Freedom of Association and Antidiscrimination Law: An Imperfect Reconciliation

The topic of this essay[1] is to identify the proper role of antidiscrimination laws in relation to the general principle of freedom of association, which is itself a subset of the basic principle of freedom of contract. Historically, the usual understanding was that the principle of freedom of association enjoyed pride of place in the social hierarchy, such that the antidiscrimination principle was employed only in select contexts, primarily as a counterweight to monopoly power in a broad class of public utility and common carrier settings. The basic position had strong underpinnings in the constitutional provisions that protected equally property rights, economic liberties, and freedom of speech. One unified vision protected all forms of productive human behavior.

Today, the intellectual climate has changed, so that the antidiscrimination norm has far greater appeal in at least several directions.

First, the constitutional landscape has changed, so that property rights and economic liberties receive only the lowest level of protection under the rational basis test. Freedom of association, grounded in the First Amendment, still receives greater levels of protection, thereby creating a huge cleavage in constitutional status in the many cases where speech, and economic liberty overlap. As a consequence of the new constitutional order, there is now no real objection to the application of that norm to restrain the behavior of employers and landlords, but never employees and tenants, in how they conduct their business. The principle has thus become role-specific rather than perfectly general, and thereby opens the door to favoritism, faction and intrigue.

Second, the list of impermissible types of discrimination has expanded since the basic articulation of the principle in one of the watershed events of the 20th century, the passage of the Civil Rights Act of 1964—which Clay Risen accurately described as “the Bill of the Century” in a book by that name.[2]

Today, the basic list of the forbidden forms of discrimination now includes not only the old standbys of race, religion, natural origin and sex (the term used in the 1964 Act), but also age, disability, and sexual orientation. As the categories of impermissible grounds have multiplied, so too has the law ramped up enforcement mechanisms to implement the law’s substantive commands. Right now it authorizes administrative action by various government agencies, most notably the Department of Justice, the Office of Civil Rights in the Department of Education, and the Equal Employment Opportunity Commission. But there are civil rights divisions in virtually every government agency, most notably in the agencies that regulate housing, education, and employment.

Public enforcement in these areas has been marked by an aggressive uptick in the set of judicial and administrative remedies that, as is often true in the administrative state, exert great pressure on various firms and associations to toe the government line. Today there is always the looming threat of a wide range of legal sanctions, including threats of individual enforcement actions or of more general administrative guidances, both of which are difficult to challenge in court if there has not been a “final agency action”, which under the common reading of the Administrative Procedure Act must be completed before the government action or regulation can be challenged in court.

Substantively, the application of the antidiscrimination norms is not tethered by any requirement that the government or a private claimant show there was any conscious intent to discriminate. It is easy for administrative agencies and courts to impose sanctions on practices that look innocent enough in themselves, but which are said to have a disparate impact on any of various groups—often measured only by statistical inference, without any direct evidence. Once these broad forms of liability are recognized, it has become routine today for government agencies to demand that various businesses and educational institutions introduce compliance officers—often former government employees who have worked on discrimination issues—into their ranks, and vest them with large powers of oversight within the organizations.

Government agencies often have the power to impose heavy fines on noncompliant targets, which are also subject to individual suits and class actions for violations of the broad substantive norms in question. The attitude behind this movement is that errors of over-enforcement pale into insignificance as compared to those of under-enforcement, so great is the supposed societal interest in preventing various forms of discrimination. Once in place, this institutional colossus becomes ever more difficult to dislodge.

And it is all a huge social mistake. The older view that ties the antidiscrimination norm to areas of monopoly power remains intellectually the preferred position. To make this case, I shall first outline the classical theory on this score, and then show how the modern efforts to impose a comprehensive antidiscrimination norm across the board generates lower economic output, greater administrative costs, and higher levels of social discord and unrest—all of which have risen under the Obama administration, most recently in the confrontations and demonstrations at Princeton, Yale, the University of Missouri, and other universities.

The outcome should come as no surprise to anyone who is schooled in the classical liberal tradition, which views the exercise of government power to be an error until it is shown to be a good. No government knows enough to impose from the center a free-floating antidiscrimination norm, and efforts to do so invite the kind of partisan politics that are always a threat to the stability of the basic social order.

