My scholarly work on the problem of race relations began with a general inquiry into the theory of economic inequality. Specifically, my 1981 paper, "Intergenerational Transfers and the Distribution of Earnings," which appeared in the journal Econometrica, introduced a model of economic achievement in which a person's earnings depended on a random endowment of innate ability and on skills acquired from formal training. The key feature of this theory was that individuals had to rely on their families to pay for their training. In this way, a person's economic opportunities were influenced by his inherited social position. I showed how, under these circumstances, the distribution of income in each generation could be determined by an examination of what had been obtained by the previous generation. My objective with the model was to illustrate how, in the long run, when people depend on resources available within families to finance their acquisition of skills, economic inequality comes to reflect the inherited advantages of birth. A disparity among persons in economic attainment would bear no necessary connection to differences in their innate abilities.

In other research, I applied this mode of reasoning to the problem of group, as distinct from individual, inequality. That analysis began with two observations. First, all societies exhibit significant social segmentation. People make choices about whom to befriend, whom to marry, where to live, to which schools to send their children, and so on. Factors like race, ethnicity, social class, and religious affiliation influence these choices of association. Second, the processes through which individuals develop their productive capacities are shaped by custom, convention, and social norms, and are not fully responsive to market forces, or reflective of the innate abilities of persons. Networks of social affiliation are not usually the result of calculated economic decisions. They nevertheless help determine how resources important to the development of the productive capacities of human beings are made available to individuals.

More concretely, one can say that an adult worker with a given degree of personal efficacy has been "produced" from the "inputs" of education, parenting skills, acculturation, nutrition, and socialization to which he was exposed in his formative years. While some of these "inputs" can be bought and sold, some of the most crucial "factors of production" are only available as by-products from activities of social affiliation. Parenting services are not to be had for purchase on the market, but accrue as the consequence of the social relations between the custodial parents and the child. The allocation of parenting services among a prospective generation of adults is thus the indirect consequence of social activities undertaken by members of the preceding generation. An adolescent's peer group is similarly a derivative consequence of processes of social networking.

I concede that this is an artificial way of thinking about human development, but the artifice is quite useful. For it calls attention to the critical role played by social and cultural resources in the production and reproduction of economic inequality. The relevance of such factors, as an empirical matter, is beyond doubt. The importance of networks, contacts, social background, family connections, and informal associations of all kinds has been amply documented by students of social stratification. In addition, values, attitudes, and beliefs of central import for the attainment of success in life are shaped by the cultural milieu in which a person develops. Whom one knows affects what one comes to know and, ultimately, what one can do with one's God-given talents.

Social capital and inequality
While all of this may seem obvious, the fact is that, prior to my work, formal theories of economic inequality had said little about the role of social background. I was the first economist to use the term "social capital" in reference to these processes by which the social relationships that occur among persons promote or retard their acquisition of traits valued in the market place. A large and growing literature has since emerged in which allowance is taken of the myriad ways that a person's opportunities to develop his natural gifts depend upon the economic achievements of those with whom he is socially affiliated. This literature suggests that unqualified confidence in the equity and efficiency of the income distribution produced by the market is not justified.

In particular, this analysis has an important ethical implication: Because the creation of a skilled work force is a social process, the meritocratic ideal should take into account that no one travels the road to economic and social success alone. The facts that generations overlap, that much of social life lies outside the reach of public regulation, and that prevailing social affiliations influence the development of the intellectual and personal skills of the young, imply that present patterns of inequality-among individuals and between groups-must embody, to some degree, social and economic disparities that have existed in the past. To the extent that past disparities are illegitimate, the propriety of the contemporary order is called into question.

I have employed this framework to explore the legitimacy question with respect to inequality between blacks and whites in America.1 (See my paper, "A Dynamic Theory of Racial Income Differences," in Women, Minorities and Emplyment Discrimination, ed. P.A. Wallace (Lexington Books, DC Heath, 1977.) In a theoretical example, I showed that, notwithstanding the establishment of a legal regime of equal opportunity, historically engendered economic differences between racial groups could well persist into the indefinite future. I concluded that the pronounced racial disparities to be observed in American cities are particularly problematic, since they are, at least in part, the product of an unjust history, propagated across the generations by the segmented social structures of our race-conscious society.

