August 1996 € Volume 23 € Number 4

Color-Blind Policies Are Not Enough

A Case in Favor of Affirmative Action

By Jonathan Caulkins

This essay argues that affirmative action is needed. It assumes that one valid goal of public policy is to enhance equality of opportunity for all citizens. The principal claim is that "color blind" policies cannot be expected to produce equal opportunity. Together the assumption and claim suggest that affirmative or "active" policies deserve consideration.

I do not argue that all policies pursued under a banner of affirmative action meet the criteria for affirmative action as laid out, for example, in the Affirmative Action Review commissioned by President Clinton. Nor do I claim that every policy which meets those criteria will enhance equality of opportunity. Rather, I argue that in the absence of any such affirmative policies, traditionally disadvantaged groups will not enjoy equal opportunity.

The method of argument is to make that case in four contexts, or models of society, that seem a priori to be less and less likely to require affirmative action. To the extent that those models strike the reader as plausible abstractions and to the extent that I succeed in arguing that color-blind policies would be inadequate in those contexts, then I will have proved my claim and supported the conclusion that affirmative action is needed.

Note: I discuss affirmative action as if the traditionally empowered group were always whites and the only disadvantaged group were blacks. Obviously that is gross simplification of the multicultural nation in which we live. I do this for brevity and clarity, not because I think affirmative action is only an issue of blacks and whites, and trust that the reader can generalize the argument to other relevant groups.

Definition of Affirmative Action
Before proceeding, it is important to clarify what affirmative action is and what it is not, or should not be. Extreme, sometimes egregious, examples of affirmative action gone awry make the most dramatic news stories and are the most likely to be challenged in court. I suspect many people believe these extremes to be the norm, both in the sense of being normal and in the sense of being the normative policy goal.

I neither can nor wish to defend such extremes, and focus instead on what I take as the current, official, statement on affirmative action, the Report to the President, whose introduction states:
"The goal of any affirmative action program must be to promote equal opportunity. Affirmative action measures recognize that existing patterns of discrimination, disadvantage, and exclusion may require race- or gender-conscious measures to achieve that equality of opportunity. [Yet] affirmative action cannot supersede the concept of merit. [No] unqualified person can be preferred over another qualified person in the name of affirmative action."

The document lists two tests that an affirmative action program must pass. First, does it work? Second, is the program fair in the sense that: (1) it does not use quotas, (2) race or gender-neutral options have been analyzed and rejected as inadequate, (3) it is flexible in the sense that less extensive and intrusive alternatives were analyzed and rejected, (4) the measure is limited in duration and reviewed periodically, and (5) the effect on nonbeneficiaries is sufficiently small and diffuse so as not to burden unduly their opportunity.

Model No. 1: Whites are Racist

Suppose that whites are racist in the sense that they will discriminate in favor of whites over blacks, even if it means hiring a less qualified worker, enrolling a less qualified student, etc. Unchecked, this would tautologically lead to unequal opportunity.

One might suggest that affirmative action is not needed; the government could simply follow a color-blind policy. However, in a democratic, free-market society it is not enough for the government alone to eschew such racism. The United States has a population of 260 million and, hence, has 260 million decision makers, any one of whom can thwart equality of opportunity.

Of course discrimination can and has been outlawed, but such laws are unlikely to erase all traces of racism for at least three reasons. First, our legal system follows a decision rule, "innocent until proven guilty," which is biased toward inaction. The inability to prove something beyond a reasonable doubt does not prove that action did not occur.

Second, eliminating all discriminatory actions would not eliminate discrimination because discrimination often manifests in the absence of an action, e.g., not hiring a qualified minority because he or she is a minority. It is generally harder to use the courts to punish lack of action.
Third, discrimination involves diffuse defendants and victims. Being rejected by one prospective employer is hardly grounds for a law suit; a pattern of repeated rejections is required. Suppose someone is rejected by 50 different employers. Which one should be sued? Also, it may not be in the interest of any one individual to file such a suit. Court costs are high, the aggrieved individuals may have limited resources, and even if the plaintiff "wins," gaining the right to work for someone you just sued could be a Pyrrhic victory.

Although I would not expect anti-discrimination laws alone to create equal opportunity, much more aggressive enforcement of these laws might represent a useful middle ground for parties on either side of the affirmative action debate. Few would oppose punishing actions that are illegal as well as immoral, and the threat of such sanction from a high profile campaign might make racist individuals cautious about acting on their prejudice.

