Discrimination and Disparities

by Thomas Sowell

  Similar gross disparities have also been found between the number of species of fish in the Amazon region of South America, compared to the number in Europe: “Eight times as many species of fish have been caught in an Amazonian pond the size of a tennis court as exist in all the rivers of Europe.”55

  IMPLICATIONS

  What can we conclude from all these examples of highly skewed distributions of outcomes around the world? Neither in nature nor among human beings are either equal or randomly distributed outcomes automatic. On the contrary, grossly unequal distributions of outcomes are common, both in nature and among people, in circumstances where neither genes nor discrimination are involved.

  What seems a more tenable conclusion is that, as economic historian David S. Landes put it, “The world has never been a level playing field.”56 The idea that it would be a level playing field, if it were not for either genes or discrimination, is a preconception in defiance of both logic and facts. Nothing is easier to find than sins among human beings, but to automatically make those sins the sole, or even primary, cause of different outcomes among different peoples is to ignore many other reasons for those disparities.

  Geographic differences are one among other factors that make for a skewed distribution of outcomes. Coastal peoples have long tended to be more prosperous and more advanced than people of the same race living farther inland, while people living in river valleys have likewise tended to be more prosperous and more advanced than people living up in the mountains.57

  Most of the most fertile land in the world is in the temperate zones and little or none in the tropics.58 Areas that are both near the sea and in the temperate zones have 8 percent of the world’s inhabited land area, 23 percent of the world’s population, and 53 percent of the world’s Gross Domestic Product.59

  Neither genetics nor discrimination is either necessary or sufficient to account for such skewed outcomes. This does not mean that either genes or discrimination can simply be dismissed as a possibility in any given circumstance, but only that hard evidence would be required to substantiate either of these possibilities, which remain testable hypotheses, without being foregone conclusions.

  Given how widely, how long and how strongly each of these two explanations—that is, genes or discrimination—has dominated thinking, laws and policies in various parts of the world, it is no small matter to escape from having painted ourselves into a corner with either of these sweeping preconceptions.

  Two of the monumental catastrophes of the twentieth century—Nazism and Communism—led to the slaughter of millions of human beings, in the name of either ridding the world of the burden of “inferior” races or ridding the world of “exploiters” responsible for the poverty of the exploited. While each of these beliefs might have been testable hypotheses, their greatest political triumphs came as dogmas placed beyond the reach of evidence or logic.

  Neither Hitler’s Mein Kampf nor Marx’s Capital was an exercise in hypothesis testing. While Karl Marx’s vast three-volume economic treatise was a far greater intellectual achievement, “exploitation” was at no point in its 2,500 pages treated as a testable hypothesis, but was instead the foundation assumption on which an elaborate intellectual superstructure was built. And that proved to be a foundation of quicksand. Getting rid of capitalist “exploiters” in Communist countries did not raise the living standards of workers, even to levels common in many capitalist countries, where workers were presumably still being exploited, as Marxists conceived the term.

  Discrimination as an explanation of economic and social disparities may have a similar emotional appeal for many. But we can at least try to treat these, and alternative theories, as testable hypotheses. The historic consequences of treating beliefs as sacred dogmas beyond the reach of evidence or logic should be enough to dissuade us from going down that road again, despite how exciting or emotionally satisfying political dogmas and the crusades resulting from those dogmas can be, or how convenient in sparing us the drudgery and discomfort of having to think through our own beliefs or test them against facts.

  * ⅔ x ⅔ x ⅔ x ⅔ x ⅔ = 32/243

  * As of 1940, just under half of the women in the Terman group were employed full time. Lewis M. Terman, et al., The Gifted Child Grows Up: Twenty-Five Years’ Follow-Up of a Superior Group (Stanford: Stanford University Press, 1947), p. 177.

  * Distinguished economist Richard Rosett was another example. See Thomas Sowell, The Einstein Syndrome: Bright Children Who Talk Late (New York: Basic Books, 2001), pp. 47–48. The best-selling author of Hillbilly Elegy was another. See J.D. Vance, Hillbilly Elegy: A Memoir of a Family and Culture in Crisis (New York: HarperCollins, 2016) pp. 2, 129–130, 205, 239.

  * More than half a century before the collapse of Eastman Kodak, economist J.A. Schumpeter pointed out that the most powerful economic competition is not that between producers of the same product, as so often assumed, but the competition between old and new technologies and methods of organization. In the case of Eastman Kodak, it was not the competition of Fuji film, but the competition of digital cameras, that was decisive. For Schumpeter, it was not the competition of firms producing the same products, as in economics textbooks, that was decisive. In Schumpeter’s words, “it is not that kind of competition which counts but the competition from the new commodity, the new technology, the new source of supply, the new type of organization (the largest-scale unit of control, for instance)—competition which commands a decisive cost or quality advantage and which strikes not at the margins of the profits and the outputs of the existing firms but at their foundations and their very lives.” Joseph A. Schumpeter, Capitalism, Socialism, and Democracy, third edition (New York: Harper & Brothers, 1950), p. 84.

