Is there a middle ground on affirmative action, an oasis between radical color-blindness on the right and racial quota-mongering on the left? As President Clinton prepares to unveil his conclusions on the subject, it's hard not to sympathize with his political predicament, but hard also not to anticipate his speech with a sense of dread. Having raised expectations so dramatically, he no longer has the luxury of embracing contradictory positions, or retreating into euphemisms. But is his task impossible? It's an urgent question, made all the more so by the inadequacy of the two alternatives.
Color-blindness, for all its moral and political appeal, is not really a practical option. When asked point-blank, few conservatives are honestly willing to accept the widespread resegregation that would follow from a rigid ban on racial preferences. As the gap between the lsat scores of white and black candidates at the University of Texas shows, the end of affirmative action would mean, in many cases, a return to lily-white universities and workplaces (see "Is Affirmative Action Doomed?" TNR, October 17, 1994). Moreover, even if the Congress decided tomorrow to prohibit discrimination in favor or against any person on the basis of race or gender, the questions at the heart of affirmative action would endure: How do you identify " discrimination"? Should it be inferred from intent or result? Are informal quotas the safest way of avoiding it? Both the California civil rights initiative and the federal statute endorsed by Senator Bob Dole, for example, forbid discrimination but don't manage to define it.
To make matters worse, affirmative action bureaucracies and beliefs are so entrenched that a sweeping color-blind law might lead to a renewed attack on objective standards. When Congress forbade race-norming in the Civil Rights Act of 1991, for example, many employers dropped their aptitude tests altogether, or set the cut-off points for all candidates at the lowest possible levels: they were more concerned with keeping up appearances than obeying the spirit of the law. The desire to "look like America," in short, has become too much a part of our social fabric to be suppressed by a legal rule.
The left-wing attempt to legislate equality of outcomes is just as utopian. The expectation that each race and gender should, always and everywhere, be represented in numbers that reflect its position in society at large is doomed to fail, and to stir racial resentment at the same time. Blaming white racism for the failure of minorities to be proportionally represented in all professions is no longer entirely plausible. If more than 40 percent of the freshman class at Berkeley is Asian American, and if more than 30 percent of law professors at Berkeley are Jewish, this sort of "over-representation" reflects the fact that talents and ambitions are distributed differently in different fields.
Similarly, if only a fraction of women are now CEOs as the tendentious federal Glass Ceiling Commission reported in March, this is hardly a result of discrimination alone. It takes twenty-five years of full-time work to rise to the top of a field; and women have only recently worked full-time in large numbers. (In 1968, women left work for ten years, on average, after their children were born; in 1987, the average was six months.) The cookie-cutter ideal of proportionalism does an injustice not only to white men, but to women and men of all ethnic groups who, through their own efforts, have succeeded beyond their numbers.
Some factions in the Clinton administration, unfortunately, refuse to acknowledge that proportionalism has costs. Consider the Labor Department's draft report, released on March 23, which purports to review all federal cases involving reverse discrimination between 1990 and 1994. Prepared by Alfred Blumrosen of Rutgers University, the report concludes that reverse discrimination cases represent less than 3 percent of all discrimination cases; that plaintiffs prevailed in "only" 30 percent of the twenty-one cases reported; and that "nothing in these cases would justify dismantling the existing structure of equal employment opportunity programs."
The report is gratuitously sarcastic--it ridicules the unsuccessful plaintiffs with headlines such as "the sleeping supervisor" and "the professor who didn't like competition"--and statistically meaningless. First, the sample of twenty-one cases is so small the results might as well be random. And even if it's true that plaintiffs win in 30 percent of reverse discrimination cases, the number is hardly low. As law professors George Priest and Benjamin Klein have pointed out, when one party in a civil suit is much more likely to win than the other, the case tends to settle before going to trial. Therefore, if 30 percent of reverse discrimination plaintiffs prevail in court, the number is comparable to the percentage of plaintiffs who win medical malpractice cases (32 percent) and product liability cases (42 percent). This doesn't mean that doctors aren't committing malpractice, or employers aren't discriminating against white men. It means that doctors and employers, when clearly guilty, are eager to avoid trials that they're likely to lose.
