A. Freedom of association as a species of freedom of contract

The principle of freedom of association has a long and storied history in both legal and political discourse. This principle holds that each person has the right to associate, or not to associate, with any other person or persons, for good reason, bad reason, or no reason at all. The major advantage of freedom of association is that the rule facilitates the creation of complex social institutions that are generated by the repeated application of the same generative principle.Footnote ¹ Accordingly, once two or more individuals decide to associate with each other, they may exercise in combination the same right to do or not do business with any other person or group of persons. Small groups can thus form larger ones, just as larger ones can recognize special divisions within themselves. This traditional model still holds for all different types of contract: if two or more people contract to form a relationship (commercial or otherwise), an outsider may not deliberately interfere with their relationship. This simple contract model facilitates a variety of socially and economically productive transactions. By definition, any voluntary transaction excludes—or discriminates against—all nonparties to that transaction. Under the standard legal theory, the secret to social welfare is to increase the velocity of transactions by reducing the transaction costs that stand in the way of mutual gain through voluntary association.Footnote ²

B. The freedom of association in libertarian and classical liberal theories

The principle of freedom of association has been under heavy attack since the rise of the New Deal; critics view the principle as creating major opportunities for the illicit exclusion of some individuals from some voluntary organizations. For the moment, I want to put aside this frontal assault on the principle of freedom of association to concentrate on two conflicting accounts of this principle, namely, the libertarian and classical liberal conceptions of freedom of association.Footnote ³ The difference here is important because the qualified classical liberal acceptance of freedom of association is able to meet criticisms that look more plausible when leveled against the more uncompromising libertarian version.

Strong libertarian theory, such as that developed in Robert Nozick's Anarchy, State, and Utopia, does not allow for any use of forced exchanges via taxation or eminent domain. As such, it places few limitations on the application of the freedom of association principle. The most important limitation is that associations—now called conspiracies—that intend to use force or fraud against third persons can be punished in their own right, even if they have not achieved their goals. But that principle rests on the same narrow conception of a harmful externality that lies behind the basic libertarian theory, which places a prohibition on force and fraud because of the damage they do to other individuals.

Classical liberal theory also gives strong support to the principle of freedom of association, but hedges it in with two further exceptions, relating to the law of antitrust and common carriers—both of which pertain to the use of monopoly power. That focus should not be surprising, since any theory of limited government starts with the objective of controlling the monopoly power that the state enjoys over the use of force within its territory.Footnote ⁴ The usual response to the monopoly problem is not to debar the state from acting at all, nor to prevent it from singling out individuals for punishment, fines, and prohibitions because they have committed a wrong involving force or fraud. Every coherent theory of government power accepts at least a version of the police power that permits some state control over the use of property and the exercise of liberty in order to protect “safety, health, morals, and general welfare of the public.”Footnote ⁵ But with respect to its general commands, the classical liberal position seeks to work its way between two extremes—the first a system that requires unanimous consent, and the second a system that relies solely on majority will. The first system creates the risk of political paralysis, and the second the risk of factional domination of minority interests.Footnote ⁶ One—I dare say the best—solution to this is to allow the majority to call the shots so long as it is bound, under a nondiscrimination principle, to the same obligations as the minority; this is a position that often expresses itself in the ideal of equal justice before the law.Footnote ⁷

I will not attempt here to work out the implications of classical liberalism for the design of government on such vital issues as federalism and separation of powers. But I think the classical liberal view helps us understand why private monopolies will generate at least some concern, even if they cannot use force against other individuals. In this connection, the most sensible application of the antitrust laws forbids horizontal associations among competitors that are intended to restrict output and raise prices. Here the simplest remedy is to require each firm to act as a stand-alone competitor. The situation becomes cloudy, however, with firms in markets that are incapable of division, so that no competitive market solution is available. The key historical illustration of these “natural monopolies” was the common carrier, a large and important category, generally neglected in philosophical writings, which could include stagecoaches, inns, railroads, telecommunication networks, and internet operating systems. Although modern technology has reduced the number of settings in which firms can exert monopoly power,Footnote ⁸ in earlier eras the dominant position of inns, coaches, railroads, and telephone companies brought forth a legal regime that made refusals to deal prima facie unlawful, and imposed a duty on the common carrier to deal with all comers on reasonable and nondiscriminatory terms.

Yet beware of the crucial limitations. That duty never applied in simple two-party transactions. The common carrier rule only limited the refusal to deal in select cases. It never limited the ordinary rights of property owners to exclude others from their land, clubs, or vehicles. To warrant either forced association or forced entry—and the two are intimately linked—two conditions have to be satisfied simultaneously. First, the firm must enjoy some monopoly control, which derives either from a state franchise or some physical configuration that creates a natural monopoly, such as a harbor with room for only one dock company.Footnote ⁹ Second, the firm's activities must involve the dispensation of routine and fungible services that in no way turn on the distinctive characteristics of the party who receives them. Phone companies do not have professional or intimate relations with their individual customers. Any customer that follows the rules and pays the stipulated rates is entitled to service. The electric, water, and gas companies fall under the same basic regime. And in earlier days the same applied to innkeepers and railroads, neither of which holds monopoly power today, save perhaps in some niche markets.

The traditional common carrier rule on required service sought to steer a middle path between two extremes.Footnote ¹⁰ The first extreme was arbitrary price discrimination. The common carrier rule's prohibition against discrimination guarded against allowing firms to charge differential rates to two customers who received the same class of service. There is a spirited disagreement as to whether this principle is preferable to price discrimination, which has two advantages over uniform rates. In some cases at least, price discrimination allows back into the marketplace individuals who could afford to pay less than a uniform monopoly price for services, but only if high demanders are charged higher prices to pick up the shortfall in revenue. The overall mix of prices could be revenue-neutral to the firm. Or price discrimination could allow the firm to recover additional revenues and thereby encourage more rapid entry of new businesses into the market. This form of price discrimination, if left unregulated, allows firms to use their pricing power to achieve collateral “advantages,” including discrimination on such questionable grounds as political affiliation, personal connections, and race. The strategy can work because the customer has no other provider to turn to for the same service. Thus, the usual compromise is to permit certain distinctions over classes of service (e.g., business and residential phone services), but not among individual customers within these categories.