  Chapter 2

  DISCRIMINATION: MEANINGS and COSTS

  Some people are said to have discriminating tastes when they are especially discerning in detecting differences in qualities and choosing accordingly, whether choosing wines, paintings or other goods and services. But the word is also used in an almost opposite sense to refer to arbitrary differences in behavior toward people, based on their group identities, regardless of their actual qualities as individuals.

  Both kinds of discrimination can result in large differences in outcomes, whether judging people or things. Wine connoisseurs can end up choosing one kind of wine far more often than another, and paying far more for a bottle of one kind of wine than for a bottle of the other.

  Something similar can often be observed when it comes to people. It is common, in countries around the world, for some groups to have very different outcomes when they are judged by others in employment, educational and other contexts. Thus different groups may end up with very different incomes, occupations and unemployment rates, or very different rates of admissions to colleges and universities, among many other group disparities in outcomes.

  The fundamental question is: Which kind of discrimination has led to such disparate outcomes? Have differences in qualities between individuals or groups been correctly discerned by others or have those others made their decisions based on personal aversions or arbitrary assumptions about members of particular groups? This is ultimately an empirical question, even 20 if attempts to answer that question evoke passionate feelings and passionate certainty by observers reaching opposite conclusions.

  Another way of saying the same thing is: Are group disparities in outcomes a result of internal differences in behavior and capabilities, accurately assessed by outsiders, or are those disparities due to external impositions based on the biased misjudgments or antagonisms of outsiders?

  The answers to such questions are not necessarily the same for all group disparities, nor necessarily the same for a given group at different times and places. Seeking the answers to such questions is more than an academic exercise, when the ultimate purpose is to enable more fellow human beings to have better prospects of advancing themselves. But, before seeking answers, we need to be very clear about the words we use in asking th
e question.

  MEANINGS OF DISCRIMINATION

  At a minimum, we need to know what we ourselves mean when we use a word like “discrimination,” especially since it has conflicting meanings. The broader meaning—an ability to discern differences in the qualities of people and things, and choosing accordingly—can be called Discrimination I, making fact-based distinctions. The narrower, but more commonly used, meaning—treating people negatively, based on arbitrary assumptions or aversions concerning individuals of a particular race or sex, for example—can be called Discrimination II, the kind of discrimination that has led to anti-discrimination laws and policies.

  Ideally, Discrimination I, applied to people, would mean judging each person as an individual, regardless of what group that person is part of. But here, as in other contexts, the ideal is seldom found among human beings in the real world, even among people who espouse that ideal.

  If you are walking at night down a lonely street, and see up ahead a shadowy figure in an alley, do you judge that person as an individual or do you cross the street and pass on the other side? The shadowy figure in the alley could turn out to be a kindly neighbor, out walking his dog. But, when making such decisions, a mistake on your part could be costly, up to and including costing you your life.

  In other contexts, you may in fact judge each person as an individual. But that this depends on context means that people have already been implicitly pre-sorted by the context, and only after that pre-sorting are they then judged as individuals. For example, a professor entering a classroom on the first day of the academic year may judge and treat each student as an individual. But that same professor, walking down a lonely street at night, may not judge and react to each stranger on the road ahead as an individual.

  The students in a college classroom are not likely to be a random sample of the full range of variations found in the general population, and are more likely to represent a narrower range of people assembled there for a narrower range of purposes, and with a narrower range of individual characteristics, as well as being in a setting less dangerous than a dark street at night.

  In short, one of the differences between the applicability of Discrimination I and Discrimination II is cost—and this is not always a small cost, nor a cost measured solely in money. Everyone might agree that Discrimination I is preferable, other things being equal, because it means making decisions based on demonstrable realities. Nevertheless, one may still be aware that other things are not always equal, and sometimes other things are very far from being equal.

  Where there is a difference in costs when choosing between Discrimination I and Discrimination II, much may depend on how high those costs are, and especially on who pays those costs. People who would never walk through a particular neighborhood at night, or perhaps not even in broad daylight, may nevertheless be indignant at banks that engage in “redlining”—that is, putting a whole neighborhood off-limits as a place to invest their depositors’ money. The observers’ own “redlining” in their choices of where to walk may never be seen by them as a different example of the same principle.

  In short, Discrimination I can have prohibitive costs in some situations, especially when it is applied at the individual level. However, Discrimination II—the arbitrary or antipathy-based bias against a group, is not the only other option. Another way of making decisions is by weighing empirical evidence about groups as a whole, or about the interactions of different groups with one another.

