the victim's degree of care is unimportant, such as product adulteration or illegal dumping of waste. Without liability the employer would actually encourage the torts of his agents. If the victim's care is unimportant strict liability is the least-cost avoider rule. Sexual harassment by a supervisor, by contrast, does not benefit the employer and can be avoided by the victim who plays a crucial role by reporting violations. Despite these differences, the Arlen-Kraakman analysis is relevant to harassment. A rule that allows evidence generated by internal disciplinary procedures to be used against the employer or fails to reduce the employer's liability in proportion to his efforts will deter efforts he would otherwise make to discover whether harassment has occurred. Based on this analysis, it will be seen that the Court's rule in Ellerth is perverse. IV. The Supreme Court's Rule in Ellerth The Supreme Court's Ellerth rule for hostile environment sexual harassment, quoted at the start of this article, does not simply put liability on the employer in case of supervisor harassment, nor does it fit Arlen and Kraakman's classification. It is duty-based liability, but rather than the plaintiff's having to show that the defendant was negligent, the defendant must show that he was not negligent. The burden of proof is significant. Think how the Ellerth rule would apply to automobile accidents. Under a negligence standard, the plaintiff would have to show that the defendant was negligent- that he had gone through a red light, for example. The plaintiff could not simply provide the court with evidence that his car was damaged by the defendant. Under the Ellerth rule, the plaintiff would provide the evidence that the defendant had damaged his car, and having presented that evidence, he would be done. The defendant would have to show that (a) he, himself, had been taking reasonable care when he drove, and (b) the plaintiff was negligent. Both points (a) and (b) would be difficult to prove both because evidence from the time of the accident might be unavailable and even if it were not, showing that the level of care was a reasonable level could be difficult, given that an accident did happen. Note too, that even if the defendant could show that the plaintiff had negligently driven through a red light, that would not excuse the defendant from showing that he was driving with care. Rather, the defendant would have to show that even though the plaintiff was at fault, the defendant could not have prevented the accident by reasonable care to correct for the plaintiff's mistakes. When the harm is due to sexual harassment, under a negligence rule, the plaintiff would have to show that the employer was negligent--- that the employer could have undertaken to suppress the supervisor's harassment but did not do so. An example of this might be evidence that the employee complained to a superior who failed to investigate. Under the Ellerth rule, the plaintiff just needs to show that harassment occurred. The defendant employer must then show that he took reasonable care to avoid and correct harassment, and that the plaintiff did not take reasonable care. If little evidence is available the employer loses. If it is not clear what "reasonable" means, then even an employer with plentiful evidence will have difficulty in showing that his care was reasonable. In hindsight, it may be difficult to convince a sympathetic jury that any level of care is reasonable if the employer's care failed to work and allowed harassment to occur in a particular instance. Indeed, to see the difficulties in the Ellerth rule we need only look at Ellerth and Faragher, the two cases to which it was first applied. In both, the Supreme Court ruled that sexual harassment had occurred but that since no tangible employment action such as firing occurred the employer could raise the defenses of reasonable employer behavior and unreasonable employee behavior. In Ellerth, even without evidence that Burlington's sexual harassment policy was faulty, Burlington must show on remand that it undertook all reasonable steps-though it is not clear what steps those are-to prevent and correct sexual harassment. This means that Ellerth can obtain a settlement from Burlington even if her case would be weak once the evidence is assembled. Ellerth has some chance of winning because of the issue of whether Burlington's behavior was reasonable, simply because what is reasonable is so ill-defined. She also has a smaller burden of proof and hence smaller legal costs. Her threat to carry the case through to trial is therefore credible, and since Burlington has more to lose by continuing expenditures on the case than she does, even if Burlington is more likely than not to win, Ellerth is in a good bargaining position in settlement negotations. More generally, in cases like this, putting the burden on the defendant will, by increasing prospective defendant legal costs, make settlements more advantageous to plaintiffs, which in turn will lead more plaintiffs to file suit, even if their cases are weak. In Faragher, the City of Boca Raton did have a formal policy concerning sexual harassment. Two months before plaintiff Faragher quit, another woman, a former employee, wrote to the City Personnel Director, who began an investigation that ended in the disciplining of the two harassing supervisors. 68 Faragher did not formally complain, but informally told the third of the supervisors, who advised her that the City did not care. 69 From this, could a court say as a matter of law that the City had a reasonable policy against sexual harassment, and that Faragher was unreasonable in not taking advantage of it? Or would more factfinding have to occur, to determine whether there were other steps the City should have taken? The Court said: While the City would have an opportunity to raise an affirmative defense if there were any serious prospect of its presenting one, it appears from the record that any such avenue is closed. The District Court found that the City had entirely failed to disseminate its policy against sexual harassment among the beach employees and that its officials made no attempt to keep track of the conduct of supervisors like Terry and Silverman. The record also makes clear that the City's policy did not include any assurance that the harassing supervisors could be bypassed in registering complaints. App. 274. Under such circumstances, we hold as a matter of law that the City could not be found to have exercised reasonable care to prevent the supervisors' harassing conduct. Unlike the employer of a small workforce, who might expect that sufficient care to prevent tortious behavior could be exercised informally, those responsible for city operations could not reasonably have thought that precautions against hostile environments in any one of many departments in far-flung locations could be effective without communicating some formal policy against harassment, with a sensible complaint procedure.70 This is surprising and shows that lower courts will