The second extreme was arbitrary wealth redistribution. The classical regimes of public regulation were not redistribution schemes designed to make one class of customers richer at the expense of another. To prevent that redistribution from happening, the nondiscrimination principle, understood in economic terms, requires rate differentials to reflect differences in the cost of providing services. That result always holds in competitive markets. No life insurance company, for example, can charge the same premium to two individuals in different risk categories. Faced with that strategy, the low-cost customer would go elsewhere: the insurer stands to lose money by selling insurance priced at some blended rate, but only to the high-risk customer. To prevent this adverse selection, the insurer charges different premiums to customers in different risk categories. That option is not available to regulated firms without any exit right because a firm requires the approval of some state commission before it is allowed to withdraw, either in whole or in part, from business. Accordingly, a regulator may easily require the firm to serve one class of customers at below cost, while charging higher rates to customers in other classes. Customers in the second group subsidize customers in the first group, as happens today routinely with the heavy business subsidies of residential phone service. The overall system therefore seeks to minimize the economic and political distortions that arise whenever individuals or firms are forced into win/lose arrangements. Sound forms of common carrier regulation work, over time, to avoid both monopoly profits and ruinous cross-subsidies. Associational interests (apart from lobbying organizations) among customers are nonexistent for most kinds of business—for example, power and communication—and have only limited resonance in such areas as transportation, where some voluntary sorting might take place. Even in the Old South, moreover, most firms resisted segregated facilities that government mandates forced upon them.Footnote ¹¹

But the common carrier rule never applied to intimate associations, clubs, businesses, or religious organizations. In these contexts, classical liberal theory collapses into libertarian theory, because both categorically reject any system of forced association that impairs the autonomy of group members and the efficiency of group operations when no monopoly problems are involved. By definition, some persons want to get into groups that wish to keep other individuals out. Transaction costs are typically low. If the group's enlargement were on net beneficial, then the membership would accept the outsiders wholly without regard to external coercion. But expanded membership hardly means a greater level of utility across the board. The gains to the outsiders are likely be smaller than the losses to the insiders, or otherwise the expansion of the group would have occurred by voluntary means. Wholly apart from issues of scale—that is, the case where the cost rises for additional individuals who are admitted as membersFootnote ¹²—it is easy to see why limiting membership makes sense. The greater the diversity of views, the harder it is for any governance structure to overcome the bitter antagonisms that arise within the ranks. A religious organization cannot long survive if its members do not agree on mission, doctrine, or liturgy. That is why Protestants broke off from Catholics, Southern from Northern Baptists, and so forth.

The classical liberal position recognizes that these internal tensions impose real costs on the association and its members. If those tensions are perceived before organizations are formed, then any rule of universal inclusion could block the emergence of the organization in the first place. Rather than take inclusion as a strong presumptive norm, the classical liberal approach treats membership exclusivity as its prime objective. The aim is to make sure that all groups, regardless of their beliefs or membership requirements, can sprout up at will. In the long run, that approach will produce more vibrant organizations with a greater range of objectives and styles, all of which can survive because of their ability to cater to their members. The implicit assumption is that the expanded range of options works better than any legal regime that uses nondiscrimination rules to pry open membership. The desire to be inclusive at the back end creates a large, if invisible, barrier to entry at the front end. In sum, under classical liberalism, freedom of association becomes an inseparable and universalizable element of freedom, limited only by the concern with monopoly power. There are no efforts to limit the associational principle to certain subclasses of organizations, whether delineated by purpose, size, membership requirements, or anything else.

A. Expressive organizations: The Boy Scouts and the NAACP

The classical liberal resolution of the tension between regulation and freedom of association works as a convenient foil to the modern approach that limits freedom of association for reasons that have nothing to do with coercion, fraud, or monopoly. It is easy to identify two different attitudes toward this issue, one which is strongly sympathetic to the right of association, and another which is deeply suspicious of it. Oddly enough, these two different attitudes manifest themselves in connection with the role of discrimination in state power. To set the antidiscrimination principle in context for an extended discussion of gay rights, it is useful to see how the freedom of association principle evolved under the First Amendment before it came in conflict with the modern antidiscrimination laws.

In one sense, the common theme is the extent to which a principle of freedom of association deals with membership in voluntary organizations. The best introduction into this study is NAACP v. Alabama (1958),Footnote ¹³ which upheld on freedom of association grounds a First Amendment challenge to Alabama's efforts to gain the membership list of the local NAACP chapter during the height of the civil rights movement. Another important case is the Supreme Court's decision in Boy Scouts of America v. Dale (2000),Footnote ¹⁴ which upheld the right of the Boy Scouts to exclude gay members; this case raises an explicit challenge to the classical liberal principle of freedom of association because its facts are so stark and simple.

In the NAACP case, the Supreme Court had to decide whether the attorney general of Alabama could require the disclosure of the membership lists of the NAACP. The case itself was brought under the First Amendment, on the ground that the Alabama statute unconstitutionally burdened the freedom of association of the NAACP's members, by requiring the unwanted disclosure. Justice John M. Harlan recognized the close connection between the substantive right to freedom of speech and the related principle of freedom of association in a passage that requires close attention:

Justice Harlan's laudable argument rests on a conscious equivocation of terms. His use of “assembly” in the first sentence leaves the impression that the association right has an explicit textual foundation in the First Amendment. Instructively, the word “assembly” does not appear in the First Amendment, which references only “the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.” Even if the second clause is not a limitation on the first, the words “to assemble” in context read much more naturally as the ability to meet in public to discuss various issues. “To assemble” does not sound like the right to form associations that meet in private to plan and organize with respect to a full range of “political, economic, religious or cultural” issues. The transformation of the constitutional guarantee of a right “to assemble” is complete in Justice Harlan's second sentence, which substitutes the broader term “association” for the narrow term “assembly.”

It is, however, worth stressing that Justice Harlan's conclusion would be correct even if the First Amendment stopped after the protection of “freedom of speech, and of the press.” The basic classical liberal synthesis recognizes that individuals cannot work alone to achieve their ends, but must cooperate in order to obtain the benefit of gains from trade. It would be odd to read the First Amendment to say that any isolated individual could speak against government action, but two people could not pool their resources toward the same end. And once they do, the terms of partnership are theirs to decide; if the membership decides to specialize its functions to increase its output, those secondary decisions are protected as well. Nor should it make the slightest difference if some individuals contribute only cash to the joint venture while others contribute their time and imagination. To look at the speech of one person in isolation from that of others is to put on blinders. The right to engage in any form of productive labor directed toward any expressive end carries with it an associative right. The right of association toward the constitutionally protected ends follows by necessary implication from the basic constitutional text.