Thus I would argue, as a matter of social ethics, that the government should undertake policies to mitigate the economic marginality of those languishing in the ghettos of America. This is not a reparations argument. When the developmental prospects of an individual depend on the circumstances of those with whom he is socially affiliated, even a minimal commitment to equality of opportunity requires such a policy. In our divided society, and given our tragic past, this implies that public efforts intended to counter the effects of historical disadvantage among blacks are not only consistent with, but indeed are required by, widely embraced democratic ideals.

Color-blind extremists
This argument leads naturally to the question of whether affirmative-action policies are necessary and justified. To emphasize that racial group disparities can be transmitted across generations through subtle and complex social processes is not necessarily to endorse employment or educational preferences based on race. (I will offer in due course a number of reasons to think that these policies should be curtailed.) But recognizing the importance of social segmentation does cause one to doubt the ethical viability, and indeed the logical coherence, of "color-blind absolutism"-the notion that the Constitution requires government agents to ignore the racial identity of citizens. Ironically, recent claims by some conservatives to this effect bear an eerie resemblance, in form and in substance, to the similarly absolute claims of some card-carrying civil libertarians on behalf of a "wall of separation" between church and state.

Consider that, as a practical matter, the government cannot enforce laws against employment discrimination without taking note of a gross demographic imbalance in an employer's work force. Yet the government's requiring that employment data be reported by race is already a departure from pure color-blind behavior. So too is the practice, nearly universal in the public and private sectors, of targeted outreach efforts designed to increase the representation of blacks in the pool of persons considered for an employment opportunity. Accordingly, the more intellectually consistent of the color- blind absolutists now recommend, as logic would require, that we repeal the civil-rights laws and abandon even those efforts to achieve racial diversity which do not involve preferential treatment. But is that stance consistent with fairness?

More subtly, how can a college educator convey to students the lesson that "not all blacks think alike," with too few blacks on campus for this truth to become evident? Were an American president to assemble a cabinet devoid of racial minority representation, would not the legitimacy of his administration rightly be called into question? What prison warden could afford to ignore the possibility that racial friction among his inmates might threaten the maintenance of order within his institution? Perhaps this is why presidents, prison wardens, and college educators do not behave in a purely color-blind fashion in our divided society.

Coming up with cases that challenge the absolutist claim is not difficult. Can the police consider race when making undercover assignments? Can a black public employee use health insurance benefits to choose a black therapist with whom to discuss race-related anxieties? Can units in a public housing project be let with an eye to sustaining a racially integrated environment? What about a National Science Foundation effort that encourages gifted blacks to pursue careers in fields where few now study? Clearly, there is no general rule that can resolve all of these cases reasonably.

I would venture to say that the study of affirmative action has been too much the preserve of lawyers and philosophers, and has too little engaged the interests of economists and other social scientists. It is as if, for this policy, unlike all others, we could determine a priori the wisdom of its application-as if its practice were always either "right" or "wrong," never simply "prudent" or "unwise." However, although departures from color-blind absolutism are both legitimate and desirable in some circumstances, there are compelling reasons to question the wisdom of relying as heavily as we now do on racial preferences to bring about civic inclusion for black Americans.

Logical stereotyping
One such reason for questioning the wisdom of affirmative action is that the widespread use of preferences can logically be expected to erode the perception of black competence. This point is often misunderstood, so it is worth spelling out in some detail. The argument is not a speculation about the feelings of persons who may or may not be the beneficiaries of affirmative action. Rather, it turns on the rational, statistical inferences that neutral observers are entitled to make about the unknown qualifications of persons who may have been preferred, or rejected, in a selection process.

The main insight is not difficult to grasp. Let some employer use a lower threshold of assessed productivity for the hiring of blacks than whites. The preferential hiring policy defines three categories of individuals within each of the two racial groups which I will call "marginals," "successes," and "failures." Marginals are those whose hiring status is altered by the policy-either whites not hired who otherwise would have been, or blacks hired who otherwise would not have been. Successes are those who would be hired with or without the policy, and failures are those who would be passed over with or without the preferential policy. Let us consider how an outsider who can observe the hiring decision, but not the employer's productivity assessment, would estimate the productivity of those subject to this hiring process.