have trouble administering the Ellerth rule. It suggests that whether having a written sexual harassment policy for the kind of behavior at issue in these cases is not worth the cost of the paper it is written on. It would be the exceptional workplace in which an employer would approve of such behavior, and formal policies are not necessary to state the obvious. Nor in Faragher is there any indication that a more cumbersome procedure would be useful. When a complaint was made by somebody else to the administrator in charge of personnel matters, the complaint was handled well. Nevertheless, though the City already has a "formal policy," the Court wants a "sensible complaint procedure," which it seems would include "assurance that the harassing supervisors could be bypassed in registering complaints." We are not told how much assurance is enough. As Justice Thomas says in dissent: The Court today manufactures a rule that employers are vicariously liable if supervisors create a sexually hostile work environment, subject to an affirmative defense that the Court barely attempts to define. This rule applies even if the employer has a policy against sexual harassment, the employee knows about that policy, and the employee never informs anyone in a position of authority about the supervisor's conduct. . . . . Although the Court recognizes an affirmative defense--based solely on its divination of Title VII's gestalt, see ante, at 19--it provides shockingly little guidance about how employers can actually avoid vicarious liability. Instead, it issues only Delphic pronouncements and leaves the dirty work to the lower courts.71 The Ellerth rule suffers not only from vagueness, but also poor incentives. Indeed, the vagueness discussed in the previous paragraphs will exacerbate the incentive problem, since an unclear rule weakens the incentives for employer and employer alike to act in accord with the rule's purpose. If trying to act in accord with a legal rule might still result in liability, because the prescription of the rule is unclear, the rule will inevitably have less influence. Even where the rule is clear it provides poor incentives. Making it difficult for employers to escape liability gives them little incentive to try, particularly in light of the disincentive to collect information that Arlen and Kraakman discuss. This is particularly true of the rule that holds the employer strictly liable for quid pro quo harassment that results in a tangible employment action such as demotion.72 Even if (a) the victimized employee does not report the harm, (b) the employer could not possibly have prevented it without notice from the employee, and (c) the supervisor's employment action hurts the employer more than the employee, the employer is still liable.73 74 Incentives for the supervisor and the employee are also poor. The Ellerth rule reduces the expected costs to supervisors by increasing the cost to employers, whose deeper pockets will make them the focus of litigation. Moreover, if employer and supervisor are joint defendants, the supervisor will benefit from legal expertise provided by the employer, who might otherwise happily leave him to his own resources. A. Possible alternatives to the Ellerth rule What alternative might the Court consider? One alternative would be for the Court to define reasonableness. A basis for determining what is reasonable might be to discover what kind of policies employers have for their employees to report other kinds of supervisor misbehavior that hurt the employer, such as embezzlement, alcoholism, sleeping on the job, promoting friends instead of good workers, and so forth. These would be suitable models because we could expect employers to take reasonable steps to guard against behavior when they bear the costs. Since only part of the cost of sexual harassment falls on the company, the law could require companies to bring their sexual harassment policies up to the level of their policies for these other offenses. If employers do not have formal policies for employee reports of supervisor alcoholism, this suggests that they do not find it cost effective to establish formal policies. A similar inference might be drawn regarding sexual harassment. Another alternative would be to treat sexual harassment the same as other intentional torts, and make the employer liable only when the supervisor is acting within the scope of employment. The most important test of whether the supervisor is acting within the scope of his employment would be if the employer ratified his actions or gave him the apparent authority by not acting when aware of them. This would be very similar to the standard of "deliberate indifference," which is similar to criminal recklessness and has been applied to decide whether a prison is liable for the actions of one prisoner against another. As Justice Souter wrote in Farmer vs. Brennan: We hold . . . that a prison official cannot be found liable under the Eighth Amendment for denying an inmate humane conditions of confinement unless the official knows of and disregards an excessive risk to inmate health or safety; the official must both be aware of facts from which the inference could be drawn that a substantial risk of serious harm exists, and he must also draw the inference.75 Judge Posner has suggested that this standard be applied to Title IX sexual harassment cases involving harassment of one high school student by another: eliberate indifference by the school in a case of one student sexually harassing another would mean that the school (1) actually knew of (2) hostile or offensive conduct likely to interfere with the victim's education, and (3) deliberately did nothing, or took steps that it knew would be ineffectual, to protect the victim, (4) without excuse (for it might be difficult or even impossible to take effective measures).76 Requiring that an employer actually knew of supervisor sexual harassment and could reasonably have responded yet did not do so would give the employee an incentive to provide notice, and would give the employer incentive to act on this notice. Retaining liability of the supervisor would, together with the threat of punishment by the employer, still give him incentive to refrain from harassment. Vexing questions-but inevitable ones if we are to use the law to control an evil-- would remain of what constitutes sexual harassment and what constitutes excuse, but by putting the burden of proof on the plaintiff, litigation would be kept to a minimum, and used only where the employer clearly could have avoided the harm at low cost. B. The Problem of Bureaucratization The standard of ratification or deliberate indifference discussed above avoids another problem which may be exacerbated by both the appellate and Supreme Court. If by reasonable care we mean the efficient level of care, or industry practice, or what common sense dictates, reasonable care may mean no explicit care at all. Consider the analogy to the problem of supervisors who demand kickbacks from employee's wages, using threats of firing