Harlan's reference to the due process clause of the Fourteenth Amendment is a bit of American constitutional law legerdemain, well established at the time, to the effect that the due process clause “incorporates” the First Amendment. This means that the clause applies the First Amendment against the state governments rather than just against the federal government.Footnote ¹⁶ And Harlan's final sentence notes, consistent with classical liberal theory, that the rights of association for speech are not bounded by some specific subject matter but cover “political, economic, religious or cultural matters,” uniformly. There is, accordingly, no need to decide which of these four ends the NAACP advanced, since all were equally protected. Harlan's last reference to “the closest scrutiny” meant that the Court would look on its own initiative to see whether the purported evil that the legislation addressed was in fact an evil. If so, the Court would then ask the question of whether the means chosen fit tightly with the ends, to guard against the potential overbreadth of the statute. The upshot was that the NAACP could claim the associational rights of its members and keep its membership lists from the prying eyes of Alabama state officials, at a time when segregation was still in its heyday. And it was critical, as Harlan noted, that the NAACP be able to assert this claim:

In the end, therefore, the autonomous structure of the organization was protected against government intrusion on principles that are wholly consistent with the classical liberal view toward limited government. There was simply no justification for the desired state intrusion into the affairs of a private organization—all the more so because the organization was dedicated to undoing the political regime that sought to regulate its behavior. The NAACP was not engaged in plans for the violent overthrow of government, so it does not come close to falling under the force or fraud rationales for limiting associative freedom.

B. Boy Scouts of America v. Dale

The issue of freedom of expressive association raised itself in quite a different environment in Boy Scouts of America v. Dale, which puts the issue of gay rights front and center. James Dale entered the Boy Scouts at the age of eight in 1978 and rose to the level of Eagle Scout some ten years later. By all accounts, he excelled in his work. In 1989, he became an adult scout who was appointed an assistant scoutmaster of Troop 73, a position of trust within the organization. Shortly thereafter, he acknowledged publicly that he was gay. In response, the Boy Scouts revoked his membership in the organization, explaining in response to his direct inquiry that the Boy Scouts “specifically forbid membership to homosexuals.”

In one sense, the case looks different from NAACP because it does not involve government efforts to suppress political opposition. But the motivation for the government action is not the decisive question. What matters is the right of organizers to accept or refuse members as they see fit. The basic principle of freedom of association should allow the Boy Scouts to determine their own membership, unless it can be established that the Scouts had acted in breach of contract, or engaged in the threat or use of force or fraud, which they had not. For organizations of this type, the question of monopoly power is quite beside the point.

Nonetheless, New Jersey's Law Against Discrimination (LAD), which proceeds from very different assumptions, reached a very different conclusion. Similar to the Civil Rights Act of 1964,Footnote ¹⁸ but going far beyond it, New Jersey's Law Against Discrimination provides:

This statute is broader than the 1964 Civil Rights Act by virtue of a 1991 amendment, which added the phrase “affectional or sexual orientation” to the list of characteristics on the basis of which it is forbidden to discriminate.

There is only one applicable exception contained within the New Jersey statute, which reads:

Everyone agreed that the Boy Scouts could not be regarded as “distinctly private,” with its membership of five million plus boys; and it was, quite by design, not affiliated with any religious institution. The classical liberal account I have outlined does not distinguish between religious and secular associations. It follows, therefore, that this statute limits freedom of association in ways that are unacceptable. Since the statute's language is clear, the only question of its legality was whether the prohibition it contained would fail a constitutional challenge under the First Amendment guarantees of freedom of speech and association. Within the framework of antidiscrimination law more generally, that had to be regarded as an uphill battle when the case was first brought, because of an unbroken line of Supreme Court and lower court cases that had rejected any and all challenges to the Civil Rights Act.

Just that fate befell the challenge to the Law Against Discrimination in the New Jersey Supreme Court. Chief Justice Deborah Poritz, writing for the unanimous New Jersey Supreme Court, held that the LAD survived the constitutional challenge. In reaching this decision, she noted that the state had offered a compelling justification for its antidiscrimination law by pointing out in its legislation the social evils that flowed from discrimination, including “economic loss; time loss; physical and emotional stress; and in some cases severe emotional trauma, illness, homelessness or other irreparable harm resulting from the strain of employment controversies; relocation, search and moving difficulties; anxiety caused by lack of information, uncertainty, and resultant planning difficulty; career, education, family and social disruption; and adjustment problems.”Footnote ²¹ She then added a further commentary:

There is much to disagree with in this defense of the LAD. At a social level, it is hard to credit various forms of discrimination with the potent consequences that the legislature attaches to them. To be sure, no one disagrees that there are such consequences when private force is used to keep certain individuals out of gainful occupations or trades, a pattern and practice that characterized so much of the segregated South. But unless it could be shown that all individuals in concert engage in horrific behavior with respect to particular groups, the larger claim represents the worst kind of intellectual cherry-picking: Look at the bad outcomes in individual cases, but systematically ignore the gains from trade that are generated for all persons by competitive markets in all areas of life.

Nor is Poritz's grim account consistent with the enormous social transformation of attitudes toward gay and lesbian individuals over the past forty years. No longer do psychiatrists describe homosexuality as a clinical disease or disorder, and the expansion of gay and lesbian organizations—all rightly protected by the principle of freedom of association—has been remarkable and effective. It is simply a mistake to ignore the huge social transformation in attitudes and behaviors, which the principle of freedom of association has facilitated, so as to create a kind of supposed market failure to justify state intervention. The one real grievance of gay and lesbian individuals lies, ironically, in exactly the opposite direction, namely, in the continued state prohibition against same-sex marriages, a prohibition which regrettably has gained ground in recent years.

Yet the reason this prohibition really does harm is exactly the reason why the antidiscrimination law does little or no good. Dale, and others similarly situated, may be excluded from some organizations. But they are, by the same token, able to join or create thousands of others that welcome them into membership. Not so with the legal prohibition on same-sex marriages. The state has a monopoly position through its control over the licensing process, which is not limited by an antidiscrimination norm that would require treating gay couples by the same standards as straight couples. The upshot is that so long as democratic institutions bar gay marriages, the same-sex marriage option is effectively foreclosed because there is nowhere else to turn, even if a substantial minority of the state electorate disagrees with the legislative choices. We should never forget that this political minority is effectively silenced by legislation that enacts the dominant social opinion into law. Prohibitions on same-sex marriage belong, therefore, in the cross-hairs of any classical liberal theory, which must question why the state is in the business of issuing marriage licenses in the first place. The antidiscrimination norm in employment and public accommodations does not remove this barrier to voluntary arrangements.

Poritz's opinion falls prey to all these misunderstandings because it unthinkingly rejects outright the classical liberal distinction between state and private action, or, in the context of private actions, the distinction between monopoly and competitive situations. Instructively, Poritz uses the word “excluded” in the passive voice, and never bothers to mention which people were excluded by what groups, and what alternatives remained open to them. Her opinion just brands exclusion as an invidious social practice, with some consequences she does not like, without asking how the principle of freedom of association fares in the long run by giving other opportunities to people who have lost out in particular transactions. Nor does her opinion explain why full participation in social, economic, and political life is somehow not achieved via the civil capacity to vote, to enter into contracts, to join organizations, to give testimony in courts, and so on—rights that are fully consistent with the like liberties of all other individuals.Footnote ²³ Her account simply speeds us along the path toward expanded state power.