After acknowledging the need for a middle ground, Clinton would be in a perfect position to define it. He might start with a clear distinction: mandatory race and gender preferences, imposed by federal statutes and executive order, are more troubling than those adopted voluntarily by private employers. Rather than trying to take a position on all affirmative action in every sphere of society, Clinton could limit himself to examining the programs that he is directly responsible for administering: the 160 federal race and gender preferences collected by the Congressional Research Service at Senator Dole's request.
In his more moderate moments, Clinton has toyed with two alternatives to the current regime: class-based preferences and affirmative action along the military model. The class-based alternative (see "Class, Not Race," by Richard Kahlenberg, TNR April 3) would be especially perverse for a president who is concerned, above all, with racial diversity. A college board survey, described in Andrew Hacker's Two Nations, reveals the problem starkly. In 1992 the average combined SAT score for black students whose parents earn more than $70,000 a year was 854, twenty-five points lower than the average SAT for white students whose parents earn less than $20,000. This statistic fatally undermines the premise that disadvantage is a useful proxy for race; and it suggests that need-based preferences, honestly applied, would replace middle-class black students with lower-class white students. Like the color- blind rule, it would produce virtually all-white universities.
Class-based preferences become even more perverse when they're extended beyond university admissions. Imagine two candidates for a promotion to office manager, one of whom got the entry-level job ten years ago because his father was a sharecropper but who is now, thanks to the preference, a middle- class professional. Can he dramatically resurrect his former disadvantage to leapfrog over a colleague named Cabot who, despite a Skull and Bones pedigree, is currently his social and economic equal? The prospect is absurd in a nation founded on the idea that social class is always mutable; and the administration, presumably, will give class-based preferences a discreet burial.
Clinton's second model for affirmative action--the military--is also problematic. The most obvious flaw in the analogy is the abundance in the military of able African American recruits. More than 5 percent of all black men between the ages of 18 and 21 now apply for military service--nearly twice the rate of white applicants. David Armor of George Mason University estimates that if the military admitted all applicants on the basis of a lottery, about half of the force would be African American. But ever since the 1970s, the military has required all candidates to take aptitude tests, and eliminated those in the bottom quartile. During the 1980s and 1990s, about one-half of all black candidates, and one-third of white candidates, were ineligible to join the military, mainly because of low scores.
Because the military has so many able minorities to choose from, it can eliminate the weakest candidates and still achieve a disproportionately large number of African American cadets. Nearly one-third of the enlisted men in the Army are now black, more than three times the percentage of blacks in civilian society. This is hardly a useful model for university admissions or federal employment, where the far smaller numbers of minority applicants create hydraulic pressure to apply different standards to white and black candidates.
The authoritarianism of the military also makes it an unlikely model for the country at large. To enforce racial harmony, the Army has imposed a range of social controls that would hardly be tolerated in civilian society, including extensive surveillance techniques, "shock treatment" sensitivity training and ruthless suppression of racial epithets. Soldiers are not allowed to join unions or strike in the middle of battles. It's possible, therefore, to admire the military's successes in race relations while recoiling at the thought of trying to duplicate its methods in a liberal democracy.
Despite the obvious problems with the analogy, the military does have some useful lessons for Clinton's affirmative action review. The White House has prominently cited the work of Charles Moskos of Northwestern University, who recently met with Clinton. Moskos points to three core principles of affirmative action in the military (see "How Do They Do It?" TNR, August 5, 1991). First, the Army eschews quotas, but it does have goals. Promotions boards are encouraged to promote minority and female officers equivalent to their percentages in the promotion pool, not to the number of minorities and women in the Army as a whole. Second, white and black candidates are held to the same aptitude standards: there is no hint of separate promotion lists. Most important of all, there are no "timetables" to meet the goals. When there are not enough qualified candidates, Moskos claims, the Army refuses to relax its standards.
But having endorsed Moskos's three principles of affirmative action in the military, Clinton will need an iron will to enforce them. For many of the 160 race and gender preferences administered by the federal government violate all three of Moskos's principles. Most dramatically, Executive Order 11,246 and its accompanying regulations require federal contractors to adopt "goals and timetables" pegged to rigid expectations of proportional representation: exactly what Moskos says the military rejects. Similar requirements pervade the federal register. "On a statistical basis," says a Labor Department regulation, "staff representative of the racial and ethnic characteristics in the workforce shall be distributed in substantially the same proportion among (1) all job groups' ... and (2) all offices in the plan(s)." Even more crudely, Education Department guidelines provide that whenever there is a significant "under-representation or overrepresentation" of protected groups, the imbalance will be presumed to result from discrimination.