them. What is the reasonable level of care for the employer to deter this kind of behavior? Everyone already knows that it is wrong, and that the employer would disapprove if he knew of it, so promulgation of an anti-kickback policy would incur a useless cost. Promulgation might backfire, in fact, because, having seen express disapproval of only this kind of obviously bad behavior, supervisors might conclude that other kinds of graft are allowed. Nor is it cost-effective for the employer to monitor the bank accounts of all its employees or send out complaint forms weekly in order to detect the one supervisor in a thousand who may be trying to extort kickbacks. The reasonable level of care by the employer to prevent kickbacks is to do nothing until the issue arises, but to listen to employee complaints and investigate if appropriate. But would this seem reasonable to a judge or jury after the fact? Looking not at 999 honest supervisors and 1 dishonest one, but at only the dishonest one, it is easy to fall into the trap of thinking that it is reasonable to monitor the supervisor closely.77 The same argument can be applied to many other forms of supervisor misbehavior. Although the employer could reproduce the entire U.S. Code and state statutes in a company policy for supervisor behavior, it is more reasonable to do nothing and trust that supervisors know the company does not approve of illegal behavior. Similarly, the reasonable level of employer care to prevent and correct sexual harassment may be zero. But whether a jury will accept that is another matter, especially when the plaintiff is a sympathetic and beautiful woman, the behavior alleged is grossly immoral, and the defendant is a large corporation. Yet another problem exists, however, in the kind of company policies that the Court is encouraging. Consider the following part of the Ellerth rule: While proof that an employer had promulgated an anti-harassment policy with a complaint procedure is not necessary in every instance as a matter of law, the need for a stated policy suitable to the employment circumstances may appropriately be addressed in any case when litigating the first element of the defense. And while proof that an employee failed to fulfill the corresponding obligation of reasonable care to avoid harm is not limited to showing any unreasonable failure to use any complaint procedure provided by the employer, a demonstration of such failure will normally suffice to satisfy the employer's burden under the second element of the defense.78 The Court suggests that the employer can plead as partial defenses (a) the existence of a formal policy against sexual harassment, with grievance procedures, and (b) the failure of the employee to use the grievance procedure. Employers therefore have the incentive to ask their legal staffs to prepare comprehensive and lengthy grievance procedures on sexual harassment. This would impede the company in disciplining supervisors for sexual harassment, thereby reducing the company's efficiency, in order to reduce the threat of lawsuits.79 Defensive legalism is also a problem because of the incentives of in-house legal staffs. Their motive is to protect the company from litigation, since they will be blamed for not providing enough warning, but they will ignore the adverse efficiency effects. They will advise extraordinary caution, and try to document their advice so they can prove later that the legal staff are not to blame for the lawsuits that occur.80 The problem of defensive legalism is even more apparent in the lower courts, which are perhaps more used to dealing with the details of implementation of new social policies. Consider the following excerpts from two of the most divergent 7th Circuit opinions in Ellerth. Judge Posner writes: And since everyone knows by now that sexual harassment is a common problem in the American workplace, the employer ought in addition to take, in advance of specific cases of harassment, preventive measures against it, as by adopting and announcing a policy against sexual harassment and creating a discreet and convenient machinery by means of which victims can obtain relief without exposing themselves to retaliation. These are the responsibilities that a negligence standard imposes.81 Judge Flaum says: Pursuant to a negligence standard for supervisory harassment, we would presumably require a posted anti-sexual harassment policy, a grievance procedure that allows a complainant to circumvent the supervisory chain of command, and prompt remedial action. . . . While posted policies and grievance procedures are important, I believe that the remedial goals of Title VII demand more. Companies' efforts to deal with sexual harassment should be systemic and proactive, rather than discrete and reactive. We know that companies can implement grievance procedures and discipline wayward employees; but we also know that companies can hire, train, and promote employees with an eye toward preventing undesirable behavior. In the abstract, a negligence standard conceivably could account for a company's systemic efforts to promote a workplace free of sexual harassment. Employers who had not done enough to reduce the likelihood of harassment throughout the workplace would be found negligent, even if they had no notice that a specific employee was a harasser.82 A wide spectrum of judges, it seems, would look coldly on a company that did not have a formal grievance procedure for sexual harassment although the behavior is widely known to be illegal and even more widely considered immoral. Companies must have special policies for sexual harassment that they do not have for other kinds of complaints about supervisor or co-worker misbehavior.83 Thus, the end result of the Supreme Court's decision may be both more sexual harassment and less corporate efficiency, the only winners being personnel officers and lawyers. V. CONCLUSIONS I have tried in this article to give a view of sexual harassment law based on objectives and incentives. This view looks at results one might expect from different laws rather than at their stated intents. The current law cannot be justified as promoting equality between the sexes, protecting employees against employers, or protecting employers against supervisors, except to the extent that it allows employees redress against breach of contract by employers who provide worse working conditions than expected or employers and employees redress against supervisors who take advantage of their position to engage in unauthorized harassment. Even then, it is open to doubt whether the incentive benefits of civil suits are worth the legal costs. A third objective is the elimination of mental externalities under which members of the general public suffer disutility when employees are harassed or otherwise required to endure mistreatment with a sexual element. This objective is quite reasonable from an efficiency viewpoint, and has real-world plausibility, since many members of the public do indeed seem highly offended by sexual harassment, even, in fact, if