When Boy Scouts of America v. Dale reached the U.S. Supreme Court,Footnote ²⁴ the justices switched course by upholding the freedom of association claim on a narrow five-to-four vote for those expressive organizations protected by the First Amendment. The Dale majority, speaking through Chief Justice William Rehnquist, was not in the business of remaking American constitutional law so as to vindicate a neo-Lochner freedom of contract norm for all voluntary associations.Footnote ²⁵ Thus, it faced a difficulty that does not trouble the defenders of the antidiscrimination law, which is how and why to draw the line between expressive and nonexpressive organizations. Perforce its discussion becomes murkier—just what is an expressive organization, and why should it matter? Of course, the Boy Scouts are concerned with inculcating moral values in the young. Does that make the organization expressive? Does it matter that its own code of conduct required young people to be “morally straight,” but did not explicitly condemn homosexual conduct? How is that to be set off against the Scouts' consistent refusal to accept gay scout leaders?

To Justice John Paul Stevens, writing in dissent, this studied equivocation provided the wedge for sustaining the antidiscrimination law. He wrote as if large polyglot organizations were incapable of expressing unified positions. Yet the Boy Scouts' equivocation on moral questions forced them, in his view, to forfeit their associational freedoms: “In light of [the Scouts'] self-proclaimed ecumenism, furthermore, it is even more difficult to discern any shared goals or common moral stance on homosexuality.”Footnote ²⁶ Justice David Souter's brief opinion echoed the same theme: “no group can claim a right of expressive association without identifying a clear position to be advocated over time in an unequivocal way.”Footnote ²⁷ Unfortunately, the dissenting views leave it unclear whether organizations like the Boy Scouts have any associational freedoms that would allow them to resist, for example, the disclosure of membership lists to the state.

The patchwork response of the two dissents fares poorly against the classical liberal alternative. That approach protects the freedom of association for all reasons, whether expressive or nonexpressive, and thus need not police a line that is not worth drawing in the first place. But even if the line between expressive and nonexpressive associations has to be drawn, as is the case in Dale, on which side of the line does any group like the Scouts that has multiple purposes fall? Certainly the insistence on a clear position makes little sense. Sprawling organizations like the Boy Scouts take evasive positions for a good reason. At over five million strong, the Scouts cannot please everyone.Footnote ²⁸ Speak too loudly against gays, and they lose a large portion of their socially liberal membership base. Speak too softly, and a significant fraction of conservative parents might be unwilling to trust their sons to the Scouts. This fragile coalition can stay together only if internal dissent does not boil over into actual schism. With shades of Aristotle,Footnote ²⁹ this equivocation is a form of moderation that allows individuals with divergent viewpoints to remain under the same tent, as part of a cohesive organization, instead of tenants in an office building who share utilities and common spaces. Associational freedom should never be reserved to those who take extreme positions. Just because the First Amendment protects the Ku Klux KlanFootnote ³⁰ does not mean, as Chief Justice Rehnquist wrote, that it offers no protection to more popular groups.Footnote ³¹ Indeed, he hit the nail on the head when he wrote: “The fact that the organization does not trumpet its views from the housetops, or that it tolerates dissent within its ranks, does not mean that its views receive no First Amendment protection.”Footnote ³² The unpopular Rehnquist decision does a far better job in balancing the equities than the more fashionable dissents of Justice Stevens and Justice Souter.

C. The Americans with Disabilities Act

The Dale case poses an interesting challenge to the modern liberal synthesis because it relies on the higher scrutiny of First Amendment law to return the law of free speech to the classical liberal synthesis. Many forms of regulation are not directed to matters of expression, but to other activities; and, regarding these ordinary activities, the level of constitutional scrutiny is so low that virtually no constitutional oversight is even attempted under current law. Thus, while the classical liberal principle of freedom of association goes beyond expressive cases, the protection of the First Amendment, no matter how broadly construed, does not. To see how this shift in orientation works, it is instructive to examine the vast legal apparatus against discrimination in public accommodations on various grounds. The public accommodations provisions on race under Title II of the 1964 Civil Rights Act were somewhat narrower than the analogous provisions of New Jersey's Law Against Discrimination. Insofar as they applied to race, they were implemented with barely a peep of protest, save from rabid segregationists who could never have imposed their will on established businesses solely by market pressures.Footnote ³³ Segregation by race is a clunky way to do business in that it generates immense amounts of ill-will, so that most businesses affirmatively welcomed public accommodations regulation because it “forced” them to desegregate without having to face the wrath of local hoodlums who knew that they could burn down the premises without fearing any response by state police. These firms faced no expensive cost subsidies that desegregation would introduce.

The antidiscrimination norm in public accommodation generated a very different response under the 1990 Americans with Disabilities Act (ADA), which prohibits discrimination on grounds of disability both in employment and in access to a wide range of facilities. The statute starts with a definition of the general duty not to discriminate against disabled individuals either in employment or in the operation of various facilities.Footnote ³⁴ The obligation to take into account the interests of handicapped persons is not absolute, but only requires (in a term of art) that “reasonable accommodations” be made, the account of which is governed by complex statutory provisions.Footnote ³⁵ Thereafter, there is an offset with respect to those accommodations that create “undue hardship”—yet another term of art.Footnote ³⁶ It is worth noting that no charitable interpretation can give either of these phrases a clear meaning. Both invoke a cluster of factors, expressed in vague terms, which necessarily speak about matters of degree, all of which are relevant and none of which are conclusive. No firm can get permits for new construction unless its operations are ADA-compliant under the most detailed and exhaustive of legal regimes.

The public accommodations requirements for disabled people present a wholly different landscape precisely because the cost of service to severely disabled persons can and does differ by orders of magnitude from the costs of equivalent service to other individuals. It is just those cost differentials that justify rate differentials under the classical liberal theory. To be sure, no firm is required to remove all cross-subsidies, and many would not choose to do so, within limits. Basic business sense and decency require ramps for wheelchairs in airports and hospitals, even if ramps for wheelchairs are also often used for baby strollers and luggage carts. The high density of use and the multiple classes of users overcome any hesitation about funding these expenditures from general revenues. Some steps toward improved handicap access would be routinely provided even in the absence of any disability statute. The measures taken would surely be less, however, than the champions of the disabled would like.