Notice that a lower hiring threshold for blacks causes the outside market to reduce its estimate of the productivity of black successes, since, on average, less is required to achieve that status. In addition, black failures, seen to have been passed over despite a lower hiring threshold, are thereby revealed as especially unproductive. On the other hand, a hiring process favoring blacks must enhance the reputations of white failures, as seen by outsiders, since they may have been artificially held back. And white successes, who are hired despite being disfavored in selection, have thereby been shown to be especially productive.

We have thus reached the result that, among blacks, only marginals gain from the establishment of a preferential hiring program-they do so because the outside observer lumps them together with their superiors, black successes. They thus gain a job and a better reputation than they objectively deserve. Moreover, among whites, only marginals are harmed by the program, for only they lose the chance of securing a job and only they see their reputations harmed by virtue of being placed in the same category as white failures. In practical terms, since marginals are typically a minority of all workers, the outside reputations of most blacks will be lowered, and that of most whites enhanced, by preferential hiring. The inferential logic that leads to this arresting conclusion is particularly insidious, in that it can serve to legitimate otherwise indefensible negative stereotypes about blacks.

A new model of affirmative action
Another reason for being skeptical about the practice of affirmative action is that it can undercut the incentives for blacks to develop their competitive abilities. For instance, preferential treatment can lead to the patronization of black workers and students. By "patronization," I mean the setting of a lower standard of expected accomplishment for blacks than for whites because of the belief that blacks are not as capable of meeting a higher, common standard. In the 1993 article "Will Affirmative Action Eliminate Negative Stereotypes?" that appeared in the American Economic Review, Stephen Coate and I show how behavior of this kind can be based on a self-fulfilling prophesy. That is, observed performance among blacks may be lower precisely because blacks are being patronized, a policy that is undertaken because of the need for an employer or admissions officer to meet affirmative-action guidelines.

Consider a workplace in which a supervisor operating under some affirmative- action guidelines must recommend subordinate workers for promotion. Suppose further that he is keen to promote blacks where possible, and that he monitors his subordinates' performance and bases his recommendations on these observations. Pressure to promote blacks might lead him to de-emphasize deficiencies in the performance of black subordinates, recommending them for promotion when he would not have done so for whites. But his behavior could undermine the ability of black workers to identify and correct their deficiencies. They are denied honest feedback from their supervisor on their performance and are encouraged to think that one can get ahead without attaining the same degree of proficiency as whites.

Alternatively, consider a population of students applying to professional schools for admissions. The schools, due to affirmative-action concerns, are eager to admit a certain percentage of blacks. They believe that to do so they must accept black applicants with test scores and grades below those of some whites whom they reject. If most schools follow this policy, the message sent out to black students is that the level of performance needed to gain admission is lower than that which white students know they must attain. If black and white students are, at least to some extent, responsive to these differing expectations, they might, as a result, achieve grades and test scores reflective of the expectation gap. In this way, the schools' belief that different admissions standards are necessary becomes a self-fulfilling prophecy.

The common theme in these two examples is that the desire to see greater black representation is pursued by using different criteria for the promotion or admission of black and white candidates. But the use of different criteria reduces the incentives that blacks have for developing needed skills. This argument does not presume that blacks are less capable than whites; it is based on the fact that an individual's need to make use of his abilities is undermined when that individual is patronized by the employer or the admissions committee.

This problem could be avoided if, instead of using different criteria of selection, the employers and schools in question sought to meet their desired level of black participation through a concerted effort to enhance performance, while maintaining common standards of evaluation. Call it "developmental," as opposed to "preferential," affirmative action. Such a targeted effort at performance enhancement among black employees or students is definitely not color-blind behavior. It presumes a direct concern about racial inequality and involves allocating benefits to people on the basis of race. What distinguishes it from preferential hiring or admissions, though, is that it takes seriously the fact of differential performance and seeks to reverse it directly, rather than trying to hide from that fact by setting a different threshold of expectation for the performance of blacks.

