Skip to main content

Affirmative Action: Don’t Mend It or End It – Bend It

Affirmative action policy—by which I mean ethno-racial preferences in the allocation of socially valuable resources—is even more divisive and unsettled today than at its inception more than 30 years ago.

Affirmative action’s policy context has changed dramatically since 1970. One change is legal. Since the Supreme Court’s 1978 Bakke decision, when Justice Lewis Powell’s pivotal fifth vote endorsed certain “diversity”-based preferences in higher education, the Court has made it increasingly difficult for affirmative action plans to pass constitutional muster unless they are carefully designed to remedy specific past acts of discrimination. Four other changes—the triumph of the nondiscrimination principle; blacks’ large social gains; evidence on the size, beneficiaries, and consequences of preferences; and new demographic realities—persuade me that affirmative action as we know it should be abandoned even if it is held to be constitutional.

s we know it” is the essential qualifier in that sentence. I propose neither a wholesale ban on affirmative action (“ending” it) nor tweaks in its administration (“mending” it). Rather, I would make two structural changes to curtail existing preferences while strengthening the remaining ones’ claim to justice. First, affirmative action would be banned in the public sector but allowed in the private sector. Second, private-sector institutions that use preferences would be required to disclose how and why they do so. These reforms would allow the use of preferences by private institutions that believe in them enough to disclose and defend them, while doing away with the obfuscation, duplicity, and lack of accountability that too often accompany preferences. Affirmative action could thus be localized and customized to suit the varying requirements of particular contexts and sponsors.

Triumph of the Nondiscrimination Principle

Why is change necessary? To explain, one must at the outset distinguish affirmative action entailing preferences from nondiscrimination, a principle that simply requires one to refrain from treating people differently because of their race, ethnicity, or other protected characteristics. Although this distinction can blur at the edges, it is clear and vital both in politics and in principle.

When affirmative action became federal policy in the late 1960s, the nondiscrimination principle, though fragile, was gaining strength. Preferences, by contrast, were flatly rejected by civil rights leaders like Hubert Humphrey, Ted Kennedy, and Martin Luther King, Jr. In the three decades that followed, more and more Americans came to embrace nondiscrimination and to oppose affirmative action, yet as John Skrentny shows in his Ironies of Affirmative Action, federal bureaucrats extended affirmative action with little public notice or debate. Today, nondiscrimination, or equal opportunity, is a principle questioned by only a few bigots and extreme libertarians, and civil rights law is far-reaching and remedially robust. In contrast, affirmative action is widely seen as a demand for favoritism or even equal outcomes.

Social Gains by Blacks

Blacks, the intended beneficiaries of affirmative action, are no longer the insular minority they were in the 1960s. Harvard sociologist Orlando Patterson shows their “astonishing” progress on almost every front. “A mere 13% of the population,” he notes, “Afro-Americans dominate the nation’s popular culture…. [A]t least 35 percent of Afro-American adult, male workers are solidly middle class.” The income of young, intact black families approaches that of demographically similar whites. On almost every other social index (residential integration is a laggard), the black-white gap is narrowing significantly; indeed, the income gap for young black women has disappeared.

Even these comparisons understate black progress. Much of racism’s cruel legacy is permanently impounded in the low education and income levels of older blacks who grew up under Jim Crow; their economic disadvantages pull down the averages, obscuring the gains of their far better-educated children and grandchildren. These gains, moreover, have coincided with the arrival of record numbers of immigrants who are competing with blacks. To ignore this factor, economist Robert Lerner says, is like analyzing inequality trends in Germany since 1990 without noting that it had absorbed an entire impoverished country, East Germany. In addition, comparisons that fail to age-adjust social statistics obscure the fact that blacks, whose average age is much lower than that of whites, are less likely to have reached their peak earning years.

My point, emphatically, is not that blacks have achieved social equality—far from it—but that the situation facing them today is altogether different than it was when affirmative action was adopted. Advocates, of course, say that this progress just proves that affirmative action is effective; hence it should be continued or even increased. But this post hoc ergo propter hoc reasoning is fallacious and ignores the policy’s growing incoherence and injustice.

Size, Beneficiaries, and Consequences of Preferences

When we weigh competing claims for scarce resources—jobs, admission to higher education, public and private contracts, broadcast or other spectrum licenses, credit, housing, and the like—how heavy is the thumb that affirmative action places on the scales? This is a crucial question. The larger the preference, the more it conflicts with competing interests and values, especially the ideal of merit—almost regardless of how one defines merit.