The battle in practice centers on the extent of the subsidy, and resistance to government-mandated subsidies is now on the rise, albeit covertly. Every community now contains a disability rights group that demands that every public or private bus or train be equipped with wheelchair hoists, or that every public toilet or remote rocky resort be made wheelchair accessible.Footnote ³⁷ On this score, we would do well to remember the account that Philip K. Howard (founder and head of The Common Good) has given of the human-rights-driven fiasco of trying to introduce public toilets into New York City. The only workable toilets were too small to accommodate wheelchairs. The larger toilets that could accommodate wheelchairs blocked public roads and also had to remain locked so that they could not be used for prostitution and other illicit sexual activities. A new institution that provided benefits for most but not for all was dutifully blocked, even though more public toilets reduced unwanted odors in public places—a classic public good.Footnote ³⁸

Unfortunately, disability groups typically object to all measures of interest to them that make some people better off if they do not make their members better off as well. Across-the-board improvements that lead to greater inequality of fortune are somehow out of bounds. Recall that traditional rules on common carriers only required equal charges for persons for whom the cost of service was, roughly speaking, identical. The new version of public accommodations ignores that equal-cost constraint and thus spawns huge subsidies that deter sensible innovation in the absence of any monopoly power. Here is one easy test of why the subsidies induced by the ADA make little sense: Suppose that disabled individuals could receive in cash, pro rata, the extra money needed to make the new public toilets accessible to them and safe for everyone else—say, by posting a twenty-four-hour guard at each facility. Then ask how these individuals would vote in an election in which they would all be bound by the consensus outcome, so that all free-riding problems would be eliminated. Would most individuals vote to spend the money on those collective improvements, or would they choose to spend it in other ways? The betting here is that they would find better uses of the money, which is a good reason not to make this particular expenditure. Better to do without the accessible toilets and, if need be, make a more sensible public expenditure from which disabled individuals could derive either exclusive or disproportionate benefits.

At this point, the dangers of myopic antidiscrimination legislation exceed the dangers of exclusion by private parties. The insistence on the modern mandate that requires private firms to provide reasonable accommodations for handicapped individuals that do not impose “undue hardship,” for which no clear standard can be articulated, should be abandoned as unworkable.Footnote ³⁹ In a world in which no requirements were imposed, some accommodations would still be made, and most of these would be reasonable even if they involved some net cost to the firm. But in a world in which public enforcement agencies and laws define what counts as “reasonable accommodations” and “undue hardship,” the natural equilibrium shifts. All too many concessions are made when benefits are printed in bold, large type and burdens in regular, small type. The University of Chicago Law School has its own monument to the folly of the current disability regime in the form of an expensive and ugly five-stop elevator that is virtually never used to move wheelchairs from one level of the school's auditorium space to another. It is not an isolated example. The costs for handicap compliance are large, with the common estimates from architectural and construction firms usually running between 20 and 25 percent of project costs, often to satisfy the needs of at most one or two users.Footnote ⁴⁰ We are not speaking of small effects from the aggressive enforcement of the antidiscrimination laws.

A. Their basic structure

The current antidiscrimination norms exert a powerful hold over all forms of business and social arrangements, subject only to a modest, and vulnerable, exception for expressive organizations. It is critical to understand, therefore, how the antidiscrimination requirement should be applied to private associations and public accommodations. The risk that these laws can go off the rails, as it were, should caution us against their application to routine businesses, churches, schools, universities, charities, and voluntary associations. Unfortunately, the ambition of the civil rights movement is not constrained by these niceties. Rather, it is premised on the resolute insistence that private and public discrimination are morally objectionable in a broad range of social institutions.

One effort to make the modern antidiscrimination laws more palatable is to call them “human rights laws,” in the hopes that this new label can give them a degree of universality that helps legitimate them in the eyes of the world at large. That is the verbal strategy of the New York Human Rights Act (1945), whose substantive provisions track those of New Jersey's LAD, in the broad definition of public accommodations and the narrow exclusions from the basic antidiscrimination command for distinctively private organizations.Footnote ⁴¹

My point here is not to quarrel with the details of coverage. Within their chosen sphere, the drafters of the human rights statutes have not misconstrued their own mission. No analyst may know the exact intensity that people attach to choosing their social groupings, but we do have a strong intuitive sense of their implicit orderings, which helps explain the exceptions found in the human rights acts. It would be odd to allow a facility manager to keep healthy people from working in his restaurant. But it would be equally odd to say that he is bound not to discriminate against infected food-handlers who could spread diseases to innocent customers. Likewise, it would be an odd human rights statute that allowed people to take sex into account in providing financial advice, but not in counseling for victims of rape or sexual harassment. The law to one side, we all share some general sense of the social differences between general business friendships and intimate personal relations. From a public choice perspective,Footnote ⁴² it is quite inconceivable that any winning coalition could reverse the widespread agreement over these implicit orderings by requiring that individuals in the marriage market not discriminate on the grounds of race, age, sex, or sexual orientation.

B. Why these human rights?

Why does any government get into the “human rights” business at all? To be sure, the phrase “human rights” is a masterstroke of political rhetoric: Who can oppose a statute that protects “human rights,” especially if it contains no mention of human duties? Each of us is human, and thus the rights we seek to protect are only those belonging to ourselves. No one would argue that the human rights laws are fundamentally misconceived by denying, perversely, all possibility of human rights as such. Yet the choice of these human rights raises deeply troublesome questions about these statutes. These questions go to the universality of the purported rights, the specification of their correlative duties, their impact on affirmative action programs, and their effect on the production of useful information.

Universality. Although human rights acts establish many rights, nobody claims that they contain the entire corpus of individual rights conferred by any legal system, ancient or modern. People possess many other rights, such as rights of personal autonomy and conscience, rights of religion, rights of free speech, rights to contract, and rights over their own persons and property. Some of these rights relate to economic liberties; others are critical to political participation; some matter for individual self-realization; some play into all three. All of these rights generally are highly important. Yet it is never quite explained why these rights fail to qualify for inclusion in human rights statutes. If we do not regard these traditional liberties as human rights, then how should we classify them? Are they second-class rights? Rights that only exist at the whim of the legislature, which may revise or redefine them at will?

The lack of universality in modern human rights laws is also evident by the incompleteness in the list of covered organizations. The classical liberal position draws no distinctions among lawful organizations with respect to their size, their membership, or their ends. All lawful organizations are protected against forced association, on the ground that those who are excluded from one organization continue to have a wide range of other organizations to choose from. But the modern position is much less sure of its grounds. As noted, the New York and New Jersey antidiscrimination statutes say that distinctly private organizations are not covered: clearly, the law is not intended to prevent people from discriminating on the grounds of sex, race, or religion in choosing marriage partners. All this is sensible enough. But the narrow exceptions do not make the case for the basic prohibition. As we move from groups of two in intimate settings to larger groups in various relationships, just where should the principle of freedom of association give way to the nondiscrimination principle? The allowance for expressive organizations, discussed in Section III above, raises the simple question of whether businesses with distinct credos should be regarded as expressive or whether they should fall into some other category reserved for ordinary businesses. It is hard to draw any line or to justify any line once drawn.