the victimized employee were to be amply compensated monetarily. Criminal penalties might well be a better instrument to use to achieve this objective, however, and legislatures should consider replacing our present system of private enforcement. The courts' attention has been most occupied with issues of allocating liability to employers when supervisors engage in unauthorized harassment. Under the common law of agency employers have not been punished for intentional torts by their agents which went against the employers' interests. This is consistent with the policy of punishing the least-cost avoider of harm. The least-cost avoider is whoever perpetrates the harassment-- the supervisor, if there is one, or the employer, if it is a smaller organization. If the perpetrator is unavailable, then we face the harder question of how to allocate the loss between the employer and the employee. In the absence of information acquisition costs, we would simply put the loss on the least-cost-avoider, whichever party that might be. In practice, a good rule would be to put the loss on an employee who does not notify the employer that harassment is occurring, or on the employer if the employee shows that harassment is occurring. In between is a murkier region in which the employee complains, but without clear evidence of supervisor harassment. Letting losses lie where they fall in such cases would at least avoid the high costs of litigation and defensive administration of businesses. The Supreme Court's rule in Ellerth, however, leaves the question of liability unclear, and by shifting the burden of proof on the defendant to show lack of negligence, without defining negligence, it encourages litigation and inefficient defensive behavior with little likelihood of significantly reducing the amount of actual harassment that occurs. 1 118 S Ct. 2257 (1998). 2 118 S Ct. 998 (1998). 3 118 S Ct 2275 (1998). 4 A number of areas of law pertain to sexual harassment. The common law applies via contract and tort law. Since sexual harassment is a workplace offense, it can often be viewed as breach of contract. Since it is an injury to a particular individual, it can be viewed as a tort, under the heading of intentional infliction of emotional distress. State fair employment statutes often apply. See Terry Dworkin, Laura Ginger & Jane Mallor, Theories of Recovery for Sexual Harassment: Going Beyond Title VII, 25 San Diego L Rev 125 (1988). For a practical exposition of the law, see William Petrocelli and Barbara Repa, Sexual Harassment on the Job (Nolo Press, 1998). A standard treatise is Barbara Lindemann and Paul Grossman, Employment Discrimination Law (American Bar Association,1996). 5 477 US 57 (1986). A parallel law of sexual harassment is based on Title IX of the Civil Rights Act, which provides that "no person in the United States shall, on the basis of sex, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any education program or activity receiving Federal financial assistance." 20 U.S.C. § 1681. The Supreme Court ruled in Franklin v. Gwinnett County Pub. Schs., 503 US 60 (1992) that a school can be liable for damages for sexual harassment under Title IX. 6 One dictionary defines "harass" as: "1: to worry and impede by repeated raids (harassed the enemy) 2a: EXHAUST, FATIGUE b: to annoy persistently." Webster's Ninth New Collegiate Dictionary, First Digital Edition (Merriam-Webster, 1992). 7 Indiana Code § 35-45-4-3 (1998), "Patronizing a prostitute" says "A person who knowingly or intentionally pays, or offers or agrees to pay, money or other property to another person: (1) For having engaged in, or on the understanding that the other person will engage in, sexual intercourse or deviate sexual conduct with the person or with any other person; or (2) For having fondled, or on the understanding that the other person will fondle, the genitals of the person or any other person; commits patronizing a prostitute, a Class A misdemeanor. However, the offense is a Class D felony if the person has two (2) prior convictions under this section." The language seems to apply to quid pro quo sexual harassment, but I have not heard of any arrests being made. 8 42 U.S.C. § 2000e-2(a)(1). 9 42 U.S.C. § 2000e-5. 10 Civil Rights Act of 1991, 42 U.S.C. § 1981 (a). 11 42 U.S.C. § 1988 (b). 12 Oncale resigned after taking a smoking break without permission and becoming involved in an argument with the harsassing supervisor, who reprimanded him and said he would "run him off'. For graphic details of the case, see the Brief for the Respondents, Oncale v. Sundowner, 1996 U.S. Briefs 568. Note that Oncale's allegation is not of mere horseplay, but of genuine threats of homosexual assault; he said, "I felt that if I didn't leave my job, that I would be raped or forced to have sex." ) Oncale, 118 S Ct. 998 (1998) at __, quoting Oncale's deposition). 13 Ellerth v. Burlington Industries, Inc., 123 F3d 490, 510 (7th Cir 1997) (en banc). 14 118 S Ct. at ___. 15 "For this accusation of trespass it is, that gives the court of king's bench jurisdiction in other civil causes, as was formerly observed; since, when once the defendant is taken into custody of the marshall, or prison-keeper of this court, for the supposed trespass, he, being then a prisoner of this court, may here be prosecuted for any other species of injury. Yet, in order to found this jurisdiction, it is not necessary that the defendant be actually the marshall's prisoner; for, as soon as he appears, or puts in bail to the process, he is deemed by so doing to be in such custody of the marshall, as will give the court a jurisdiction to proceed." William Blackstone, Commentaries. Book 3: Of Private Wrongs, Chapter 19: Of Process. 16 Justice Thomas's one-sentence concurrence in Oncale emphasizes that this is the legal principle applied even there: " I concur because the Court stresses that in every sexual harassment case, the plaintiff must plead and ultimately prove Title VII's statutory requirement that there be discrimination "because of sex."' Oncale, 118 S Ct. 998 (1998) at __. 17 One court, at least, has suggested this is the case. See Barnes v. Costle, 561 F2d 983, 990 (DC Cir 1977). 18 Sexual reproduction does not require more than one sex. It means that two individuals each contribute half of the genes to the offspring, as opposed to asexual reproduction, in which the one parent contributes all of the genes. Both amebas and flatworms (planaria) have just one sex, but flatworms reproduce sexually. See Ralph Buchsbaum, Animals Without Backbones (2nd ed., 1948) at 19, 115. Differences in attractiveness between individuals are intrinsic to sexual reproduction, because the value of the genes to be contributed to the joint offspring differ in value. 19 For a careful analysis of the allocational effects of sexual liasons at work, see Gertrude Fremling and Richard Posner, Status Signaling and the Law, with Particular Application to Sexual Harassment, unpublished manuscript,July 11, 1998 (Status Signalling). 