The best data concern higher education admissions where (for better or for worse) schools commonly use standardized test scores as a proxy for aptitude, preparation, and achievement. William Bowen and Derek Bok, the former presidents of Princeton and Harvard, published a study in 1999 based largely on the academic records of more than 80,000 students who entered 28 highly selective institutions in three different years. Affirmative action, they claimed, only applies to these institutions, although a more recent study suggests that the practice now extends to some second- and even third-tier schools.

Selective institutions, of course, take other factors into account besides race. Indeed, some whites who are admitted have worse academic credentials than the blacks admitted under preferences. Still, Bowen and Bok find a difference of almost 200 points in the average SAT scores of the black and white applicants, and even this understates the group difference. First, the deficit for black applicants’ high school grade point average (GPA), the other main admission criterion, is even larger. Thomas Kane finds that black applicants to selective schools “enjoy an advantage equivalent to an increase of two-thirds of a point in [GPA]—on a four-point scale—or [the equivalent of] 400 points on the SAT.” Second, although the SAT is often criticized as culturally biased against blacks, SAT (and GPA) scores at every level actually overpredict their college performance. Third, the odds were approximately even that black applicants with scores between 1100 and 1199 would be admitted, whereas the odds for whites did not reach that level until they had scores in the 1450-1499 range. With a score of 1500 or above, more than a third of whites were rejected while every single black gained admission. The University of Michigan, whose affirmative action program is detailed in a pending lawsuit, weighs race even more heavily than the average school in the Bowen and Bok sample. At Michigan, being black, Hispanic, or Native American gives one the equivalent of a full point of GPA; minority status can override any SAT score deficit. And a recent study of 47 public institutions found that the odds of a black student being admitted compared to a white student with the same SAT and GPA were 173 to 1 at Michigan and 177 to 1 at North -Carolina State.

These preferences, then, are not merely tie-breakers; they are huge—

How much of blacks’ impressive gains is due to reduced discrimination resulting from changing white attitudes and civil rights enforcement, as distinct from preferences? How would they have fared had they attended the somewhat less prestigious schools they could have attended without preferences? What would the demographics of higher education be without those preferences? We cannot answer these vital questions conclusively. We know that black gains were substantial even before preferences were adopted, that preference beneficiaries are overwhelmingly from middle- and upper-class families, and that most black leaders in all walks of life did not go to elite universities. We also know that many institutions are so committed to affirmative action that they will find ways to prefer favored groups—, legacies, athletes, and others—no matter what the formal rules say. Although California voters banned affirmative action in state programs, their politicians press the university system to jigger the admission criteria until it finds a formula that can skirt the ban and produce the “correct” number of the favored minorities (excluding Asians, who are thought not to need the help).

New Demographic Realities

The moral case for affirmative action rests on the bitter legacy of black slavery, Jim Crow, and the violent dispossession of Native Americans. Yet the descendants of slaves and Native Americans constitute a shrinking share of affirmative action’s beneficiaries. Political logrolling has extended preferential treatment to the largest immigrant group, Hispanics, as well as to blacks from Africa, the Caribbean, and elsewhere, Asians and Pacific Islanders, and in some programs to women, a majority group.

Some affirmative action advocates acknowledge this problem and want to fix it. Orlando Patterson, for example, would exclude “first-generation persons of African ancestry” but not “their children and later generations—in light of the persistence of racist discrimination in America.” He would also exclude all Hispanics except for Puerto Ricans and Mexican Americans of second or later generations and would exclude “all Asians except Chinese-Americans descended from pre-1923 immigrants….” With due respect for Patterson’s pathbreaking work on race, his formula resembles a tax code provision governing depreciation expenses more than a workable formula for promoting social justice.

Centuries of immigration and intermarriage have rendered the conventional racial categories ever more meaningless. The number of Americans who consider themselves multiracial and who wish to be identified as such (if they must be racially identified at all) was 7 million in the 2000 census, including nearly 2 million blacks (5 percent of the black population) and 37 percent of all Native Americans. This is why advocacy groups who are desperate to retain the demographic status quo lobbied furiously to preempt a multiracial category.

In perhaps the most grimly ironic aspect of the new demographic dispensation, the government adopted something like the one-drop rule that helped enslave mulattos and self-identifying- whites before Emancipation. Under OMB’s rules, any response combining one minority race and the white race must be allocated to the minority race. This, although 25 percent of those in the United States who describe themselves as both black and white consider themselves white, as do almost half of Asian-white people and more than 80 percent of Indian-white people. The lesson is clear: making our social policy pivot on the standard racial categories is both illogical and politically unsustainable.


