Affirmative action is a classic example of the administrative state’s pathology. Congress in the Great Society era outlawed “discrimination based on race.” Rather than define “discrimination,” however, it delegated this lawmaking power to an independent agency, the Equal Employment Opportunity Commission (EEOC). Its bureaucrats’ definition of discrimination was far more sweeping than what most congressmen thought they had voted to outlaw, and in important respects antithetical to the standard of a color-blind Constitution invoked so often in the debates over the 1964 Civil Rights Act. The Supreme Court, however, deferred completely to the regulators’ understanding of “discrimination,” ignoring the original intent of Congress.

Another stream of affirmative action arose from executive orders that extend back to World War II. (Indeed, it was President John F. Kennedy’s executive order that introduced the phrase “affirmative action” into civil rights policy.) These orders required any federal contractor to devise nondiscrimination plans, which morphed into statistical “goals and timetables”—quotas, in all but name. By the 1980s these programs had become so entrenched that President Ronald Reagan, who could have acted on his expressed desire to abolish them “with the stroke of a pen,” refrained from doing so, largely at the behest of his Big Business supporters who preferred the predictability, and reduced exposure to lawsuits over personnel practices, that came with the quota regime.

* * *

June and David O’Neill’s book, The Declining Importance of Race and Gender in the Labor Market, does two useful things. It describes the operation of the affirmative action regime, and it accounts for the persistence of racial and gender disparities in the labor market. The affirmative action regime affects other areas (e.g., education and voting) and applies to other groups, such as the disabled. But the O’Neills, economists who have worked extensively in both the academic and government worlds, provide indispensable clarity by focusing on race and gender as they relate to employment. June O’Neill was the director of the Congressional Budget Office in the late 1990s and is the director of the Center for the Study of Business and Government and the president of the Committee for a Responsible Federal Budget. David O’Neill has worked for the Census Bureau, New York Federal Reserve Bank, and the General Accounting Office. Both teach at Baruch College.

The authors argue persuasively that affirmative action is based upon gross but memorable statistical distortions about, for example, blacks earning only 60% as much as whites, or women 80% of what men are paid. In the prevailing “disparate impact” theory, discrimination encompasses far more than conscious acts of unequal treatment by prejudiced employers. Rather, discrimination exists whenever ostensibly race- or gender-neutral qualifications screen out more blacks than whites, more women than men. Disparate impact is a potent reminder that statism depends on its etymological cousin—statistics—because statistics reduce individual human persons to homogenous entities that administrators can shape to their liking.

The O’Neills provide an illuminating account of how the government-contract compliance program works. The acronyms of these agencies, originally presidential committees but now housed in the Department of Labor, are a wonderful sedimentary record of 20th-century bureaucracy. Franklin Roosevelt established the President’s Committee on Fair Employment Practice (commonly called the FEPC). Harry Truman followed with the President’s Committee on Government Contract Compliance (PCGGC). Dwight Eisenhower set up the President’s Committee on Government Contracts (PCGC—the missing “C” was said to indicate a profound change in policy). Kennedy gave us the President’s Committee on Equal Employment Opportunity (PCEEO). Lyndon Johnson launched the Office of Federal Contract Compliance (OFCC). It is now the Office of Federal Contract Compliance Programs (OFCCP).

* * *

Suffice it to say that OFCCP’s methods are highly capricious. Contractors face compliance review for reasons that appear to be, according to the O’Neills, “random” or “more nearly perverse,” thereby spawning consulting firms like the Equal Employment Advisory Council to help firms deal with OFCCP compliance reviews. The law itself offers little help since administrative law judges handle all disputes. “The importance of this restriction on the scope of judicial review is that the constitutional validity of any of OFCCP’s various procedures, such as affirmative action with goals and timetables, is never questioned.” OFCCP is a perfect example of what The Federalist called “the accumulation of all powers, legislative, executive, and judiciary, in the same hands.” Congress delegates legislative powers to it. Despite being in the Department of Labor, it escapes presidential control. The judiciary routinely accedes to it.

The Civil Rights Act of 1964 should be the cornerstone of our civil rights policy, having provided the element of consent missing from the evolving system of executive orders. Title VII of the Act dealt with employment discrimination, and also outlawed discrimination on the basis of “sex,” a category later added to the government contract system. Title VII certainly meant to avoid racial preferences and quotas. (Nothing could be clearer from the legislative history. Hubert Humphrey famously swore that if anyone could find “any language which provides that an employer will have to hire on the basis of percentage or quota related to color, race, religion, or national origin, I will start eating the pages one after another, because it is not in there.”) It also meant to avoid the administrative model of the New Deal, especially that of the National Labor Relations Board (NLRB). EEOC’s powers were initially so limited that liberals despaired over its inability to issue “cease and desist” orders, the main weapon wielded by NLRB. Instead, individual victims of discrimination could bring lawsuits if EEOC could not work out an amicable settlement with their employers. Republicans hoped that courts rather than bureaucrats would prevent a law meant to secure individual rights from turning into a system of racial group entitlements. It was a vain hope. By the 1970s the courts had bought into what EEOC affirmative action mastermind Alfred W. Blumrosen called the “law transmission system”: Congress delegates sweeping powers to enforce a vague standard, bureaucrats like himself make the rules, and courts enforce them.