To see the gap between contemporary human rights statutes and earlier visions of human rights, compare the modern statutes with one key civil rights statute. I refer here not to the Civil Rights Act of 1964, but to the earlier and wiser human rights statute, the 1866 Civil Rights Act,Footnote ⁴³ passed by a Reconstruction Congress in the aftermath of the Civil War. There is no overlap in the civil rights protected by the 1866 act and the modern human rights acts. The 1866 statute gave to all citizens rights such as the right to contract, to hold property, to convey real estate, to testify in court, and to sue or be sued.Footnote ⁴⁴ It essentially guaranteed civil capacity—the right to participate in a social order organized under the law of property, contract, and tort. The 1866 Civil Rights Act primarily dealt with economic rights and liberties. It is also possible to clearly delineate the correlative duties to these enumerated rights. If A and B have the right to contract, then C is duty-bound not to block their joint endeavors with the use of force or fraud; if A is competent to give testimony in court, then no one can prevent his appearance; and so on down the line.

None of these rights is mentioned, let alone affirmed, in modern civil rights legislation. Certainly no one holds that the Bill of Rights has depreciated over time, even though most of its liberty-based claims are in tension with modern human rights laws. To put this another way, if the rights specified in human rights acts are truly fundamental, why historically have they taken so long to surge to the forefront of anyone's legislative agenda? In the United States, their influence dates only from the 1960s. The same pattern holds around the world.

I think that their delayed introduction stems from the simple fact that no one regards any right to be free of discrimination in ordinary social or business transactions as truly fundamental for his or her own well-being and security. Consider this question: On an all-or-nothing basis, how would you choose between having only the protection of a human rights act, with its various prohibitions on discrimination, or the common law prohibition against the use of force? I know exactly which rights I would disregard. I would allow people to refuse to hire me for any reason at all, so long as the state protected my bodily integrity. I doubt that in any real-world setting a single person would choose differently. Nobody would prefer the risk of being killed by a “nondiscriminatory” murderer to a flat prohibition against the use or threat of force. The result does not change if we compare the modern statutory guarantees against discrimination to the classical liberal right to make contracts to dispose of one's labor or property. Once again, the basic right to trade occupies the more fundamental niche, no matter what the structure of the marketplace. To deny people the right to trade forces them to rely exclusively on their own resources and negates the enormous gains from the division of labor. Any antidiscrimination norm is therefore heavily parasitic on the basic right to trade, which it seeks to limit in unproductive ways, and largely for misguided reasons. The antidiscrimination norm cannot survive as a stand-alone principle.

The lack of universality in modern human rights laws is further revealed by the limited class of persons to whom they apply. The older civil rights statutes applied to all citizens regardless of the role that they occupied in some particular relationship. The 1866 statute, for example, sought to end discrimination on the grounds of race by guaranteeing to “citizens, of every race and color” the same right “to make and enforce contracts, to sue, be parties, and give evidence, to inherit, purchase, lease, sell, hold, and convey real and personal property, to full and equal benefit of all laws and proceedings for the security of person and property, as is enjoyed by white citizens.”Footnote ⁴⁵ In one sense, this provision is not universal because it does not extend its protection to aliens, which is in line with the uneasy tradition that, within each nation, citizenship gives advantages in the economic realm that can be denied to foreigners. That is certainly the case with respect to the United States Constitution, where the privileges or immunities clause of the Fourteenth Amendment, adopted two years after the 1866 Civil Rights Act, provides that “No State shall make or enforce any law which shall abridge the privileges or immunities of citizens of the United States.”Footnote ⁴⁶ The 1866 act is, in one sense, a nondiscrimination provision because it refers to the various rights that are protected under the classical liberal framework, but only protects them to the extent that they are enjoyed by white persons. The purpose of that limitation is to make sure that various limitations on these rights can be imposed so long as it is done in an even-handed fashion with respect to race. Yet, at the same time, the one result that the Fourteenth Amendment does not bring about is to override the key notion of freedom of association by denying to all private persons the right to choose whom they wish to contract with in the first place.

The dominance of the antidiscrimination norm makes it easy for courts to convert the classical liberal conceptions that animated the 1866 Civil Rights Act into a robust version of the antidiscrimination principle. The Warren Court adopted this interpretation of Section 1982,Footnote ⁴⁷ the modern statutory successor to the 1866 act, in Jones v. Alfred H. Mayer Co. (1968).Footnote ⁴⁸ The Jones decision held that Section 1982 barred all forms of private discrimination in the sale or renting of real estate on the grounds of race, by virtue of its plain meaning. Yet the decision never acknowledged the sensible reading of Section 1982, namely, that if white persons could sell, or refuse to sell, to white or black persons, then black persons could sell, or refuse to sell, to white or black persons. Far from imposing a nondiscrimination obligation on persons of all races, Section 1982 expressly negates that obligation.Footnote ⁴⁹ Yet once the error in Jones is put aside, nothing in the 1866 Civil Rights Act or any of its successors has the built-in role-specific characteristic of modern human rights statutes. The modern human rights statutes all protect, for example, tenants against landlords, or employees against employers. The 1866 act and its successor did not favor one side of the relationship. Rather, they allowed any person to decide whether to become a landlord or a tenant, an employer or an employee, with the same level of contractual protection always enjoyed by whites.

Correlative duties. The modern human rights laws are also deficient by being inattentive to the duties that are correlative to the rights created. Rights are claims over scarce resources—human, physical, or intangible. Any claim only makes sense if the assignment of a right to A necessarily imposes duties on at least some B. Footnote ⁵⁰ A mature system of rights thus requires, at a minimum, internal coherence. It is not sufficient to identify the holder and the object of the right; it is also necessary to specify who is bound and to what correlative duties. Accordingly, classical jurisprudence always paired human rights with correlative human duties. To the extent that one person's sphere of action is expanded, the liberty of action of all other individuals is necessarily contracted. In designing a system of rights, the advantages that we conceive of and create for certain individuals should more than offset the disadvantages thereby imposed upon others. Today's mantra of human rights refuses, rhetorically, to acknowledge the existence, and certainly the weight, of these correlative duties. No one wants to defend the “Reconfigured Human Rights and Duties Act,” much less the “Human Duties Act.” Yet why the lack of candor?