Even a remote possibility that eliminating affirmative action would resegregate our society deeply distresses almost all Americans. Nothing else can explain the persistence of a policy that, contrary to basic American values, distributes valuable social resources according to skin color and surname. But to say that we must choose between perpetuating affirmative action and eliminating it entirely is false. To be sure, most suggested reforms—using social class or economic disadvantage rather than race, choosing among minimally qualified students by lottery, and making preferences temporary—impracticable or would make matters worse. Limiting affirmative action to the descendants of slaves and Native Americans would minimize some objections to the policy but, as Patterson’s proposal suggests, would be tricky to implement and would still violate the nondiscrimination and merit principles.

Most Americans who favor affirmative action would probably concede that it fails to treat the underlying problem. Black applicants will continue to have worse academic credentials until they can attend better primary and secondary schools and receive the remediation they need. A root cause of their disadvantage is inferior schooling, and affirmative action is simply a poultice. We must often deal with symptoms rather than root causes because we do not know how to eliminate them, or consider it too costly to do so, or cannot muster the necessary political will. If we know which social or educational reforms can substantially improve low-income children’s academic performance, then we should by all mean adopt them. But this does not mean that we should preserve affirmative action until we can eliminate the root causes of inequality.

I propose instead that we treat governmental, legally mandated preferences differently than private, voluntary ones. While prohibiting the former (except in the narrow remedial context approved by the Supreme Court), I would permit the latter—but only under certain conditions discussed below. A liberal society committed to freedom and private autonomy has good reasons to maintain this difference; racial preferences imposed by law are pernicious in ways that private ones are not. To affirmative action advocates, it is a Catch-22 to bathe benign use of race now after having used it against minorities for centuries. But to most Americans (including many minorities), affirmative action is not benign. It is not Catch-22 to recognize what history teaches—that race is perhaps the worst imaginable category around which to organize political and social relations. The social changes I have described only reinforce this lesson. A public law that affirms our common values should renounce the distributive use of race, not perpetuate it.

There are other differences between public and private affirmative action. A private preference speaks for and binds only those who adopt it and only for as long as they retain it. It does not serve, as public law should, as a social ideal. As I explained in The Limits of Law: Essays on Democratic Governance (2000), legal rules tend to be cruder, more simplistic, slower to develop, and less contextualized than voluntary ones, which are tailored to more specific needs and situations. Legal rules reflect interest group politics or the vagaries of judicial decision; voluntary ones reflect the chooser’s own assessment of private benefits and costs. Legal rules are more difficult to reform, abandon, or escape. Voluntary ones can assume more diverse forms than mandated ones, a diversity that facilitates social learning and problem solving.

Still, many who believe in nondiscrimination and merit and who conscientiously weigh the competing values still support affirmative action. If a private university chooses to sacrifice some level of academic performance to gain greater racial diversity and whatever educational or other values it thinks diversity will bring, I cannot say—nor should the law say—that its choice is impermissible. Because even private affirmative action violates the nondiscrimination principle, however, I would permit it only on two conditions: transparency and protection of minorities. First, the preference—its criteria, weights, and reasons—must be fully disclosed. If it cannot withstand public criticism, it should be scrapped. The goal is to discipline preferences by forcing institutions to reveal their value choices. This will trigger market, reputational, and other informal mechanisms that make them bear more of the policy’s costs rather than just shifting them surreptitiously to nonpreferred applicants, as they do now. Second, private affirmative action must not disadvantage a group to which the Constitution affords heightened protection. A preference favoring whites, for example, would violate this condition.

The Commitment to Legal Equality

For better and for worse, American culture remains highly individualistic in its values and premises, even at some sacrifice (where sacrifice is necessary) to its goal of substantive equality. The illiberal strands in our tangled history that enslaved, excluded, and subordinated individuals as members of racial groups should chasten our efforts to use race as a distributive criterion. Affirmative action in its current form, however well-intended, violates the distinctive, deeply engrained cultural and moral commitments to legal equality, private autonomy, and enhanced opportunity that have served Americans well—even though they have not yet served all of us equally well.

Peter H. Schuck is professor of law at Yale University and New York University. This article is drawn from a chapter in a forthcoming book, Diversity in America: Keeping Government at a Safe Distance (Harvard).

Get daily updates from Brookings