Model No. 2: Whites and Blacks are Equally Racist

One might counter that Model No. 1 is inaccurate because it assumes blacks never discriminate against whites. If equal percentages of blacks and whites discriminated against the other group, would equal opportunity prevail without affirmative action? Not necessarily. Suppose only whites were in a position to make decisions about who to hire, who to admit to college, and who to promote. Then only blacks, not whites, would be denied opportunities because of discrimination. Clearly whites do not hold all positions of power. Equally clearly, by a wide range of measures, blacks do not hold their proportionate share.

In the political sphere, there has never been a black president or vice president of the United States. Only one sitting Senator is black. Only 2.3 percent of local elected officials are black, well below blacks' percentage (12.6 percent) of the population as whole.

Wealth and professional status confer power in our society, and only 1.5 percent of black families are among the wealthiest 5 percent of American families. The 1993 median black family income ($21,542) is just 55 percent that of white families ($39,300). Blacks are under-represented in better paying professions such as being a lawyer (3.3 percent), physician (4.2 percent) or engineer (3.7 percent).

It has been said that "knowledge is power." Fewer blacks than whites complete four or more years of high school (72.9 percent vs. 82.0 percent), and fewer complete four or more years of college (12.9 percent vs. 22.9 percent). Blacks earned 4.5 percent of all doctorates conferred in 1993. [The statistics cited in this section are taken from the Statistical Abstract of the United States, 1995.]
Even if whites and blacks had equally discriminatory intent, whites would have more chances to act on their prejudice, so blacks would not enjoy equal opportunity.

Model No. 3: Whites and Blacks are Equally Racist and Equally Powerful

Even if whites and blacks were equally prone to discriminate and had similar chances to exercise those discriminatory feelings, blacks still would not necessarily have equal opportunity simply by virtue of their being a minority in American society.

To see why, consider two stylistic examples. First, suppose that 10 percent of each race would discriminate against a member of the other race if given the chance, everyone has eight neighbors, and neighborhoods are racially integrated. Then blacks are more than five times more likely to suffer from a neighbor's discrimination than whites.

The probability that any one neighbor of a white person discriminates is the probability they are black (12.6 percent) times the probability the person is racist (assumed to be 10 percent in this example) or 1.26 percent. Hence, the probability that at least one of a white person's eight neighbors discriminates against him or her is (1 - P{a neighbor doesn't discriminate}8) = 1 - 0.98748 = 9.6 percent.

For a black person, any given neighbor is much more likely to be of the other race (87.4 percent, not 12.6 percent) and hence to discriminate (8.74 percent, not 1.26 percent), so the probability that a black suffers discrimination from at least one neighbor is (1 - (1 - 8.74%)8) = 51.9 percent.
If equal proportions of whites and blacks would discriminate against a member of the other group, then in a mixed society, blacks are much more likely to experience discrimination than is a member of the majority group.

As prelude to the second example, note that in 1991, 13.5 million of the 125 million workers in the United States (roughly 1 in 9) were black. Imagine a situation in which nine equally qualified individuals, eight white and one black, each applied for nine jobs about whose relative desirability everyone agrees. That is, all nine would produce the same rank order from most (No.1) to least (No. 9) desirable. Consider how the expected desirability of one's job varies for whites and blacks under different regimes.

Consider initially a regime in which blacks are discriminated against. The black applicant would get the least desirable position (No. 9). The eight white applicants are, by assumption, indistinguishable. Collectively they'll take the eight better jobs. Who gets exactly which job would be determined randomly, so the expected job quality for each white is 4.5, on a scale where 1 is best and 9 is worst.

Consider next a color-blind regime. Since all applicants are equally qualified, they all have equal likelihood's of getting any job, and a priori the expected quality rating of the job each gets is a 5.
Finally consider an aggressive affirmative action regime which favors blacks. The black applicant would get the most desirable job (No. 1); the expected rating of the job received by each white candidate drops from 5 to 5.5.

Note that the presence or absence of discrimination, either traditional discrimination or affirmative action, makes a much greater difference for a member of the minority than for someone in the majority. The reduction in opportunity an individual white experiences moving from a color blind regime to aggressive affirmative action is much less than the reduction in opportunity experienced by a minority individual moving from a color blind society to one that discriminates against members of his or her group.