20 Nevertheless, a court has found discrimination in such circumstances. Toscano v. Nimmo, 570 F Supp 1197 (1983) (holding in favor of women who alleged that although she was not asked for sexual favors the woman who was promoted in competition with her succeeded because she did grant them to the supervisor). Whether because of quid pro quo's or for other reasons, attractive people, both men and women, do better in the workplace. See Daniel Hamermesh and Jeff Biddle, Beauty and the Labor Market, 84 Am. Ec. Rev. 1174 (1994). 21 The employer can, of course, threaten not to hire the person. This, however, leaves the person no worse off than if the employer did not exist in the first place. Using the word "threat" there is like using it to say that the employer threatens not to pay the person unless they agree to work for him; the threat is just to withhold benefits, not to inflict harm. We do not regard a woman as being hurt if someone comes up to her and offers to pay her a thousand dollars if--but only if-- she will wash his car, threatening not to pay her if she refuses. The situation here is analogous. 22 I have overstated the case here. The supervisor does have some power to secretly punish, because he has certain private knowledge about the employee's ability, and he can lie about that knowledge. He can, for example, say in an annual personnel report that his private opinion, based on what he has observed, is that the employee is less talented than would appear to outsiders. Sticks of this kind, however, are limited to private information, whereas the carrots of undeserved privilege and promotion can more safely be used when information is public. 23 It is sometimes observed that a university department has far more than its share of unpleasant faculty. This may be a rational response to unfortunate initial conditions-- if the department's social atmosphere is poisonous to begin with, that department loses far less by hiring an obnoxious new professor than another department which starts with cordial and civilized relations among its faculty. 24 Moreover, employer and employee cannot get around the problem by a written waiver of the employee's right to sue. Such an ex ante waiver of a Title VII right is void. See Fremling and Posner, Status Signaling, note 22 at 25. 25 See, for example, Christopher Jencks, The Homeless (Harv. U. Press, 1994). One small example illustrates the point. The City of Bloomington requires that any bedroom shared by three or more people must be at least 200 square feet. A graduate student with a wife and two children was therefore evicted from his one-bedroom apartment, his appeal to stay being unanimously rejected by the Board of Housing Quality Appeals. One local property manager said this case was "cut and dry" and "To me, it's something I deal with every day, and I'm very cognizant of certain ordinances regarding family size and regarding square footage." Family's Request to Extend Lease is Denied, Bloomington Herald-Times, May 16, 1998, p. A1. Www.hoosiertimes.com. 26 The employer is also vulnerable to disappointment in the abilities and effort of the employee. Under employment at will, employer and employee both agree to accept the risk of incurring reliance expenditures and then being disappointed, since the contract can be terminated by either party at any time. 27 This argument may also be found in the short discussion of sexual harassment in Richard Posner, An Economic Analysis of Sex Discrimination Laws, 56 U Chi L Rev 1311, 1332, where, however, Judge Posner expresses skepticism that the reduction in supervisor wages would really outweigh the increase in employee wages and the reduction in productivity. 28 See Restatement of Agency (Second), @ 399, quoted below. 29 This is merely an application of the standard economic idea of an externality, e.g. "When the actions of one agent affect the environment of another agent other than by affecting prices, we will say that there is an externality," Hal Varian, Microeconomic Analysis, 1st Edition (1978) at 203 (italics as in the original). For further elaboration, see Eric Rasmusen, Of Sex and Drugs and Rock'n Roll: Law and Economics and Social Regulation, 21 Harv. J. L. & Public Policy : 71-81 (1997); Eric Rasmusen, The Economics of Desecration: Flag Burning and Related Activities, 27 J. Leg. Stud. 245-270 (1998). 30 Insider trading, at least, is often discussed in terms of morality, but in terms of absolute morality rather than the preferences of the people in society. David Bayne, for instance, has written extensively on the duty of the insider not to hurt those with whom he trades, but the basis for the argument is not that an insider who hurts people is offensive; even if everyone else in society wickedly enjoyed the insider's treatment of others, the insider's duty would remain unaltered. David Bayne, Insider Trading: The Essence of the Insider's Duty, 41 Kansas L. Rev. 315 (1992). Similarly, Anthony Kronman notes that his approach to the duty of lawyers to conduct themselves responsibly is not based on the cost-benefit calculations of economists, even though the conclusions of the two approaches are often identical. Anthony Kronman, The Fault in Legal Ethics, 100 Dickinson L. Rev. 489 (1996). Still another example of policy recommendations based on morality, but on morality independent of the utilitarian calculus of externalities is the recent book by Geisler and Turek on moral regulation of activities such as homosexuality and abortion. Norman Geisler and Frank Turek, Legislating Morality, (Bethany House, 1998). Traditional morality is not based on what people want, but on what people ought to do. Often, however, arguments from morality will reach policy results similar to those from mental externalities, since if a large number of people strongly hold certain moral beliefs, the mental externalities generated will justify social rules satisfying their preferences. 31 Oncale, 118 S Ct at ___ (1998). 32 In Indiana, for example, stalking is defined as "a knowing or an intentional course of conduct involving repeated or continuing harassment of another person that would cause a reasonable person to feel terrorized, frightened, intimidated, or threatened and that actually causes the victim to feel terrorized, frightened, intimidated, or threatened" and is a Class D felony. Indiana Code § 35-45-10-1, 5 (1998). 33 An example of a vaguely worded statute is the crime of "Intimidation": "A person who communicates a threat to another person, with the intent that: (1) The other person engage in conduct against his will; or (2) The other person be placed in fear of retaliation for a prior lawful act; commits intimidation, a Class A misdemeanor." Indiana Code § 35-45-2-1 (1998). 34 Civil Rights Act of 1991, 42 U.S.C. § 1981 (a). 35 We will return to this below in the discussion of penalties for sexual harassment. 36 The neglect of discrimination in favor of attention to sex is the theme of Vicki Schultz, Reconceptualizing Sexual Harassment, 107 Yale L J 1683 (1998). 