The online version of the American Heritage Dictionary of the English Language (2000) defines discrimination as, "Treatment or consideration based on class or category rather than individual merit; partiality or prejudice." Discrimination is a broad and multidimensional concept that covers all acts of preferring one thing, person, or situation over another (Block and Walker 1982, p. 6). In this broad sense, discriminatory behavior can occur within many economic or social activities of daily life. For example, the preference of a high school basketball coach for a taller player over a shorter one in selecting a team or an employer paying an African American worker less than a white worker for the same work would both fall under the heading of discrimination. While the latter act carries an unambiguously negative connotation, few people would consider the former act to be malevolent. Thus, discriminatory behavior does not always imply injustice or prejudice. While understanding this distinction is important, a more relevant discussion of discrimination should emphasize the types of discriminatory acts that are socially and economically unjust, the type of acts that have caused the word discrimination to gain an unambiguously negative meaning. Denying or restricting equal opportunity in housing, education, and employment to members of a certain demographic group, such as African Americans, women, or other minority groups, constitutes an act of discrimination that violates common notions of social and economic justice and points to a need for policy intervention.
LABOR-MARKET DISCRIMINATION
One of the most common forms of discrimination, labor-market discrimination refers to differential treatment of workers within the labor market. It occurs when individual workers with identical productivity characteristics are treated differently with respect to hiring, occupational access, promotion, wage rate, or working conditions because of the demographic groups to which they belong.
Taste for Discrimination Gary Becker, the 1992 Nobel Prize recipient in economics, laid the groundwork for the mainstream economic approach to the analysis of discrimination in The Economics of Discrimination (1957). Becker’s theory of discrimination represents an example of the neoclassical economics approach. He introduces the concept of taste for discrimination to translate the notion of discrimination into the language of economics. According to Becker:
If an individual has a "taste for discrimination," he must act as if he were willing to pay something, either directly or in the form of reduced income, to be associated with some persons instead of others. When actual discrimination occurs, he must, in fact, either pay or forfeit income for this privilege. This simple way of looking at the matter gets at the essence of prejudice and discrimination. (Becker 1957, p. 14)
Employer Discrimination In cases of employer discrimination, employers with a taste for discrimination act as if employing, for example, African American workers imposes psychological costs that they are willing to pay. The measure of their willingness to pay can be translated into monetary terms by the discrimination coefficient. To illustrate, suppose that the costs to an employer of employing an African American worker and a white worker are Waa and Ww , respectively. If the employer possesses a taste for discrimination against the African American worker, he will act as if the actual cost were W (1 + d), where d, a positive number, is the discrimination coefficient. The prejudiced employer will be indifferent when choosing between a white worker and an African American worker when the cost of hiring each worker is, to him, equal—that is, Ww = Waa (1 + d). A clear implication is that the African American worker will be hired by the discriminating employer only if his wage rate is below that of a white worker. More precisely, the African American worker will only be hired if his wage rate is less than that of a white worker by at least the amount of the discrimination coefficient.
Employee Discrimination The source of discrimination may also be the prejudice of fellow employees. For instance, white workers may possess discriminatory preferences against African American workers and avoid situations where they have to work alongside them. The extent of employee prejudice can be monetized by the discrimination coefficient, using an analogy parallel to employer discrimination. A white worker who is offered a wage W for a job will act as if this wage rate is only W (1 — d), where d is the white worker’s discrimination coefficient. The white worker will agree to work with African Americans only if he or she is paid a premium equal to W d.
Customer Discrimination Another source of discrimination in the labor market results from the prejudice of customers. For example, white customers may prefer to be served by white workers, which would reduce the demand for goods and services sold or served by African American workers. More formally, suppose the actual price of a good or a service is p. Then a white customer would act as if the price of this good or service were p(1 + d) when faced with an African American worker. One of the implications of customer discrimination is that it would result in a segregated workforce within a firm, with white workers being placed in positions with high customer contact and African Americans working in positions with minimal customer interaction.
Predictions of Becker’s Theory One of the predictions of Becker’s theory is that the labor market will become completely segregated over time. This prediction can be illustrated using employee discrimination. In the presence of employee discrimination, nondiscriminating and profit-maximizing employers would never choose to hire both white and African American workers because employers want to avoid paying a premium to white workers. Instead, they would hire only African American workers who offer the same productivity as whites at a lower wage.