The defense that I shall make of the traditional doctrine follows lines of inquiry that I have taken up elsewhere. It is an effort to fuse traditional natural law understandings of human flourishing with more explicit utilitarian accounts of overall social welfare by making judgments about collective welfare as a function of the welfare of individual citizens. The position starts with the basic account of freedom of association. It later turns to the special case of monopoly. Once the theoretical issues are developed, I shall illustrate the dangers of inconsistency and overreach that infuse the modern enforcement of antidiscrimination laws.

The Classical Liberal Approach to Freedom of Association

It is commonly insisted that freedom of association is one of the fundamental liberties that belong to all individuals in society. That principle is reflected in the rule that most organizations can decide whom they want to invite as members and whom they want to exclude. That principle applies not only to organizations like partnerships and corporations that are organized for profit, but also to nonprofit organizations that have as their object religious, educational or charitable activities. It should be evident, therefore, that the principle should not be regarded as a celebration of narrow financial or economic interest over other human rights. It is instead best understood as a way to facilitate decentralized cooperation by like-minded individuals.

Why cooperation? The explanation stems from the simple proposition that voluntary cooperation is never a zero-sum game, but always operates in expectation as a positive-sum game for its participants. The words “in expectation” emphasize the simple point that collective plans that seem sensible at the outset may go awry in the end, so that the venture fails and its members are left with the unpleasant task of dividing up the losses.  But people form these organizations because on average they are confident that if they pick the right partners for some business or social venture, they will achieve through their combined efforts greater benefits than they could achieve by themselves.

The inevitable failures in some cases are thus more than offset by the far greater gains that come from the cooperative ventures that succeed. The two sets of prospects mean that any system that prizes freedom of association needs to have easy means whereby individuals can join together while simultaneously supplying means to liquidate or reorganize ventures that fail, so as to allow the redeployment of the parties and resources tied up in one venture to another.

One of the great advantages of freedom of association is that the constant application of the basic principle generates ever more complex voluntary arrangements. Thus, if two individuals decide to form a simple partnership, they can thereafter admit a third. If two partnerships decide to merge, they can draft an agreement that sets out their respective rights and obligations.

The same conditions hold with respect to commercial transactions. Goods that are sold from a manufacturer to a distributor, or from a distributor to a retailer, can be broken down into smaller units and resold to other customers and businesses. The law can enforce these agreements without any inquiry into the worthiness of their content because the parties themselves determine that they are gainers from the transactions in question. Legal enforcement is thus simplified in the knowledge that with each additional level of transactional complexity, the same basic economic relationship holds: completed transactions, whether by exchange or association, are in expectation positive sum for the parties who participate in them.

To this sunny view of voluntary transactions, two qualifications must be added. The first concerns individual actions that can subvert the mutual gain between transacting parties—chiefly fraud, misrepresentation, concealment, and nondisclosure. The second involves the effects of their transactions on third parties.

This first class of actions is important, especially within a classical liberal framework, because they can undermine the mutual-gain hypothesis on which both cooperation and trade rest. It is therefore critical to stress that in any system of laissez faire markets, these behaviors must be subject to sanctions. The exact remedies—damages, fines, rescission, jail—for these forms of misbehavior are outside the scope of this  essay, but for the most part intelligent actors in all walks of life take great precautions to choose the right trading partners and to enter into the right long-term relations to minimize the risk of breakdown in social relations. The control of fraud is consistent with classical liberal principles because it increases opportunities for third parties without forcing the state to get into the risky business of deciding which types of transactions to favor and which to penalize.

The second concerns what should be done about outsiders under this regime. The implicit bias in this question is that the externalities that are created will be negative to the people who are not asked into the new venture. But the overall situation is in fact more complex because the formation of any voluntary association or specific trading relationship generates both positive and negative externalities.

The positive externalities are often ignored because in general they do not become the focal point of legal sanctions. But their influence cannot be overlooked in any general social evaluation of the traditional system. The successful conclusion of any voluntary transaction among two or more people routinely increases the opportunities for association and trade available to everyone else. Greater freedom and greater wealth go hand in hand, and these positive externalities help explain why the enforcement and protection of these associational and exchange freedoms is a proper role for the state.

Negative externalities raise different challenges. In this context, two sorts of externalities must first be distinguished—one that justifies legal intervention and one that does not. The proper invocation of state power is to prevent the use of aggression and fraud against other individuals. It is not to redress losses from offense or competition, lest the exceptions follow the rule. Take the points in order.