37 See Guido Calabresi & Douglas Melamed, Property Rules, Liability Rules and Inalienability: One View of the Cathedral, 85 Harv. L Rev 1089 (1972). Criminal punishment, social norms, and self-help are not in their taxonomy, perhaps because they offer no compensation to the victim of illegal behavior except revenge. 38 Judgement-proofness is a well-known explanation for why some offenses are crimes instead of torts, and why crimes are punished with imprisonment or corporal punishment rather than fines. See Richard Posner, Economic Analysis of Law, p. 222 (4th Ed., Little, Brown, 1992): "This means that criminal law is primarily designed for the nonaffluent; the affluent are kept in line by tort law." 39 Uniform Commercial Code 2-610, Anticipatory Repudiation and U.C.C. 2-711, Buyer's Remedies in General; Buyer's Security Interest in Rejected Goods. 40 On the advantages of stigma generally, see Eric Rasmusen, Stigma and Self-Fulfilling Expectations of Criminality, 39 J L Econ 519 (1996). 41 A prominent example of these costs is the embarassment and perjury into which President Clinton fell as a result of legitimate discovery in Jones v. Clinton, 990 F. Supp. 657 (1998 E D Ark). (This case was brought under neither Title VII nor Title IX, but rather under 42 U.S.C. § 1983, under which a government agent is liable if he deprives a woman of equal protection under color of the law.) . The problem is well put by James Stephen when he discusses why social norms are superior to law in dealing with vice: "... the expense of the investigations necessary for the legal punishment of such conduct would be enormous. It would be necessary to go into an infinite number of delicate and subtle inquiries which would tear off all privacy from the lives of a large number of persons. These considerations are, I think, conclusive reasons against treating vice in general as a crime." James Stephen, Liberty, Equality, Fraternity, 2nd edition (1874) at 151 42 Note, too, that if the supervisor damages the employer by making him liable to the employee, the supervisor can seek compensation from the supervisor if the origin of the liability was in the supervisor's breach of duty. Section 401 of the Restatement (Second) of Agency say that "An agent is subject to liability for loss caused to the principal by any breach of duty," with the comment at @401 (d) that "Thus, a servant who, while acting within the scope of employment, negligently injures a third person, although personally liable to such person, is also subject to liability to the principal if the principal is thereby required to pay damages." (We will see below,however, that sexual harassment will ordinarily not be in the "scope of employment".) 43 The reputation punishment to the supervisor may be larger relative to the supervisor's wealth even though it is smaller in absolute size. The reputation punishment to a large corporation from a well-publicized employee complaint may be extremely large in dollar terms precisely because the corporation is so large to begin with. Whether it is the absolute or relative size of the punishment that is most important has been much studied. See John Lott, Should the Wealthy be Able to `Buy Justice', 95 J. Pol. Econ. 1307-1316 (1987); David .Friedman, Reflections on Optimal Punishment, or: Should the Rich Pay Higher Fines? 3 Research in Law and Economics, ed. Richard Zerbe, 185-205 (1981); David Friedman, Should the Characteristics of Victims and Criminals Count? Payne v Tennessee and Two Views of Efficient Punishment, 34 Boston College L. Rev. .731-769 (1993).. For present purposes, we are concerned more with who should pay and how rather than with ho much. 44 There do exist solutions to the problem of the plaintiff's incentive to sue-- to grant the private plaintiff treble damages or punitive damages, for example-- but devising optimal rules that fit all situations is difficult. For one careful analysis of this kind of problem, see A. Mitchell Polinsky & Yeon Koo Che, Decoupling Liability: Optimal Incentives for Care and Litigation, 22 Rand J. Econ. 562-570 (1991). 45 This sharing of loss among innocent parties is the major theme in the law of agency as applied to breach of contract. There, often even a supervisor's wealth is insufficient to pay for the losses he creates by his irresponsible agreements with third parties. See Eric Rasmusen, The Economics of Agency Law and Contract Formation, unpublished manuscript, Indiana University, Dept. of Business Economics and Public Policy, August 1998. 46 See generally Steven Shavell, Strict Liability versus Negligence, 9 J Leg Stud 1 (1980); Robert Cooter, Unity in Tort, Contract, and Property: The Model of Precaution, 73 Cal L Rev. 1 (1985); Richard Landes and Richard Posner, The Economic Structure of Tort Law (Harv U Press,1987). 47 If we fail to impose liability for a driver who intentionally hits the pedestrian, or fail to make the pedestrian's intentional flinging himself in front of the car a defense for the driver, we will encourage harmful behavior that could easily be prevented. On strict liability generally, see Alan Schwartz, The Case Against Strict Liability, 60 Fordham L Rev 819 (1992). 48 An obscure but interesting example to the contrary from Mongolian law is the provision in the Mongolo-Oirat Regulations of 1640 that if a third party buys stray cattle from a finder, the loss is split-- the original owner is entitled to the head and the third party to the rump. Similar splits were common in other areas of accident law in Mongolia. Saul Levmore, Variety and Uniformity in the Treatment of the Good-Faith Purchaser, 16 J Leg Stud 3, 63 (1987). 49 This is the conventional economic interpretation of what is behind the famous Hand Rule of United States v. Carroll Towing Co., 159 F2d 169, 173 (2d Cir 1947). 50 Here, I part company with Judge Posner, who says, "The provisions of the Second Restatement of Agency are designed mainly for two types of case neither of which is before us. The first is the tort committed against a stranger, as where a truck driver employed by the defendant runs down a pedestrian. The second is the contract between a stranger and an agent of the defendant." Ellerth-7th Circuit, 123 F3d at 508 (1997). Judge Posner thinks torts by fellow servants and supervisors are significantly different; I think the same principles useful for analyzing them. 51 These are, of course, not the only two alternatives. The court could use comparative negligence and base the split on the efforts of each party to avoid the harm. This, however, has high transactions costs, since every effort and every loss must be precisely estimated, and both parties will contest the estimate every step of the way. 52 Lord Holt said in Jones v. Hart, K.B. 642 (1698) "If the servants of A. with his cart run against another cart, wherein is a pipe of wine, overturn the cart and spoil the wine, an action lieth against A." A's servant is acting on A's business, even though he is performing it incompetently, and so A is liable. For an economic analysis, see Alan Sykes, The Economics of Vicarious Liability, 93 Yale L J 1231 (1984). For an economic view of the scope of employment rule, see Alan Sykes, The Boundaries of Vicarious Liability: An Economic Analysis of the Scope of Employment Rule and Related Legal Doctrines, 101 Harv. L Rev 563 (1988). 