Another prediction of Becker’s theory is that discrimination cannot persist in the long run in a competitive market where firms can enter and exit freely. This is because free entry by nondiscriminating employers will force discriminating employers out of the market. For example, nondiscriminating employers will hire equally productive African American workers at a wage that is lower than that offered to white workers. Nondis-criminating firms thus have a cost advantage over discriminating firms, and the forces of competition would drive the discriminating firms out of business. As more nondis-criminating firms enter the market, the demand for minority workers will rise, which will gradually erode the wage differentials between different groups of workers. Therefore, Becker’s theory suggests that policies aimed at lessening or eliminating barriers to competition in the market place should help reduce discrimination and wage differentials.
Economists have widely tested the predictions made by Becker’s model. For example, Orley Ashenfelter and Timothy Hannan (1986), using data from the banking sector, showed that the share of women employed in a firm is negatively related to the extent of competition in a geographical area. Another study by Judith Hellerstein, David Neumark, and Kenneth Troske (2002) found that, among plants with high levels of market power, those that employ more women are more profitable than those employing fewer women; no such relationship was found for plants with low levels of market power. These findings are consistent with Becker’s prediction that discrimination can exist in situations where firms possess market power.
Although Becker’s analysis of discrimination has found wide support among economists, criticism has been raised about the predictions of the theory. For example, it has been pointed out that, despite the model’s predictions, competitive market forces have not eliminated discrimination, and wage disparities between different demographic groups have not completely disappeared.Others have shown that the prediction of a segregated market does not accord with today’s real world; furthermore, more segregated workforces tend to generate more wage inequality, not less (Mason 1999). There is also evidence of skin-tone differences in wages among under-represented minorities (Mason 2004). Finally, several studies find that market competition does not decrease the degree of discrimination practiced within a sector, as predicted by Becker (Coleman 2004; Agesa and Hamilton 2004). Alternative theories of discrimination have been proposed to provide explanations for these issues.
Statistical Discrimination Discriminatory behavior can also occur because employers have limited information about the productivity characteristics of potential employees. Therefore, employers’ hiring decisions may rely on average group characteristics based on factors such as race and gender. As a result, individuals with identical productivity characteristics will have different labor-market outcomes because of the average quality of the group to which they belong. Judging individuals on the basis of their average group characteristics is referred to as statistical discrimination.
For example, suppose an employer has to choose between a male and female job applicant. Assume further that the observable personal characteristics of these two applicants, such as age, years of education, previous work experience, test scores, and recommendation letters, are identical and that both of them performed equally well at the job interview. An employer, having to make a hiring decision between the two applicants, may decide to offer the job to the male applicant based on the employer’s belief that female workers are more likely to quit their jobs than their male counterparts because women are likely to engage in childrearing. The employer makes a decision using statistics about the average group characteristics of the applicants. It is important to note that the statistical information used by the employer may or may not be accurate. In this example, whether or not women actually have higher quit rates than men is not relevant to the hiring outcome. While the behavior of some employers engaging in statistical discrimination could be rooted in prejudice, it is also possible that these actions are based purely on nonmalicious grounds. Statistical discrimination can result from decisions that may be correct, profitable, and rational on average.
Statistical discrimination helps explain how racial and gender differences between workers of equal productivity can exist in the labor market. It also explains how discrimination can persist over time. Unlike Becker’s taste-for-dis-crimination model, the employer does not suffer monetarily from practicing statistical discrimination. On the contrary, the discriminating employer can benefit from this behavior by minimizing hiring costs. Therefore, there is no compelling reason for discrimination, and wage differentials between males and females or African Americans and whites tend to disappear in the long run in the presence of statistical discrimination.
While much of the focus on statistical discrimination concerns the labor market, such practices can be observed in different sectors of society as well. One nonmarket example of statistical discrimination is the observed racial differentials in policing patterns. John Knowles, Nicola Persico, and Petra Todd (2001) found that police search vehicles driven by African American motorists for illegal drugs and other contraband far more frequently than those of white motorists. If the motive behind this police behavior is the belief that African Americans are more likely to commit the types of traffic violations that police use as pretexts for vehicle searches, this type of behavior is an example of statistical discrimination.
They assert that employers should realize that their beliefs are incorrect if average group differences are perceived but not real. If, on the other hand, these differences are real, then employers should develop methods to measure future performance accurately rather than engaging in discriminatory behavior, especially in a world with strict antidiscrimination laws.