In addition to creating coercive incentives to achieve racial proportionalism, Executive Order 11,246 has failed to achieve its purported goal of expanding opportunity for needy minorities. The most careful survey of the econometric literature, conducted by George Rutherglen of the University of Virginia, suggests that the federal executive order has modestly increased the number of African Americans employed by federal contractors (in the neighborhood of .15 percent annually), but has had little long-term effect on wages. If Clinton is looking for a dramatic gesture as proof of his good faith, therefore, he should repeal Executive Order 11,246, a relic of the Johnson era that has outlived its purposes.
What about the other race and gender preferences administered by the federal government? Clinton, of course, has no power to repeal congressional statutes. But since he has promised to review the entire range of federal race and gender programs, determining "what works and what doesn't," he will have to take a position on the rest of the 160 racial preference programs compiled by the Congressional Research Service. Despite the daunting length of the list, the preferences tend to be minute variations on similar themes; and they can be divided into three broad categories--roughly corresponding to the education market, capital market and labor market.
The first and least controversial category consists of federal aid targeted at "historically black colleges and universities." Between 1992 and 1995, for example, historically black colleges were exempted from a regulation disqualifying schools with a high default rate from receiving student loans. Other programs involve set-asides for federal grants: 15 percent of National Science Foundation funds for faculty exchanges are targeted at institutions with "a substantial percentage of minority students." But, although the educational merits of historically black colleges are hotly disputed (according to the American Council on Education, their retention rates for black students are now no higher than predominantly white colleges), all of them are at least formally open to whites as well as blacks. This sort of aid seems less discriminatory than the minority scholarship at the University of Maryland whose constitutionality is being challenged because it is limited to blacks. Clinton can defend the programs in this category as affirmative action at its most benign.
The second broad category on the list is based on the assumption that women and minorities have a harder time getting access to capital than white men do. For example, a long list of federal agencies "encourage" grant recipients to use minority and female-owned banks, including the Departments of Agriculture, Labor and Veterans Affairs, the Rural Economic Development Program and the special Supplemental Food Program for Women, Infants, and Children. The Resolution Trust Completion Act (which expires this year) included a " minority preference in acquisition of institutions in predominantly minority neighborhoods." And other regulations award "preference points" to expand bank ownership by minorities and women. Are these preferences really necessary? The Equal Credit Opportunity Act, passed in 1974, forbids lending discrimination on the basis of race or gender; and the Community Reinvestment Act, passed in 1977, requires that banks lend in low- and moderate-income communities. Vigorous enforcement of these two laws seems like a more effective way of rooting out lending discrimination than resorting to special preferences for minority-owned banks. In 1992 there were only thirty-six banks owned by blacks and six banks owned by women in the country, which makes it unlikely that they can have much influence on the credit markets, no matter how much "encouragement" they get from the government. And the evidence is mixed about whether black-owned banks actually have a better record of lending to minorities: some studies suggest that they may be more cautious because their assets are more vulnerable.
The most questionable program in this category is the Federal Communication Commission's bidding preferences for minorities and women in auctions for cellular telephone licenses. In 1993 Congress told the FCC "to ensure that small businesses, rural telephone companies, and businesses owned by members of minority groups and women" are given the opportunity to own radio and cellular licenses. The commission, accordingly, set aside one-third of its licenses for small businesses, minorities and women. Small businesses with gross revenues under $40 million get a bidding credit of 10 percent. But the Achilles' heel of the program is that businesses owned by women or minorities get an additional 15 percent bidding credit, whether or not they are small-- that is, even if they have revenues over $40 million. The FCC defines minorities, as " t hose of Black, Hispanic Surnamed, American Eskimo, Aleut, American Indian and Asiatic American extraction."