* * *

The court approved of the EEOC’s “disparate impact” standard in 1971 in Griggs v. Duke Power Co. When the Supreme Court began to tighten up that standard in 1989, Congress responded by writing it into law in the 1991 Civil Rights Act. This provided the first explicit statutory sanction to affirmative action, though it’s doubtful the statute conforms to either the Constitution or rudimentary concepts of fairness and equal justice before the law.

So EEOC, like OFCCP, deals with disparate impact or statistical discrimination. The cynical abandonment of any concern for individual rights can be seen in EEOC’s “80%” rule. EEOC’s formulas divine how many women, blacks, Hispanics, etc., would be employed by a perfectly nondiscriminatory employer. If an employer’s employment practices meet 80% of those benchmarks, the commission will leave him alone. Individual justice does not matter so long as the statistical racial aggregates are close enough. We might call this the Four-fifths Compromise of affirmative action.

Accounting for current disparities in racial and gender outcomes requires rigorous statistical analysis, and sometimes the O’Neills cannot resist the social scientist’s temptation to “show all work.” Lay readers can, for example, ponder this formula, which measures the effect of OFCCP policies on the earnings of black workers: Wi = a + b1(C/T) + b2(Ai) + b3(Ei) + b4(R) + b5(I)

But the O’Neills’ careful statistical inquiries lead overwhelmingly to the same conclusion: almost none of the racial and gender disparities are the result of unequal treatment. Once you control for obvious variables such as age, level of education, regional economic and demographic differences, and years of work experience, the disparities between demographic subgroups become statistically insignificant. In gender terms, it is obvious that the choices that women make to leave the work force or reduce their hours for the sake of their children account for most of the gap between their compensation and men’s. And federal laws and policies have not done much to reduce unequal treatment—apart from the first few years after Title VII was enacted, especially in the South. Just as the Head Start program has failed, expensively, to give poor children head starts, our Equal Employment Opportunity apparatus has effected a net reduction in employment opportunities afforded equally and fairly.

* * *

But even if the O’Neills persuaded everyone that statistical disparities in the labor force cannot be ascribed to unequal treatment, few advocates of affirmative action would endorse a change in policy. One of the most important variables is performance on the Armed Forces Qualification Test, “an achievement test of verbal and mathematical skills [that] reflects the quality and quantity of schooling received (as well as the effects of parents’ education, since human capital is acquired in the home as well as in school).” Affirmative action advocates would respond that unequal results on the AFQT are themselves the result of unequal treatment in the past—that we are still dealing with the effects of slavery, segregation, and historical discrimination.

With regard to sex, the very fact that we call it “gender” gives the game away. “Gender” is not “sex,” a natural, biological difference. Thus we cannot say that women really “choose” to leave the work force or reduce their hours to care for their children. This apparent choice, like all other apparent sex differences, is really a coercive trick played on them by men over the centuries.

But even if true, to treat the causes of inequality—to correct and compensate for all the ways that historical legacies fall short of our moral standards—would be difficult and expensive. Likewise, to sort through individual cases of discrimination, to treat individuals as individuals, is difficult and expensive. This is why many employers have made their peace with affirmative action—it is cheaper and simpler than justice.

In the old days, too, employers engaged in racial discrimination based on statistics, assuming (often plausibly) that black workers were, on average, less efficient than white workers. Though they thus lost the services of those individual black workers who were more efficient than some white workers they hired, in the aggregate such discrimination made economic sense. The average overcomes the individual. Affirmative action simply inverts the incentives for statistical discrimination: employers must now take into account the costs of OFCCP and EEOC enforcement actions.

* * *

Although it is cheaper and easier to deal with the effects than with the causes of discrimination, the consequences of concentrating solely on the effects are perverse. Affirmative action is a palliative that only reinforces the causes of inequality. In the old days, they said that a black worker had to be twice as good as a white worker in order to get the same job. Now, they think (but can’t say that) he only has to be half as good. So why bother to be better?

The O’Neills make a similar point with regard to a new frontier of affirmative action, that of “comparable worth.” Congressional Democrats introduced the “Paycheck Fairness Act,” which would require firms to raise the wage of female-dominated jobs. The O’Neills rightly observe that this would establish a system of “politically administered wages,” which “would likely impede women’s progress. Women have been moving into nontraditional occupations such as medicine, law, and even engineering.” This outcome-equalizing program “would discourage women and men from investing in the skills to enter these higher paying fields.”

But “comparable worth” was so over-the-top that even the Ninth Circuit Court of Appeals rejected it in a 1985 decision written by Anthony Kennedy, who now sits on the Supreme Court. The Supreme Court this term addressed affirmative action in voting and education, but no serious challenge to affirmative action in employment is on the horizon. Perhaps June and David O’Neill’s impressive The Declining Importance of Race and Gender in the Labor Market will help to bring one about.