Regarding aggression or force, there is no need to define these notions in some artificial or narrow sense. Both Roman law and common-law systems draw the same basic line.   Protection against force not only covers cases of hitting or burning. It also covers creating a trap in which someone else can fall, or concealing poison in the food or drink of another person. Force (in its broader sense, coercion) also includes threats of force, including those that give people a choice between their money or their lives, when entitled to both.

The simple explanation here is that these transactions are, at our best empirical guess, all negative-sum. The losses on the one side are huge, and the gains on the other side are small. The third-party effects are at best a wash in these cases, because even if the aggressor improves his position with third parties, the family, friends, and associates of the victim have doubtless suffered far greater losses.

It is not credible to think that these external effects, taken in full, can reverse the negative polarity associated with the original transactors, which is why the prosecution of aggression in discrete cases never requires an individuated knowledge of third-party effects. These are so highly correlated with the net loss that coercion creates for the parties to the transaction that it is not worth spending a fortune to track down the supposed effects on billions of people. Indeed, it is not even worth tracking down their effects on close family and friends of the parties to decide whether the conduct should be subject to some form of legal sanction.

The situation is far different with respect to offense that one person takes to the action of others from the mere knowledge that such actions are undertaken. Offense is ubiquitous, so if this counts as a form of actionable harm, anyone can get his or her way in the world by getting ever angrier at the actions of other individuals. The more you stoke the fires, the stronger your claim, which is a form of moral hazard that no set of social institutions can long survive. It is thus clear that any commitment in favor of speech does not allow the offense exception to drown it out, which is why the flag-burning cases, among others, cannot be decided in ways that legitimate the grievances of these or any other protesters.

The same point applies to persons who feel aggrieved when (subject to an exception for common carriers and public utilities) they are excluded from the activities of others.  Once universalized, their claims reduce others’ freedom to a hollow shell: those activities that are widely or universally approved may be freely undertaken. Other types of action will be sharply limited.

It is just this two-sided view of the world that has fueled the student unrest at places like Yale, Princeton and Missouri. Leftwing students attack with the most vicious language those whose cautious words give them deep offense. That two-sided standard kills free speech and inquiry because it makes clear that the willingness to control offensive speech cannot work in a neutral fashion but necessarily favors one group while squashing the sentiments of others. Standards of decorum can never be content-based.

A market competitor’s injury in the marketplace presents a variation on the same theme, namely, that it would be too socially destructive to adopt any freewheeling view of the harm principle. Some see a genuine puzzle in allowing a merchant to recover $100 in damages from someone who broke his machine tool, but denying him the recovery of even a penny from someone who took his entire customer base, costing millions, by offering superior crops at lower prices.

These complaints are commonly heard in the context of foreign trade, where the willingness to send business overseas is often said to make firms traitors to their home nation. But the criticism is wholly misguided in both the domestic and foreign context, for trade is a positive-sum game between the parties that improves the opportunities of other persons. To let competitive losses count is in effect to allow private rights of action that are inverse to overall social welfare.

The point has real historical punch. It is one of the great tragedies of the New Deal period that claims of “ruinous” competition fell on all-too-receptive ears. In rapid succession, New Deal legislation—the National Labor Relations Act, the Fair Labor Standards Act, the Motor Vehicle Act, the Robinson Patman Act, the Civil Aeronautics Act, and the Agricultural Acts, to mention only schemes at the federal level—institutionalized legal cartels that produced systematic social losses insofar as the gains to cartel members were smaller than were the social losses to the disparate consumer groups that suffered from these statutory interventions.

Both the Roman and the common law sensed the danger in this approach. In the traditional legal system, the phrase damnum absque iniuria—harm without legal injury—covered those cases where the private loss was acknowledged, but was systematically denied any legal protection at all. Other lawyers referred to these as noncognizable or nonactionable harms, to the same effect. At the same time, economists spoke of “pecuniary externalities,” an utterly impenetrable term meant to convey the sense that the plaintiffs’ private losses should not be recognized because these losses are negatively correlated with social welfare, precisely because they are always offset by positive gains.  But unless that line is maintained, competition and coercion collapse into each other, resulting in a devastating extension of state power.

It is sometimes assumed, however, that the refusal to grant all individuals and firms protection against competitive injury necessarily means that the state is also helpless to deal with the overall issue of monopoly: the control by a single seller of some particular good. That conclusion follows from a narrow libertarian framework in which only force and fraud are actionable, because voluntary cartel formation by rival traders involves neither. Nonetheless, from the earliest time, the common law—this issue did not arise in the Roman context—set its face against contracts “in restraint of trade,” that is, those contracts that sought to limit entry into given markets. By the conventional account, a monopoly reduces output, raises prices, and results in systematic social losses when third-party effects are taken into account. These social losses are not as large as those from the threat or use of force, but they are large enough to matter, which is why the exercise of monopoly power is commonly, if inexactly, called a form of coercion.