Model No. 4: No One is Racist

Suppose no one is racist in the sense that everyone simply acts rationally in his or her own self-interest. People hire or recruit whoever they expect to produce the greatest benefit to them and their organization. No other factors enter into their decision. Affirmative action may be needed even in this idealized meritocratic state.

Race is an easily observed characteristic that is often correlated with more difficult to observe and job-relevant characteristics, making it "rational" to discriminate statistically. Such "rational" discrimination can reduce opportunity every bit as much as bitter, hate-based discrimination.
Furthermore, this is not necessarily a transitory problem. Inequality generated out of hatred can be perpetuated in a world populated by "rational" actors who bear no particular grudge against members of another group. For example, suppose knowledge of the reduced opportunity implied by statistical discrimination reduces the incentive for members of the minority group to invest in their human capital, thereby perpetuating the correlation between the observable trait (skin color) and the more difficult to assess job relevant characteristics. Even in a world populated by rational people who harbor no hatred, inequality can be a trapping state in the absence of race-conscious intervention.

Impact of Affirmative Action on Whites
The discussion above focused on opportunity for blacks. However, even if affirmative action improved opportunity for one group, if it simultaneously substantially reduced opportunity for another group, it might not increase equality of opportunity. So, it is important to ask how affirmative action policies affect opportunities available to whites. The answer depends on how narrowly or broadly one defines opportunity.

If one focuses on a particular hiring decision, no policy could be more equal than choosing the most qualified applicant and breaking ties randomly. Viewed in that narrow sense, any policy which explicitly recognizes traits not directly relevant to job performance would reduce equality of opportunity.

Suppose one defined opportunity more broadly as opportunity for employment in some job, not just one particular job. If all other jobs were governed by an ideal color-blind policy, then moving a given hiring decision from color blind to affirmative action would reduce equality of opportunity. But if even a small fraction of the other hiring decisions are made in ways that discriminate against blacks, then affirmative action could enhance equality of opportunity for obtaining some job, even as it reduces equality of opportunity for the particular job in question.

Finally, suppose one took seriously the argument that the prospect of unequal opportunity discouraged minorities from investing in their human capital. This would lead minorities to be underemployed in the sense of not participating in jobs which are as knowledge or skill intensive. If affirmative action encouraged minorities to stay in school longer and otherwise invest in their future employment prospects, then since knowledge and skill intensive jobs generate more value added to the economy than lower skill jobs, affirmative action might lead to more economic activity and increased employment opportunities for everyone.

The human capital accumulation argument is one reason why affirmative action might be a positive sum policy that makes every group better off, not a zero-sum redistributive issue. Michael Rothkopf offered another in a paper presented at the May INFORMS conference. It suggests that affirmative action in federal contract awards which subsidize minority applicants might, in plausible circumstances, save the government money and perhaps even lead to greater overall economic efficiency.

There is no definitive answer to the question of whether affirmative action will reduce opportunities for whites in a way that reduces equality of opportunity. One can draw different conclusions depending on whether one views opportunity narrowly (e.g., in terms of one specific job) or broadly (e.g., in terms of overall employment prospects) and on whether one considers indirect effects (such as feedback through overall economic growth).

Bottom Line
I will close with a John Rawls' style "test." Imagine you were about to be born, reincarnated if you will, and all that remains to be decided is whether you will be white or black. Your "endowment" of intelligence, courage, moral character, etc. have all been determined, as has the structure, income, and educational level of the family into which you will be born. Suppose you wanted to be president of the United States, to run a Fortune 500 firm, to win a Nobel Prize in physics or chemistry, or simply to not be murdered, otherwise victimized by violent crime, or feel the sting of a racial slur. Would you maximize your chance of achieving such goals by choosing to be white or by choosing to be black? I think one would have a better chance of achieving such goals being white.

The conclusion I draw from this "test" is that even with affirmative action blacks do not enjoy the same level opportunity as whites in the United States today. Presumably ending affirmative action would aggravate this disparity. If the goal is equal opportunity, color-blind policies are not enough. We need affirmative action.

Jonathan Caulkins is an assistant professor at the H. John Heinz III School of Public Policy and Management at Carnegie Mellon University.

E-mail to the Editorial Department of OR/MS Today: orms@lionhrtpub.com

Lionheart Publishing, Inc.
2555 Cumberland Parkway, Suite 299, Atlanta, GA 30339 USA
Phone: 770-431-0867 | Fax: 770-432-6969
E-mail: lpi@lionhrtpub.com