The answer is simple. It is much easier to forge winning coalitions by accentuating the positive. The strong constraints of scarcity are ignored, so it looks as though the new legal reform does create free lunches for all. To be sure, the classical liberal position also has explicit correlative duties, namely, those universal duties of noninterference in the state of nature, and those specific duties of compliance with voluntary obligations. But these rules have powerful efficiency properties. The primary duties among strangers are relatively easy to state, and, most critically, are scalable, so that the content of these duties is the same whether we work in a society of thousands or millions. The gains from noninterference dwarf the losses. And for particular obligations, the ability to pick trading partners helps assure joint gains. Disappointed competitors may lose, as they always do, but the increased wealth and opportunities for all enhance the overall size of the pie, creating positive externalities, which should not be ignored in the social calculus even if they are (unlike nuisances) rarely the subject of legal action.

Affirmative action. The classical liberal position and the modern human rights acts also take radically different approaches to the recurrent question of affirmative action programs, which are intended to give some individuals preferences for jobs or positions in universities, say, on the grounds of race or sex. The classical liberal view, of course, does not care why people choose to associate with others. So it does not demand any special public justification by private organizations for preferring women or minority members. Just do it, is the answer. But the flipside is that other groups, be they many or few, may favor men or members of some privileged white elite if they so choose. In practice, voluntary organizations tend to follow the patterns preferred by the defenders of affirmative action, but that hardly provides a reason for exempting only these preferred patterns from the antidiscrimination principle under some ad hoc exception that stresses the historical differences in the treatment of, say, minority groups and women. All that historical information is public, and, if they choose, private organizations can take it into account in fashioning their rules of admission. If women's clubs are permissible, then why not men's? The historical record does not go to general unique entitlements. It only influences the likelihood that men's or women's clubs will form. More specifically, there is no reason to condition the ability of women to form private associations on some proof of past discrimination against women or on some supposed social consensus that bias, either overt or subconscious, persists against members of certain groups.

Nor should we like the current system that justifies some forms of discrimination by consulting the historical record. Using past discrimination to justify the exemption of some forms of discrimination from modern human rights laws creates a real moral hazard: it leads both government and private parties to exaggerate their own past injustices against their favored groups in order to sustain preferential treatment in those groups' favor. That effort to skew the political dialog is both mischievous and unnecessary. Let any group form along any lines, and the pattern of preference gives a good sense of the social response without forcing dissenters to hew to the majority will. There is, in fact, no need for any private group to offer some special justification for its own affirmative action programs.

The situation with public institutions is somewhat more complex, for here state agencies tend to be fiduciaries of the public at large, and these agencies neither do nor should have the same freedom of action as private parties.Footnote ⁵¹ That point carries with it special power in connection with the enforcement of the basic rights to bodily integrity and private property. There can be no system of affirmative action in dealing with criminal situations, with one law of murder for whites and another for blacks. But the modern state also deals heavily in the provision of key goods, such as education, which in an ideal world should be left to private actors. And once the government is in the management business, it has to have some greater degree of discretion than it does with the enforcement of the criminal law, or other functions of the minimal state. At this juncture, I think that the distribution of popular sentiment should help influence what the state can do. So long as there is—and there is—extensive support for affirmative action programs at all kinds of private universities, there is a presumption that these programs should be allowed for their public competitors. I would not apply this test to the Jim Crow era, where all public and private decisions were governed by corrupt political institutions. But I think the relative openness of modern political choices makes it defensible to allow affirmative action programs to function in the public setting—as a second-best solution when there is no realistic way to eliminate state institutions of higher education.

Information destruction. Yet another unfortunate feature of modern human rights laws stems from their adverse impact on one of the most essential ingredients of well-functioning markets: a high level of information flow among parties to voluntary transactions. Once a human rights law designates certain characteristics as “off limits,” that law must implement ways to keep the parties subject to it—landlords, proprietors, and especially employers—from using information about those characteristics for improper purposes. In multiple ways, therefore, the task of the legal system is to keep regulated parties in the dark. Two illustrations make this basic point: one deals with testing and the other with medical records.

First, the prohibition against discrimination in employment quickly led, in the important case of Griggs v. Duke Power Co. (1971),Footnote ⁵² to an explicit limitation on systematic standardized testing by employers of their prospective and actual employees. The disparate impact, whereby black applicants systematically perform worse than white applicants, counts as a form of discrimination that can only be justified by business necessity.Footnote ⁵³ No longer can private employers rely on standardized tests, such as the Wonderlic test,Footnote ⁵⁴ which have proved valuable in a wide range of contexts. Instead, the statutory approach kills the messenger by barring the use of any standardized test in which minority employees are underrepresented unless it passes a tight definition of job-relatedness.Footnote ⁵⁵

No one claims that these tests are perfect, or that other forms of employer evaluation don't count. The only question is whether the hiring process can be made more rational with this information than without it. The current rule states that testing cannot be used even if it improves prediction of successful job performance, unless it has a virtually perfect correlation with success, which no test has. The costs of this approach are quite large because it forces the preparation, at great expense, of particularized tests that must focus narrowly on certain defined tasks. Such job-specific tests are of little value in identifying the long-term career paths of job candidates and employees. The consequence is that employers have less information to make decisions, and workers have a reduced incentive to acquire those forms of information that are not tested.

Second, with respect to medical records, the tendency toward disinformation also arises under the Americans with Disabilities Act. Information about potential disabilities generally cannot be taken into account in making decisions on hiring or promotion,Footnote ⁵⁶ but the ADA permits employers to collect the same information for medical purposes.Footnote ⁵⁷ The entire system thus requires deliberate concealment and material nondisclosure that would, if not government sanctioned, be branded as a serious form of fraud (say, for example, under the 1933 Securities Act).Footnote ⁵⁸ The consequence, of course, is the creation of large, concealed cross-subsidies, as healthier workers are forced to bear some fraction of the medical costs of their less fortunate colleagues. It may well be that some public support should be provided to workers who are unable to obtain insurance through their employers. But the current situation is highly unstable.