Congress and the FCC don't even try to justify the program in the name of diversity of the airwaves, or compensatory justice. Instead, they claim that bidding preferences are necessary because women and minorities face discrimination in the commercial credit markets. But this premise is hard to accept in a program that favors successful minorities and women, who may have an easier time getting capital than their smaller, non-minority competitors. To justify its empirical claims, the commission relies heavily on a 1992 study by the Federal Reserve Bank of Boston, which concluded that black and Hispanic applicants in the Boston area are roughly 60 percent more likely to be denied a mortgage loan than white applicants with similar income and education. But the premise of the Boston Fed study has been hotly contested. If lending discrimination were a serious problem, only the strongest black applicants would get loans; and the default rate for black borrowers would be lower than the default rate for whites. In fact, as Jonathan Macey of Cornell Law School points out, the reverse is true. A recent study by the Federal Reserve Board compared default rates among 220,000 loans insured by the Federal Housing Administration in the late 1980s. The study found that the average default rate for blacks was 9 percent, significantly above the default rate for Asians (3.2 percent), whites (4.3 percent), Hispanics (5.1 percent) and American Indians (6 percent). All this suggests that, far from discrimination against black borrowers, banks may be discriminating in favor of them.
In the scheme of things, the FCC's bidding preference might be defended as less of a displacement of the free market than crude set-asides for minority contractors. The government is providing a license to compete, rather than a guarantee of monopoly profits; and minority entrepreneurs who make bad business decisions will lose money for themselves and their investors. Nevertheless, the real question is why the government should be going out of its way to help successful minority businesses in the first place. If Clinton is really concerned about trimming away the excesses of affirmative action, then the bidding preferences for minorities and women, like the infamous tax credits for conglomerates that sell to minorities, are ripe for repeal.
The third and largest category of preferences on the Congressional Research Service list includes minority business set-asides for federal contractors. They are modeled on Section 8(d) of the Small Business Act, which requires a certain percentage of federal contracts--at least 10 percent and usually more- -to be set aside for "socially and economically disadvantaged business," defined as businesses owned by the following groups: "Black Americans; Hispanic Americans; Native Americans (American Indians, Eskimos, Aleuts or Native Hawaiians); Asian Pacific Americans (persons with origins from Japan, China, the Philippines, Vietnam, Korea, Samoa, Guam, U.S. Trust Territory of the Pacific Islands, Northern Mariana Islands, Laos, Cambodia or Taiwan)." And women, too.
As Drew Days, the U.S. solicitor general, argued as a Yale law professor, this is one of the largest and least defensible of federal affirmative action programs. Despite its rhetoric about correcting past discrimination, Congress never convincingly established a compensatory rationale for this program when it was established in 1978--there was little evidence of discrimination against recent Asian immigrants in the construction industry-- and in the succeeding two decades, the program has drifted even further from its flimsy moorings. The Department of Veterans Affairs, the State Department and the Secretaries of Commerce and Agriculture are required to "establish regular performance monitoring and reporting system to assure that goals are being achieved." Even the Resolution Trust Corporation must set aside work for minorities, although the agency didn't exist when the original set-aside was created in 1978, and so is hardly atoning for past sins.
Minority set-asides for contractors violate all three principles of affirmative action along the military model. The rigid numerical targets must be considered quotas rather than a goal. There is no effort to tailor the goal to the number of qualified firms in the relevant labor market: Colorado, which has a tiny Aleut population, must meet the same minimum targets as Alaska. Furthermore, because prime contractors do most of their own work on highway projects, only a small portion of every job is available for subcontracting to begin with. Therefore, to meet the rigid requirement that 10 percent of the entire job must be set aside for minorities, contractors must often reserve virtually all of their subcontracting work for minority firms. Finally, the program is so easily abused by front men, and has been so polluted by corruption, that it gives the rest of affirmative action a bad name.
The minority set-aside program is now being challenged before the Supreme Court. A sweeping decision in June, striking down the program as unconstitutional, would be Bill Clinton's wildest dream: it would relieve him of the need to take a position on the issue. Still, federal set-asides are so deeply entrenched that even a thunderbolt from the Court is unlikely to obliterate them. In May, 1992, for example, the U.S. Court of Appeals for the D.C. Circuit struck down the District of Columbia's minority business enterprise program, which was modeled on the federal set-asides. A month later, the D.C. City Council responded by renaming the program the "Equal Opportunity for Local, Small and Disadvantaged Business Enterprise Emergency Act" and establishing a 50 percent set-aside "to provide a preference for local and disadvantaged business enterprises in all available areas of District Government contracting."