  This is still Discrimination I, basing decisions on empirical evidence. But the distinction between the ideal version of Discrimination I—judging each individual as an individual—and making decisions based on empirical evidence about the group to which the individual belongs is a consequential difference. We can call the ideal version (basing decisions on evidence about individuals) Discrimination Ia, and the less than ideal version (basing individual decisions on group evidence) Discrimination Ib. But both are different from Discrimination II, making decisions based on unsubstantiated notions or animosities.

  To take an extreme example of Discrimination Ib, for the sake of illustration, if 40 percent of the people in Group X are alcoholics and 1 percent of the people in Group Y are alcoholics, an employer may well prefer to hire only people from Group Y for work where an alcoholic would be not only ineffective but dangerous. This would mean that a majority of the people in Group X—60 percent in this case—would be denied employment, even though they are not alcoholics.

  What matters, crucially, to the employer is the cost of determining which individual is or is not an alcoholic, when job applicants all show up sober on the day when they are seeking employment.

  This also matters to the customers who buy the employer’s products and to society as a whole. If alcoholics produce a higher proportion of products that turn out to be defective, that is a cost to customers, and that cost may take different forms. For example, the customer could buy the product and then discover that it is defective. Alternatively, defects in the product might be discovered at the factory and discarded. In this case, the customers will be charged higher prices for the products that are sold, since the costs of defective products that are discovered and discarded at the factory must be covered by the prices charged for the reliable products that pass the screening test and are sold.

  To the extent that alcoholics are not only less competent but dangerous, the costs of those dangers are paid by either fellow employees who face those dangers on the job or by customers who buy dangerously defective products, or both. In short, there are serious costs inherent in the situation, so that either 60 percent of the people in Group X or employers or customers—or all three groups—end up paying the costs of the alcoholism of 40 percent of the people in Group X.

  This is certainly not judging each job applicant as an individual, so it is not Discrimination I in the purest sense of Discrimination Ia. On the other hand, it is also not Discrimination II, in the sense of decisions based on a personal bias or antipathy toward that group. The employer might well have personal friends from Group X, based on far more knowledge of those particular individuals than it is possible to get about job applicants, without prohibitive costs.

  The point here is neither to justify nor condemn the employer but to classify different decision-making processes, so that their implications and consequences can be analyzed separately. If judging each person as an individual is Discrimination Ia, we can classify as Discrimination Ib basing decisions about groups on information that is correct for that group, though not necessarily correct for every individual in that group, nor necessarily even correct for a majority of the individuals in that group.

  A real-life example of the effect of the cost of knowledge in this context is a study which showed that, despite the reluctance of many employers to hire young black males, because a significant proportion of them have criminal records (Discrimination Ib), those particular employers who automatically did criminal background checks on all their employees (Discrimination Ia) tended to hire more young black males than did other employers.1

  In other words, where the nature of the work made criminal background checks worth the cost for all employees, it was no longer necessary to use group information to assess whether individual young black job applicants had a criminal background. This made young black job applicants without a criminal background more employable than before.

  More is involved here than simply a question of nomenclature. It has implications for practical policies in the real world. Many observers, hoping to help young black males have more employment opportunities, have advocated prohibiting employers from asking job applicants questions about a criminal record. Moreover, the U.S. Equal Employment Opportunity Commission has sued employers who do criminal background checks on job applicants, on grounds that this was racial discrimination, even when it was applied to all job applicants, regardless of race.2 Empirically, however, criminal background checks provided more employment opportunities for young black males.


  In a very different situation, even employers who have no animosity or aversions against particular groups may nevertheless engage in Discrimination Ib—empirically based generalizations—when the employer knows that various groups react differently in the presence of some other group or groups.

  Back in nineteenth-century America, for example, when there were many immigrants from Europe in the workforce, some groups brought their mutual antagonisms in Europe with them to America. To have a workforce including both Irish Protestants and Irish Catholics working together at that time was to risk distracting frictions and even violence, with negative effects on productivity. In other words, a workforce consisting exclusively of either group might be more efficient than a workforce consisting of both.

  The same principle applies where different groups have especially positive reactions to one another. For example, the employer may be indifferent as to whether the work to be done is done by men or by women, and yet be well aware that men and women are not indifferent to each other, or else the human race would have become extinct long ago.

  Therefore, in the interests of workforce efficiency, when a particular occupation is overwhelmingly chosen by women, such as nursing, the employer may be reluctant to hire a male nurse, regardless of that male nurse’s individual qualifications. Conversely, where lumberjacks are overwhelmingly male, the employer may be reluctant to hire a female lumberjack, even if she is demonstrably as fully qualified as the men.

  Observers who point out that particular individuals are equally qualified, regardless of sex, miss the point. An equally qualified individual may do the work just as well as others, but if some of the others are distracted from their work, the net effect can be a less efficient workforce. That is the empirical basis that can lead employers to practice Discrimination Ib in such situations, even if the employers have no bias or aversion to those less likely to be hired.