The key question is what remedy is allowed, and it is at this point that the place of antidiscrimination norms starts to come into focus, where they function as a remedy that in some cases is used to combat that risk, most concretely in cases where it is not feasible to block the formulation of the monopoly or break it up into competitive units. A legal monopoly is conferred by the state. A “natural monopoly,” a trickier concept still, arises when no firm has an economic incentive to enter the market once an initial firm has made investments to stake out its market position. The first firm has large sunk costs but low variable costs (that is, those needed to serve additional customers). New entrants cannot meet those low costs if they have to duplicate the high front-end costs.

Historically, this situation commonly arose for all sorts of public utilities, such as the railroads, electricity, gas, and the like. So the legal response was to strip the incumbent firm of the right to refuse any customer within its service area. That right to refuse business is absolutely critical to maintaining a competitive marketplace. Both buyers and sellers have to respond to external signals for markets to clear. If the state could dictate term price, competition could not take place at all. But once that private right is allowed, it is effectively hemmed in by the ability of customers to go elsewhere with their business if the price demanded climbs too high. At the limit, the prices of standardized goods in competitive markets converge to a single unique point where price equals marginal cost. The firm that raises its prices loses its customers to rivals. The firm that lowers prices cannot cover its costs.

In these situations, the market dynamic means that the refusal of one firm to deal is met by the decision of the consumer to go elsewhere. No firms are perfectly competitive in this restricted sense, but in most industries, close substitutes make it unwise to impose any restraint on how parties do transactions.

That position does not apply when a single firm has a legal or natural monopoly. Now customers have no close substitute to turn to. Parties will have to do without transportation, heat, or power. So there has to be some duty to deal. But the firm cannot name whatever price it chooses to meet the challenge. To charge $1,000 for a short train ride is tantamount to refusing to deal. To charge one customer $100 and another $500 for the same service at the same time is to play favorites, which could drive one firm out of the market while another prospers.

So the rule developed that the rates had to be reasonable and nondiscriminatory. The first requirement was intended to block an overall price structure that incorporated monopoly rates. The second was an effort to block favoritism with respect to customers in the same class. Yet to make sure that the firm did not go belly up, the overall set of fares had to be sufficient to cover the costs of capital, maintenance, and operations. The system of rate regulation thus emerged after the Civil War with the rise of the great industrial firm. An antidiscrimination norm was an essential part of the practice.

It should now be easy to see how this framework applies to various forms of discrimination. On the private side, one possible line of discrimination was race. Black customers, even after the end of slavery, could be charged higher rates and offered inferior services out of malice or favoritism to white customers. The duty to serve overcomes that view, and it is just this approach that explains historically why the decision in Plessy v. Ferguson (1896) to uphold racial segregation on privately owned trains flew in the face of the antidiscrimination principle that otherwise applied.

It is important to understand how minimal the restriction is on common carriers. Wholly apart from race, common carriers do not have to impose any demanding requirements in choosing their customers, whose main task is to sit quietly next to each other. It is for this reason that seats are assigned by anonymous devices, none of which ask about the compatibility among fellow passengers. So seating is by ticket, by time of arrival, by place in line, and so forth. In essence, the basic judgment is that the advantages of access swamp the occasional disadvantages of sitting in the aisle beside someone you don’t like in the middle seat. The basic norm applies and, in rare cases, some reseating is done to avoid tension. But the reason this system works so well is that the gains from stopping exclusion and discrimination are great, and the offense to individual autonomy is small.

The question then is how much further the antidiscrimination principle should apply. In ordinary business transactions, it is generally unwise to force someone to do business with a person he does not like. These arrangements depend upon trust and cooperation and are unlikely to work well when one person is able to force himself on another.

The same principle applies with greater force to those associations formed to pursue a common purpose, such as firms or clubs. In these cases, the ideal norm is no longer that people sit next to each other in peace and quiet. It is that they cooperate with each other, which they are not likely to do if they are not well disposed toward each other or do not share common ends and complementary skills. Thus, hiring a new employee, or inviting a new member into a business or social association, requires a lot more care and attention than letting someone sit side by side with a stranger on a train, which is just how all labor markets behave.