53 Illustration 4 is attached to the following comment. "The fact that an act is done in an outrageous or abnormal manner has value in indicating that the servant is not actuated by an intent to perform the employer's business. See Comment b on @ 229 and @ 245. In such cases, the facts may indicate that the servant is merely using the opportunity afforded by the circumstances to do the harm. Hence, unless the principal has violated a personal duty to the person injured, or unless he becomes liable because of the nature of the instrumentality entrusted to the servant (see @@ 212-214), he is not liable for such acts." 54 Michael J. Phillips, Employer Sexual Harassment Liability Under Agency Principles: A Second Look at Meritor Savings Bank, FSB v. Vinson, 44 Vand. L. Rev 1230 (1991). See also J. Hoult Verkerke, Notice Liability in Employment Discrimination Law, 81 Va L Rev 273 (1995). 55Hunter v. Allis-Chalmers Corp., Engine Div., 797 F2d 1417, 1422 (7th Cir 1986, Posner, J.). Although this is true of harassment, there are many intentional torts which agents do plausibly think are in the interests of the principal, and so are within the scope of employment. In Son v. Hartford Ice Cream Co. , 102 Conn. 696, 129 A. 778 (1925), defendant's truck driver, sent to a shop to collect money, used force and injured the plaintiff. The trial and appellate courts ruled for the plaintiff, since the agent was on his principal's business, and was using his force on his behalf, however misguidedly and even though he was otherwise instructed. Similarly, Judge Learned Hand held that a drunken boatswain who assaulted a sailor for lying idle in his bunk was acting within the scope of his employment. Nelson v. American-West African Line, 86 F2d 730 (2nd Cir 1936). 56Vinson v. Taylor, 753 F2d 141, 151 (DC Cir 1985). 57 How little this point is understood is shown by what even Judge Flaum, who advocates an expansive view of sexual harassment law, says in his 7th Circuit opinion in Ellerth, referring to the Restatement of Agency: "I imagine that subsections (a) and (c) would infrequently be of analytical value in a sexual harassment case, for presumably employers are not in the business of harassment, and it is not conceivable that harassment of an employee will serve the interests of an employer." It would help the employer by pleasing the harasser, who would be happier at his work. 58 See also § 219(2) of the Restatement (Second) of Agency: (2) A master is not subject to liability for the torts of his servants acting outside the scope of their employment, unless: (a) the master intended the conduct or the consequences, or (b) the master was negligent or reckless, or (c) the conduct violated a non-delegable duty of the master, or (d) the servant purported to act or to speak on behalf of the principal and there was reliance upon apparent authority, or he was aided in accomplishing the tort by the existence of the agency relation." 59 Restatement (Second) of Torts § 317 (1958). 60Lancaster v. Norfolk & Western Railway, 773 F2d 807 (7th Cir 1985), at 818. 61 In McCrink v. City of New York, 296 N.Y. 99, 71 N.E.2d 419 (1947), the City was held liable for the shooting of two people by a drunken on-duty policeman whom it knew had a drinking problem. In Easley v. Apollo Detective Agency, 69 Ill. App. 3d 920 (1979), a detective agency was held liable for attempted sexual assault of an employee after it hired a security guard, and gave him a pass key to tenants' apartments, knowing he had once been fired for sleeping on the job, had an arrest record for minor crimes, and that a female co-worker once complained he had "made eyes at her." (This was under a special Illinois statute imposing liability on detective agencies for employee's misconduct-- Ill Rev Stat., ch. 38, para. 201-10b(10) (1975) (repealed 1984)). There are also, of course, an abundance of rules requiring employers to hire only qualified workers for jobs with high potential for non-intentional torts, e.g., driving trucks or airplanes. 62 See, e.g., Mark C. Lear, Just Perfect for Pedophiles? Charitable Organizations That Work with Children and Their Duty To Screen Volunteers, 76 Tex. L Rev 143 (1997). 63 Slander law and natural disinclination to write derogatory reference letters means, of course, that references are likely to be more boring than lurid. As Judge Coffey points (Ellerth-7th Circuit, 123 F3d at 541 (1997)), state statutes often bar employers from reasonable precautions such as checking arrest records in hiring employees; Wisconsin law forbids employers from asking job applicants whether they have arrest records. Wis. Stat. Ann. §§ 111.331, 111.32, and 111.335 or from basing an employment decision on criminal convictions unless they "substantially relate to the circumstances of the particular job." Wis. Stat. Ann. § 111.335. Would a rape conviction be relevant to a job in which the applicant would work with women? The unfortunate employer won't know till his case reaches trial. See, more generally, Walter Olson, The Excuse Factory: How Employment Law is Paralyzing the American Economy (Free Press, 1997). 64 The Restatement of Agency has extensive discussions of apparent authority and ratification. See Sections 32 to 49 on apparent authority and 82 to 104 on ratification generally. These doctrines are most important in contracts made by agents on behalf of the principal, since the cost to the third parties of discovering the absence of actual authority can be too high for it to be efficient to impose that burden on them. With regard to most torts, the question of authority is unimportant, because the victim does not agree to the tort. With regard to sexual harassment, however, authority matters because (a) the tort may be a perk of the agent, reducing the amount the principal needs to pay him as discussed earlier, and (b) if the tort is authorized or seems to be authorized, it is inefficient to require the victim to learn that complaint to the principal would result in punishment of the agent. 65 Making the owner liable for the intentional torts of his agent makes him vulnerable to collusive lawsuits. The supervisor could deliberately harasses the employee in exchange for a cash payment and freedom from suit, after which the employee would sue the employer, using the supervisor as a witness who freely admits to the harassment. More simply, a supervisor could takes revenge on an employer who had somehow offended him by committing torts for which the employer would be liable. The only drawback is that the supervisor would be liable to the employer for compensation for the damages paid to the third party, but in cases where the supervisor was judgement proof his ability to commit intentional torts and shift the cost to the employer could be quite effective. 66 Jennifer Arlen and Reinier Kraakman, Controlling Corporate Misconduct: An Analysis of Corporate Liability Regimes, 72 NYU L Rev 687 (1997). 