NONCOMPETITIVE MODELS OF LABOR-MARKET DISCRIMINATION
The models of discrimination discussed above assume that firms operate in competitive markets. However, discriminatory motives can also be drawn from circumstances in which employers possess some degree of market power in wage determination. Alternative explanations of discrimination have been offered for these circumstances.
One of these alternative explanations is occupational crowding. The occupational crowding model of discrimination, advanced by Barbara Bergmann (1971), hypothesizes that labor-market disparities between African American workers and white workers (or males and females) are not due to a "taste for discrimination" by employers, but rather to a deliberate policy of segregating African American workers (or females) into lower-paid occupations. Crowding these groups into low-paying occupations and limiting their access to other occupations reduces their marginal productivity in comparison with that of white (or male) workers. At the same time, the exclusion of minorities from high-paying jobs pushes up the wages of whites, including those who might earn lower wages in the absence of discrimination.
However, profit-maximizing motives should induce some firms to start replacing higher-paid white (or male) workers with equally productive but cheaper African American (or female) workers. This process would eventually eliminate wage disparities between the two groups. In this case, the observed wage disparities can be explained by the presence of noncompetitive forces, such as barriers to worker mobility between occupations.
An alternative explanation of discrimination in the context of a noncompetitive market was developed by the British economist Joan Robinson (1933). Robinson argued that in a monopsonistic market (a labor market with a single employer), profits can be increased by discriminating against some workers if the labor supply elasticity (responsiveness to wages) of African American (or female) workers is less than that of white (or male) workers. Although this model offers a plausible explanation for monopsonistic markets, its applicability is limited because these types of markets are relatively rare.
Another explanation for discrimination was put forth by radical economist Michael Reich (1981). Reich criticized the neoclassical approach to discrimination on several grounds and offered an alternative explanation based on class conflict. According to Reich, a firm’s output and profitability depend not only on the amount of labor hired, but also on the level of collective action or bargaining power among workers within the firm. He further argued that bargaining power is a function of the wage and employment disparities between African Americans and other workers in the firm. Therefore, discriminatory practices, such as paying equally productive African American workers less than white workers or denying employment to African Americans, generates animosity in the work force, which reduces the bargaining power of workers. As a result of the workers’ weakened bargaining power, employers are able earn more profits. Unlike Becker’s model, a competitive employer is a direct beneficiary of discrimination in Reich’s model.
One prediction of this explanation is that discrimination cannot be eliminated through government interventions like affirmative action that redistribute occupational positions; it can only be rooted out by eliminating the social hierarchy, because it is the main source of ethnic conflict that results in discrimination.
Patrick Mason (1999) showed that if workers face differential conditions of labor supply, if employers are able to limit the power of labor coalitions, and if racial identity is an important factor in job competition among workers, then competitive profit-maximizing firms may persistently discriminate in their labor-market decisions.
MEASURING LABOR-MARKET DISCRIMINATION
Although it is relatively easy to detect discrimination, it is much more difficult to measure it. The difficulty arises from the fact that not all the disparity between, say, the wages of African Americans and whites is due to labor-market discrimination. These two groups may also differ in their productive characteristics because of pre—labor market discrimination (for example, African Americans receiving less and lower-quality education due to educational discrimination). Then, the disparity must be due in part to differences in the productivity characteristics between the groups. It is important to account for all of these productivity characteristics in order to obtain an unbiased estimate of the true measure of discrimination. After accounting for productivity characteristics, the residual in disparity would then represent a measure of discrimination.
CONCLUSION
Discrimination, whether it occurs in the form of denying employment, housing, or education to members of a particular group, places countless burdens on its victims and on society as a whole. These burdens are both economical and social. By provoking hostility between different groups, discrimination may undermine social harmony, leading to such undesirable social consequences as increased rates of crime. Discrimination also has adverse economic consequences because the earnings of those who face discrimination will be depressed and their career paths will suffer. As a result, the rate of poverty will increase among these groups. The negative consequences of poverty on education, health, crime, and the overall economy will further increase the costs of discrimination for society. Furthermore, many of these negative effects are likely to persist over generations. Thus, discrimination will not only place a cost on its current victims, it will also punish future generations, even if these future generations live in a society without discrimination.
Although the discussion above focuses on discrimination based on race or gender, discriminatory behaviors can and do target all minority groups, including the elderly, teenagers, the disabled, homosexuals, ethnic groups such as Hispanics, and members of certain religions such as Jews and Muslims.