For example, given that black students are far scarcer than white and Asian students in the fields of math and science, encouraging their entry into these areas without lowering standards-through summer workshops, support for curriculum development at historically black colleges, or the financing of research assistantships for promising graduate students-would be consistent with my distinction between "preferential" and "developmental" affirmative action. Also consistent would be the provision of management assistance to new black-owned businesses, which would then be expected to bid competitively for government contracts, or the provisional admission of black students to the state university, conditional on their raising their academic scores to competitive levels after a year or two of study at a local community college. The key is that the racially targeted assistance be short-lived and preparatory to the entry of its recipients into an arena of competition where they would be assessed in the same way as everyone else.

Racism and responsibility
Unfortunately, economists seem to be the only people persuaded by, or even interested in, this kind of technical argument about affirmative action. Therefore, I turn now, in my capacity as an intellectual and a citizen, to a range of moral and political considerations that may be of broader interest but that still point in the same direction. Begin with an obvious point: The plight of the inner-city underclass-the most intractable aspect of the racial inequality problem today-is not mitigated by affirmative-action policies. Defenders of racial preferences answer by claiming this was never the intent of such policies. But this only leads to my second point: The persistent demand for preferential treatment as necessary to black achievement amounts, over a period of time, to a concession of defeat by middle-class blacks in our struggle for civic equality.

The political discourse over affirmative action harbors a paradoxical subtext: Middle-class blacks seek equality of status with whites by calling attention to their own limited achievements, thereby establishing the need for preferential policies. At the same time, sympathetic white elites, by granting black demands, thereby acknowledge that, without their patronage, black penetration of the upper reaches of American society would be impossible. The paradox is that, although equality is the goal of the enterprise, this manifestly is not an exchange among equals, and it never can be.

Members of the black middle class who stress that, without some special dispensation, they cannot compete with whites are really flattering those whites, while exhibiting their own weakness. And whites who think that, because of societal wrongs, blacks are owed the benefit of the doubt about their qualifications are exercising a noblesse oblige available only to the powerful. This exchange between black weakness and white power has become a basic paradigm for "progressive" race relations in contemporary America. Blacks from privileged backgrounds now routinely engage in a kind of exhibitionism of non-achievement, mournfully citing the higher success rates of whites in one endeavor or another in order to gain leverage for their advocacy on behalf of preferential treatment. That Asians from more modest backgrounds often achieve higher rates of success is not mentioned. But the limited ability of these more fortunate blacks to make inroads on their own can hardly go unnoticed.

It is morally unjustified-and to this African American, humiliating-that preferential treatment based on race should become institutionalized for those of us now enjoying all of the advantages of middle-class life. The thought that my sons would come to see themselves as presumptively disadvantaged because of their race is unbearable to me. They are, in fact, among the richest young people of African descent anywhere on the globe. There is no achievement to which they cannot legitimately aspire. Whatever degree of success they attain in life, the fact that some of their ancestors were slaves and others faced outrageous bigotry will have little to do with it.

Indeed, those ancestors, with only a fraction of the opportunity, and with much of the power structure of the society arrayed against them, managed to educate their children, acquire land, found communal institutions, and mount a successful struggle for equal rights. The generation coming of age during the 1960s, now ensconced in the burgeoning black middle-class, enjoy their status primarily because their parents and grandparents faithfully discharged their responsibilities. The benefits of affirmative action, whatever they may have been, pale in comparison to this inheritance.

My grandparents, with their siblings and cousins, left rural Mississippi for Chicago's mean streets in the years after World War I. Facing incredible racial hostility, they nevertheless carved out a place for their children, who went on to acquire property and gain a toe-hold in the professions. For most middle-class blacks this is a familiar story. Our forebears, from slavery onward, performed magnificently under harsh circumstances. It is time now that we and our children begin to do the same. It desecrates the memory of our enslaved ancestors to assert that, with our far greater freedoms, we middle-class blacks should now look to whites, of whatever political persuasion, to ensure that our dreams are realized.

The children of today's black middle class will live their lives in an era of equal opportunity. I recognize that merely by stating this simple fact I will enrage many people; and I do not mean to assert that racial discrimination has disappeared. But I insist that the historic barriers to black participation in the political, social, and economic life of the nation have been lowered dramatically over the past four decades, especially for the wealthiest 20 percent of the black population. Arguably, the time has now come for us to let go of the ready-made excuse that racism provides. And so too, it is time to accept responsibility for what we and our children do, and do not, achieve.