67 The 1997 article is an expanded analysis of this point, which was earlier made in Jennifer Arlen, The Potentially Perverse Effects of Corporate Criminal Liability, 23 J. Leg Stud 833 (1994). See also David Dana, The Perverse Incentives of Environmental Audit Immunity, 81 Iowa L Rev 969 (1996). 68 Faragher, 118 S Ct at ____(1998). This other woman was Nancy Ewanchew, the other plaintiff in Faragher-D.C. 69 Faragher118 S Ct at ____(1998). 70 Faragher, 118 S Ct at 61,62____(1998). 71 Ellerth-U.S., 118 S Ct. at 44,55___ (1998). 72 Stacey Dansker reports in 1997 that the circuits were split on the issue of strict liability for quid pro quo sexual harassment. Stacey Dansker, Note: Eliminating Strict Employer Liability in Quid Pro Quo Sexual Harassment Cases, 76 Texas L Rev 435, 436. The Fifth Circuit affirmed, in an unpublished table decision, a district court ruling that an employer who took prompt action to correct quid pro quo harassment was not liable, and the Supreme Court refused certiorari. Sims v. Brown & Root Industrial Services, 889 F Supp. 920 (WD La 1995), 78 F3d 581 (5th Cir 1996) , dert denied, 117 S Ct 68 (1996). For the facts and procedure in this case, see Jennifer Johnson, Note, Employment Law-Are Employers Strictly Liable for Supervisor Sexual Harassment in the Fifth Circuit After Sims v. Brown & Root Industrial Services, Inc. et al., 889 F Supp. 920 (W.D. La. . 1995)? Probably Not Unless Harassment is Included in Supervisory Job Descriptions, 38 Texas L Rev 965 (1997). The Fifth Circuit opinion was unpublished, though, and Justice Kennedy writes in Ellerth that "Every Federal Court of Appeals to have considered the question has found vicarious liability when a discriminatory act results in a tangible employment action." Ellerth-U.S., 118 S Ct. at ___ (1998). He then cites Sauers v. Salt Lake County, 1 F3d 1122, 1127 (10th Cir 1993), which imposes strict liability. Justice Scalia wrote that the Supreme Court had declined to grant certiorari on the issue in Ellerth, but quid pro quo liability turns up as dictum nonetheless--as part of the Ellerth Rule cited above. . Eugene Scalia, The Strange Career of Quid Pro Quo Sexual Harassment, 21 Harv. J L & Pub Pol 307, 321 (1998). For a defense of strict liability in the context of sexual harassment by supervisors, see David Benjamin Oppenheimer, Exacerbating the Exasperating: Title VII Liability of Employers for Sexual Harassment Committed by Their Supervisors, 81 Cornell L Rev 66 (1995). 73 Strict liability for quid pro quo harassment rule has been accepted with relatively little debate perhaps because of the analogy to racial discrimination. Racial discrimination motivated by the supervisor's bias against an entire race can be systemic, resulting in a visible pattern of employment actions, a pattern more easily visible to the employer even than to outsiders or employees. Employee complaint is much less important. In recent years, suits based on individual mistreatment have become more common; see John J. Donohue & Peter Siegelman, The Changing Nature of Employment Discrimination Litigation, 43 Stan. L Rev 983 (1991). This is one example of how employer liability for the employment actions of supervisors arises in more contexts than just sexual harassment. For a broad view, see Rebecca White, Vicarious and Personal Liability for Employment Discrimination, 30 Ga L Rev 509 (1996). 74 xxx I added the racial discrimination footnote in response to the reviewer's comments. I've kept it in, since if the reviewer thought it was important, perhaps other readers will too, even tho I agree with you that discussing race diesrimination is distracting. I don't know that your footnote discussions of qpq harassment and racial discrimination are helpful because you don't tie them into your analysis. Should the type of discrimination make a difference regarding the nature of the liability? 75 Farmer v. Brennan, 511 U.S. 825, 837 (1994). 76 Jane Doe et al. v. University of Illinois, 138 F2d 653, Lexis no. 59 (7th Cir 1998). Judge Posner wrote this in dissent to a denial of en banc hearing of this case. He also distinguished this Title IX education case from Title VII cases such as Ellerth: "The analogy to Title VII is deceptive, since Title VII regulates the behavior of adults in the workplace rather than the inevitably unruly behavior of adolescents." (at Lexis no. 57). (Implicit is that a high school is more like a federal prison than like a workplace.) 77 Psychologists refer to our tendency to rely unduly on memorable rather than typical occurences as "the availability bias". See pp. 92-102 of Robin Dawes, Rational Choice in an Uncertain World, (Harcourt Brace Jovanovich,1988) for an overview and references. 78 Ellerth-U.S., 118 S Ct. at 42___ (1998). 79 University procedures regarding misgrading and cheating may provide a model. I have taught at UCLA and at Indiana University. Both have highly formalized and time-consuming procedures for dealing with student complaints of professorial misconduct in grading and with professor complaints about student cheating. Professors complain of these procedures, but we do not seem to realize that although they can consume considerable faculty time, they make complaint so difficult that the misgrading professor can hardly be punished. . Indeed, I was told by one professor at a state university that the most effective way to deal with cheating was to avoid the formal procedure for punishing the student and simply to give him an F on the final, which would put the procedural burden on the student. From the university's point of view, the objective may be to derail troublesome complaints about both cheating and misgrading. 80 This, in turn, may create evidence that imperils the employer in subsequent litigation This is indicated by a September 15, 1998 memo on Indiana University's new sexual harassment. Its Appendix A includes a series of examples such as "In the hospital hallway, a male doctor and a male student health care professional once again discuss the physical attractiveness of a female patient. A female student health care professional feels embarassed and excluded by the conversation. When she expresses her feelings of discomfort, she is told there is not harm intended and that she needs to "grow up" if she wants to be successful on this clinical rotation." Our legal staff can now say they have warned the employees, but their expansive definition of sexual harassment will make it difficult for them to defend the university when, inevitably, violations occur. 81 Posner, Ellerth-7th Circuit, 123 F3d at 511 (1997). 82 Flaum, Ellerth-7th Circuit, 123 F3d at 498 (1997). 83 Special procedures might avoid retaliation by the supervisor. However, the employee does not need much imagination to see that complaining to the supervisor's boss, as required under formal procedures, might help. This may be an argument for using some outside regulatory or police agency to detect and punish sexual harassment, since outsiders have few incentives to betray confidences. 36