Aside from the name, however, nothing else has changed. At the beginning of April, for example, the District held a "maxaccess Minority Business Conference" at the Washington Convention Center, promising "Direct access to key people awarding $2 billion in local government (MD, DC, VA) contracts each year." Although the District's refurbished program is supposedly based on need rather than race, the conference revealed that the D.C. government grants "reciprocity" to minority businesses registered in surrounding counties, where the definitions are far more lax. Fairfax County, for example, defines a minority business enterprise as a business, no matter how large or successful, controlled by "African Americans, Hispanic Americans, Asian Americans, American Indians, Eskimos and Aleuts, women regardless of race or ethnicity, all persons with a physical or mental impairment that substantially limits one or more of the major life activities of such individuals, a record of such impairments, or who are regarded as having such impairments." While claiming to focus on economic disadvantage, in short, D.C. still awards contracts on the basis of an expanding list of victimhood.
Unless the justices swoop down to his rescue, Clinton will have to take some position on the minority set-asides, which are, after Executive Order 11, 246, the largest preference program the federal government administers. By any measure--remedying past discrimination, helping the disadvantaged or preventing future discrimination--the program is hard to defend. Clinton's options are to repeal the set-asides entirely, or reformulate them as small business preferences that would benefit needy entrepreneurs regardless of race. But this would require real, rather than cosmetic, changes in contracting as it is currently practiced.
If Bill Clinton is serious about separating defensible from indefensible affirmative action, in short, he will have to eliminate most of the mandatory racial preferences that the federal government now administers. But by acting swiftly and decisively on this score, Clinton would have the political and moral capital to draw a crucial distinction: federally mandated affirmative action, which puts the U.S. government in the business of classifying its citizens by race and gender, is more offensive to the ideal of equal citizenship than private affirmative action, voluntarily adopted by individual employers. In practice, those who feel aggrieved by affirmative action may not especially care if they are victims of preferences mandated by the government or chosen by private employers. But Clinton can seize the high ground by embracing a principle of presidential neutrality as the centerpiece of his middle way: affirming the non-discrimination principle in the public sphere; and leaving it up to employers to choose or reject race-consciousness in the private sphere.
In practice, of course, affirmative action as most Americans experience it would be largely undisturbed. In corporate bureaucracies, it's easier to administer crude statistical goals than to develop nuanced rules. And vigorous enforcement of the Civil Rights Act of 1991 would continue to create strong informal pressures for employers to be exquisitely race-conscious and to adopt vigorous affirmative action plans. The act provides that employers can be found guilty of employment discrimination, even when they don't intend to discriminate, whenever their employment practices have a "disparate impact" on minority applicants. To avoid being held liable under a "disparate impact" theory, employers have a strong incentive to reject color-blind hiring strategies, and to take refuge in the safe harbors of racial proportionalism.
If Republicans are committed to color-blindness, therefore, they will have to take the debate to an entirely new level. The bill endorsed by Senator Dole would merely prohibit the federal government from discriminating in favor or against anyone on the basis of race, but would leave private affirmative action undisturbed. A principled commitment to color-blindness, by contrast, would require the Republicans to repeal, or radically refine, the Civil Rights Act of 1991, an act that four years ago many of them supported. Waving the banner of President Bush, the Republicans could try to resurrect the original meaning of discrimination: disparate treatment rather than disparate impact.
This is a battle that Democrats should relish. Flush with the glow of having ended federal classifications by race, Clinton could try to paint the Republicans as the party of resegregation. In fact, as Richard Epstein has argued in Forbidden Grounds, individual minorities might fare best if all civil rights laws that regulate behavior in private employment markets were repealed. This would leave the handful of unreconstructed racists free to marginalize themselves, and everyone else free to engage in even more aggressive preferential treatment than current law allows. But Republicans, having boxed themselves into a color-blind corner, do not have the luxury of rejecting the non-discrimination principle entirely.
Perhaps the endless distinctions this sort of middle ground would require are too legalistic to be politically compelling: federal set-asides and the executive order mandating goals and timetables should be repealed, but the Civil Rights Act of 1991 should be retained; the private sphere would look more like America than the public sphere, and so forth. Perhaps, in the next year or so, the entire apparatus of preferences and set-asides will be swept away by judicial or by congressional fiat; and the color-blind principle will be the law of the land for the first time since it was proposed and rejected during Reconstruction. But the search for a middle ground is worth the effort. No laws can save us from our obsessive race-consciousness, which is why it may be more honest simply to take government out of the business of requiring or forbidding racial preferences, except at the margins. Each sphere of society will then be free to sort out its conflicted impulses, with all deliberate speed.
Jeffrey Rosen is legal affairs editor at The New Republic and president and CEO of the National Constitution Center.