The common law responded to these situations. It could be said that exclusion from any voluntary organization is a harm to the party excluded. And so it is, but only in terms of offense. But the admission of one person into a firm or group against the will of its members is a harm to any members who now think the quality of their association is in decline. Membership situations are typically low-transaction-cost settings, so that if letting the outsider in makes the members better off, we can trust the parties to find that solution when deciding on their own criteria for group membership. This principle is quite universal, and one of its major functions is to protect small, marginal groups from being overwhelmed by majority groups of a different perspective, which is what happens once antidiscrimination laws go into action outside the context of common carriers and public utilities.

Modern Antidiscrimination Laws

The modern age of antidiscrimination laws began with the passage of the Civil Rights Act of 1964 (CRA). Insofar as the CRA, as in Title I, attacked systematic efforts to exclude people from voting on grounds of race, its adoption was all to the good.

Insofar as it was directed against discrimination in public accommodations, the CRA was a mixed blessing. As of 1964, the institutional racism embedded in segregation throttled the operation of competitive markets in many parts of the country by a combination of public and private devices intended to restrict entry. On the public side, it is quite likely that the ability of local white-dominated governments to control common carrier hookups effectively made it difficult for private firms to integrate in segregated towns. At the same time, the constant threat of private force—ignored or even abetted by local police—finished the job. Clearly the correction of a terrible situation required tougher measures than would be required if there had never been embedded forms of institutionalized racism in the first place.

Yet when the issue turned to employment relationships under Title VII, the analysis became more disturbing. The first mistake of the CRA in this regard was to prohibit in colorblind fashion all discrimination against “any individual . . . because of race, color, religion, sex, or national origin.” That principle rested on a conflation of the public and the private.

In the public realm, administering the criminal and civil law does require colorblind enforcement. It will not do to have one set of criminal rules for whites and another for blacks, and so on. The monopoly power of state enforcement pushes strongly for a colorblind attitude, lest favoritism dominate various forms of public discourse.

But the same logic does not apply to private firms or individuals in a private competitive market where different individuals and groups can adopt different approaches to the vexed questions of racial and sexual equality. To be sure, the drafters of the CRA thought the only form of discrimination that would be practiced if any were allowed would be against members of minority or underprivileged groups. But that was before the race riots that started in 1964, at which point attitudes started to change radically.

Prior to 1964, the conventional wisdom—expressed, for example, by Michael Sovern, a law professor and later Dean of Columbia Law School and President of Columbia University[3]—was that the colorblind principle worked fine for civil rights because, with time and improvement of the educational system, matters would straighten themselves out in the workplace.

Yet before the ink was dry on the page, the racial unrest that consumed much of the 1960s made it perfectly clear that the waiting strategy would not and could not work. At this point, the colorblind CRA acted as an impediment to private actions, including special outreach or internship programs, that might have helped redress the overall situation, as private businesses were reluctant to go head to head with legislation that pointed in the opposite direction.

In the 1979 decision in United Steelworkers of America v. Weber, Justice Brennan was able to twist the words of the CRA so as to allow for affirmative action programs on the ground that the short-term use of race-conscious devices paved the way for the long-term realization of a colorblind society. But Weber made it appear as if the right to run an affirmative action program was a special privilege conferred upon one group instead of an ordinary business decision that any firm or association could make for itself.

That perception only gets stronger when the matter is contrasted with the rise of disparate impact liability under the CRA, starting with the decision in Griggs v. Duke Power Co. (1971), which again used sleight-of-hand methods of statutory construction to impose a rigid disparate impact test that could only be overcome by proof of business necessity, under a standard so stringent that it could hardly ever be satisfied. The ostensible reason for this expansion was that the impulse to discriminate was so strong that the law had to take extra steps to stamp it out at a time when the pressure for affirmative action programs continued to grow. This massive expansion of the civil rights laws rendered illegal standard methods of testing whenever they yielded a disparate impact, which was commonly the case.

In a particularly unfortunate 1982 decision, Connecticut v. Teal, the Supreme Court held that the disparate impact test blocked employers’ use of testing even for states and businesses that operated bona fide affirmative action programs. Under current law, testing only passes muster if it is so task-specific that it provides no information to an employer as to the long-term potential of any particular employee (which was supplied by the well-established Wonderlic Personnel Test, used for measuring general intelligence, and the Bennett Mechanical Comprehension Test). Virtually all other examinations followed the same fate, without any identification of their supposed flaws. The result is that employers have to place greater reliance on inferior information, including casual social impressions, about all employees even when there is no trace of discrimination in the air.