To be sure, it is easy to make arguments that the gain to a worker from additional medical insurance produces a greater social benefit than the financial loss sustained by the employer. Cass Sunstein offers the hypothetical case of a poor disabled worker who is only willing to pay $20 more for an accommodation that it costs an employer $150 to supply; he reasons that the employer should bear the $150 cost, since the gain in social welfare would be greater than the cost as measured by some impoverished standard of willingness-to-pay.Footnote ⁵⁹ Everyone likes to make interpersonal comparisons of utility in the abstract. But this ostensible appeal to a social-welfare standard is wholly myopic because it ignores the full institutional setting. It does not ask whether, for example, separate facilities for disabled workers should be encouraged because they reduce the employer's average costs so as to make the alteration feasible. Build a facility for ten disabled workers, and its benefits could exceed the cost to the employer, such that in a low-transaction-cost environment, the improvement can take place. But if it cannot, then announcing that we should force the exchange has profound negative consequences. First of all, without a willingness-to-pay criterion, no one knows how real these numbers are. It is easy to inflate preferences when they need not be backed by dollars. Nor do those simple numbers take into account the effect of the change in amenities for other workers or customers—an effect which could, depending on circumstances, be either positive or negative. Nor does any simple effort to get to overall welfare without going through market transactions take into account the profound incentive effects that the systematic coercion has on employers, who will flee from hiring, especially in large numbers, disabled workers who cost them more in total compensation than they bring in return benefits.Footnote ⁶⁰

The mistake here lies in thinking that any long-term social improvement can come by using the coercive power of the state to force win/lose relationships on employers, or for that matter, on anyone else. The program is too reminiscent of the general socialist mind-set that central planning can do a better job with social welfare than individual choice. But it always backfires. Universalize this social-welfare norm and every firm could be bankrupted by constant demands to supply benefits that cost the firm more than it generates in revenue. With bankruptcy looming, employers will evolve to minimize the legal pressure, by reconfiguring their operations in ways that make them less dependent on the labor of disabled persons.

Ironically, one point that we can be confident of is that the introduction of any antidiscrimination law will introduce covert discrimination. In open markets, the ability to adjust wage and other terms allows all sorts of persons to compete by accepting lower wages, if need be. But once those wage adjustments are ruled out, we should expect employers to take, covertly to be sure, avoidance measures to minimize their exposure. We cannot improve welfare by destroying markets.

The message is therefore cautionary. If some government intervention is appropriate to deal with the additional health-care costs of disabled individuals, an explicit government subsidy is a more transparent and efficient way to handle this problem. Unfortunately, the current appetite for hidden cross-subsidies has the predicted effect. It tends to discourage firms from hiring disabled employees in the first place.

Explicit state discrimination. The aggressive enforcement of human rights laws does more than increase the levels of private discrimination. In the public sphere as well, it can create the very abuses that these laws were intended to guard against. As with affirmative action, it is critical to distinguish between the government as an enforcer of rights and the government as a manager of its own businesses. As to state regulation, the key element to success is to reduce the level of discretion in the enforcement of law. The multifactor tests under all the antidiscrimination laws necessarily increase the scope of government power to regulate private firms. State officials have to collect evidence to establish certain violations and determine remedies. They have to decide whom to sue and with whom to settle. They know that bringing any enforcement action can impose real costs on private parties, even those that prevail after painful litigation, with the attendant loss of reputation.

So who are the officials who carry out these sensitive roles? People who tend to be supporters of the human rights statutes. They, in turn, erect questionable presumptions of wrongdoing that bear little relationship to fact, or impose race- or sex-specific remedies in order to correct what they regard as unconscious forms of discrimination. There is little, however, that anyone can do to restrain the vast reservoirs of state discretion, which pose a serious threat to the rule of law.

The situation is grim on the other side as well. State agencies have to make their own hiring decisions. Yet under the current law they are vulnerable to disparate impact cases even when they have introduced affirmative action programs that parallel those in private businesses. The most perverse upending of these programs arose in Connecticut v. Teal (1982),Footnote ⁶¹ still good law, where the state agency both ran an aggressive affirmative action program (on the one hand), and conducted extensive standardized testing, with the usual black/white disparities (on the other). The logic of the Connecticut program was first to set an appropriate ratio between white and black employees, and then to select the strongest candidates from each group. One can easily quarrel with the use of explicit race-based criteria for government employment. But so long as affirmative action programs are a part of public and private life, the key question is how to maximize the efficiency of the workforce subject to that constraint, which the Connecticut program did.

In the Teal decision, however, Justice William Brennan struck down the use of the test in the case of a black applicant who fell below the cut-off line for the black group on the ground that the test did not meet the rigorous job-related standards established in Griggs. But Brennan's decision was wholly context-blind. The attack on testing in Griggs was motivated by the (not-unjustified) fear that some employers might seek to introduce race classifications through the back door under the guise of testing. But that fear was wholly negated by the unquestioned implementation of Connecticut's extensive affirmative action program. The effect of Teal was to knock out the testing that allowed for efficient sorting of job candidates. With the decline of testing, subjective factors come back into the equation, leading to various forms of intrigue in public and private workplaces, even those committed to affirmative action. It is important to recall that the key function of the antidiscrimination principle is to operate as a brake against the monopoly power of the state. Those concerns are not present in labor markets, where the government as employer is in competition with private employers. Without question, these private employers would resort to testing to get better information about their workers if left free to do so by the state. There is no reason why the state should not be allowed to operate in the same fashion as its private competitors.

The illusion of abundance. The failure to recognize the exaggerated claims about the benefits of human rights laws has the unfortunate consequence of ignoring or understating their associated costs. As long as this is the case, we will live, and suffer, under the happy illusion that the constraints of scarcity do not really matter—that we will be able to magnify certain rights without limitation, while nobody need pay the cost either directly or indirectly. That is, of course, a fantasy. How, then, does this fantasy take hold? It does so partly, I believe, through employing the following technique. Instead of talking in terms of human beings on both sides of the rights/duties divide, we depersonalize the entities on whom the duties are imposed. Individual people receive the benefits created by human rights laws, but the burdens of the laws are imposed on abstract entities such as corporations, unions, universities, and other organizations.

This mental conjuring trick rests on illusory foundations. The collectives on which we impose duties are comprised of individuals as shareholders and members. Organizations do not act themselves; they act through individuals and for individuals, be they shareholders, union members, or faculty and students. Consequently, in analyzing the impact of a statute, we must return to the postulate of methodological individualism, so congenial to the classical liberal tradition. We cannot simply match human rights to a set of correlative duties on abstract bodies. We cannot just appeal to notions of community to impose obligations on individuals and the groups they form. We must trace the implications of the statute through the entity to the particular individuals on whom those duties will be imposed.

In politics the same principle applies. It is incorrect to say that the government has a “duty” to supply benefits to its citizens. Any benefits created by the government must necessarily be backed by regulation, taxation, or the imposition of liability on other individuals. Those burdens must be recognized and weighed against the benefits, if an appropriate balance is to be obtained. Unfortunately, this balance will not be obtained if public rhetoric conceals the relevant trade-offs.

Moreover, even when the cost elements of a modern human rights statute are allowed to enter into the equation, they are never placed on the same footing as the rights side. Characteristically, in these statutes the rights are put, so to speak, in bold sixteen-point type at the top of the page while underneath, in a corner, the costs are set out in barely legible eight-point type.Footnote ⁶² If the costs turn out to be too great, they may not even feature in the discussion at all.