Once launched, disparate impact introduced a massive intrusion of government into the private marketplace. And what applied to testing could carry over to every aspect of business, for there are many practices that will have a disparate impact even if they make good business sense. The net effect has been that the effort to save employment markets from invidious discrimination diminished business opportunities across the board.

They also induced, ironically, some employers to engage in discrimination. Given that the antidiscrimination laws made it harder to fire workers in the protected classes, employers had a reason not to hire them, which was indeed correlated with race. It is an open question whether a “ban the box” campaign that prohibits prospective employers  from asking about criminal records until late in the hiring process will increase the likelihood that employers, anxious to avoid time and expense, will resort to using race and age as a crude proxy for criminal conduct. But the cross-currents here are hard to decipher because they will vary from case to case.

Sometimes firms will engage in aggressive affirmative action programs. In others, they will steer clear of workers in protected classes, especially now workers who have disabilities. That last result follows inexorably once it is understood that antidiscrimination laws, especially in cases of disability, create large cross-subsidies such that the output of healthy workers has to be large enough to offset disabled workers’ lower levels of production. The net effect is of course that the willingness to hire disabled workers has declined since the 1990 passage of the Americans with Disabilities Act.

The correct response is to repeal the statute, at which point the rate of employment for disabled workers will rise so long as wage differentials can offset the extra costs of providing for their needed services, which are compounded by the greater difficulty in firing or disciplining disabled workers for subpar performance. These extra costs should not be regarded as a fixed feature of nature, given that firms will do whatever they can to increase the productivity of their workers. In many cases, that result is best achieved by investing in common facilities, for example ramps, that help disabled workers but are not necessary for others. In some instances, the use of advanced technology systems could allow disabled workers to work at their own pace without slowing down the work efforts of others.

The current law purports to require firms to make reasonable accommodations unless they cause undue hardships. That is exactly what firms will do when they can decide how to manage their own businesses. It is exactly what government officials will not do when they are given oversight of the system without having to bear the consequences of the dislocations that it causes.

The difficulties associated with the antidiscrimination laws are not of course confined to the employment relationship. They push hard in housing and educational markets, where again government officials are quick to find discrimination whether or not it exists.  Indeed, one of the most dangerous developments of recent years has been the nonstop insistence that the prohibition against discrimination in universities, for example, requires the adoption of an elaborate bureaucracy to weed out potential sources of discrimination, including racial harassment, before it even occurs.

The settlement agreement that the Department of Education’s Office of Civil Rights imposed on the Harvard Law School represents the intolerable extremes to which overzealous bureaucrats can go to impose heavy sanctions on students and faculty in situations where they are denied the most elementary due process protections. The agreement requires Harvard to expand the definition of sexual harassment beyond recognition, and to be somehow responsible for the actions of adult students when they are not on the Harvard campus.

The university must use truncated procedures to establish guilt by a simple preponderance of evidence; consolidate into a single office all activities pertaining to investigation, prosecution, fact-finding, and appellate review; truncate rights to mount a defense by limiting the ability to discover relevant facts prior trial, or to confront adverse witnesses; and deny persons accused of serious wrongs the right to defend themselves.  These excesses all happen precisely because the interest in preventing various forms of discrimination, including sexual harassment under definitions that are far broader than the threat or use of force, are rated so high that nothing is allowed to stand in the way.

That same attitude is found all too often in dealing with the interaction between the antidiscrimination laws based on sexual orientation and the protection of religious liberty.  It is often said that discrimination on the grounds of sexual orientation is as odious as discrimination on the grounds of race. But the analogy goes in the wrong direction.  People in a free society who do not like racial discrimination are free not to associate with the individuals or firms who practice it. Indeed, in all likelihood, that is exactly what would happen if some large firm announced a policy of not hiring blacks, Muslims, or women—which is all the more reason to give these private entities the freedom to set out their own employee qualifications. The key point here is that they are never free to impose their will on others.

The most ardent defenders of the antidiscrimination norm are typically strong defenders of affirmative action programs on the ground that their form of discrimination is justified while traditional forms of discrimination are not. They are of course free to raise that argument privately in any organization they choose, where they can do so with great effect in many contexts today. But the consensus that affirmative action programs are proper should not be an excuse to impose them on those institutions that think otherwise, which should be free to follow a colorblind policy if they choose, or to give preferences to members of their own religion or ethnic group, and bear the economic and social consequences of their choices.

The recent issues with so-called human rights laws reveal the deep totalitarian impulse that activates their supporters. The stark challenge raised in these cases is the insistence that all businesses have to serve gay and lesbian customers or face heavy fines, even if same-sex marriage is against their deeply held religious beliefs. The arguments invoked to sustain that position are insupportable in any situation where the services needed—wedding cakes, wedding photography—are competitively supplied in any market.

The plea of personal anguish and offense from not being served carries no more weight in this context than it does in any other: get over it, tell your friends not to patronize the store, or write an angry letter to the local newspaper. But these last two counterstrategies have risks. Some of your friends may not agree, and the proprietors could write a letter noting that their objections are not to serving gay and lesbian couples as such, but only in areas that contradict their religious beliefs. And they can easily add that the ultimatum to either perform services against conscience or go out of business causes anguish to those whose options in the business world are starkly limited by an intolerant majority that treats the eradication of discrimination like the eradication of the plague.

This kind of intolerance has unfortunately informed most of the fierce opposition to the Supreme Court’s 2014 decision in Burwell v. Hobby Lobby Stores, Inc., in which a narrow majority of the Supreme Court upheld the refusal of faith-based businesses to supply contraceptive and abortifacient services against their own religious beliefs. The critics of Hobby Lobby often claim that the state has a compelling interest in preventing discrimination, when in fact it has no interest in combatting it collectively in competitive industries.

Equally misguided is the constant refrain that “the boss” cannot “impose” his religious beliefs on others. That would be the case if the boss insisted, as none has, that no workers be allowed to spend even their own money to purchase contraceptive devices. But there is a strong claim of state coercion against employers if the law may put them the following choice: either spend their own money for purposes of supplying services that go against their conscience, or be prohibited from supplying any employee health insurance at all.

In these settings, the utter inability to understand who is coercing whom reveals the great danger of a kind of law that forces the unwilling to yield to its commands when all they want is to be free to practice their own faith. The response that the only protection the Free Exercise Clause supplies is in religious worship gives an indefensibly narrow reading of the term “free exercise of religion,” for that term extends to all areas of life, not just those approved by some hostile public official. The standard limitations of classical liberal theory apply in this instance. The free exercise of religion, like all other freedoms, does not authorize the use of force or fraud against other individuals. The duty to serve remains in cases where the state supplies monopoly services, so that county clerk Kim Davis of Rowan County, Kentucky should be allowed to absent herself from signing the marriage licenses of gay and lesbian couples, but only so long as someone else in the office discharges that ministerial task.

The Indispensability of “Live and Let Live”

In sum, there are today deep divisions in American culture as to the proper way to organize human relations on all the lines that are covered by antidiscrimination laws. It is highly doubtful that people on opposite sides of this contentious set of issues will be able to persuade their opponents of the rightness of their own views. But it is precisely for that reason that antidiscrimination laws are such a great mistake outside of monopolies.

The point applies not only in cases where the public is evenly divided, but even more so when there is some large social consensus in favor of a political view. People who sit in the catbird seat have no reason to impose their will on the minority that disagrees with them. That isolated minority is not prepared to use force against the majority or otherwise block its activities. But the dominant majority can all too easily coerce the minority on all things great and small to bow to their will, increasing alienation along the way. In general it is always a good idea to encourage voluntary cooperation between and across different groups. It is quite another thing to mandate it.

Toleration is not a virtue that is called into play with people who are in fundamental agreement with each other. But it is an indispensable social norm in cases of large-scale disagreement, such that everyone has to learn to tolerate behaviors of others with whom they disagree. Live-and-let-live is not some idle social maxim. It is the key to social success when public sentiments are sharply divided. Where toleration works, cooperation may follow. Where toleration is rejected, civil strife will increase.

[1] My thanks to Rachel Cohn and Krista Perry, University of Chicago Law School, Class of 2016 and Julia Haines, University of Chicago Law School, Class of 2017 for their excellent research assistance.

[2] Clay Risen, The Bill of the Century: The Epic Battle for the Civil Rights Act (Bloomsbury Press, 2014).

[3] Michael I. Sovern, Legal Restraints on Racial Education in Employment (Twentieth Century Fund, 1966).