Chapter Study Outline

  • Although there are large observed differences in wages across groups, these differences are consistent with the hypothesis that:
    • [H0] Intrinsic Group Differences: The observed group differences result from the unfettered choices made by intrinsically different workers within a well-functioning nondiscriminatory labor market.
    • An attempt to quantify the severity of discrimination must apportion a proper weight to the hypothesis H0.
      • If observed intergroup differences are largely the result of voluntary choices, antidiscrimination policies will do more harm than good.

12.1 Measuring Discrimination

  • Earnings, y, are defined by y= wh, where w is the wage and h is the number of hours worked.
    • Discrimination may operate through two channels: wages or hours worked.
    • This discussion will assume a 40-hour workweek and focus only on the wage.
      • “Wage discrimination” and “earnings discrimination” are used interchangeably, since under this assumption y is directly proportional to w.
  • The raw wage gap, Δw̅, contains little information about market discrimination because it ignores underlying differences in material productivities.
    • Δ > 0 does not imply that there is discrimination, nor does Δ ≤ 0 imply that there is none.
    • It is necessary to compare the earnings of productively identical men and women, where the earnings of men are wm =α0m +α1mq and the earnings of women are wf =α0f +α1f q.
      • m = male
      • f = female
      • q = material productivity
      • α0i and α1i are parameters for each sex (I = f, m) that are to be estimated from the data using regression techniques.
      • Δw = Δα0 + Δα1 ∙ q is the difference from subtracting the equation for women’s earnings from the equation for men’s earnings, or the wage gap.
    • Wage discrimination against women arises if Δα0 ≥ 0 and Δα1 ≥ 0, where at least one of the inequalities is strict.
  • Oaxaca transformation: a procedure to compare the actual earnings of the average man with the hypothetical earnings of the average woman with the same productivity.
    • The Oaxaca transformation decomposes the raw wage gap Δ > MB as Δ = MB = AB + MA.
      • AB accounts for productivity differences.
      • MA is the residual capturing the extent of wage discrimination.
  • There are many obstacles to estimating the effects of discrimination, including self-selection bias, failure to capture the true relationship between wages and productivity, and the endogeneity of human capital.
    • Self-selection bias arises because people choose whether to work, and economists can only collect data on those in the workforce.
      • However, people may systematically vary with other factors that economists cannot observe as a result of choices like these.
      • Fortunately, econometric methods can alleviate this problem by controlling for workers’ abilities and participation decisions, thereby accounting for everyone in the population.
  • Occupational discrimination is again difficult to measure because of the trouble isolating discrimination and voluntary choice.
    • First it is necessary to measure the disparity of the occupational distribution
    • Occupational distribution: a graphical measure of the respective fractions of men and women who are employed in various occupations
    • Duncan index of dissimilarity, D: is the fraction of women who must switch occupations in order to bring about complete equality with the given male occupational distribution, where 0 ≤ D ≤ 1 and the level of dissimilarity increases with D.
      • If D = 0, the distributions are identical. If D = 1, they are completely dissimilar.
      • D = ½Σ |fi - mi|

12.2 Race

  • Since 1940, the ratio of average Black earnings to White earnings has changed with transformations in the Black labor market.
    • Up until the mid-1970s the manufacturing sector was becoming more important relative to agriculture, making high-paying manufacturing jobs available to Blacks.
    • Beginning in the mid-1970s, the growth of service industries eclipsed the manufacturing industry, resulting in job losses that were especially hard for Black workers.
    • The same period also introduced new laws to combat workplace discrimination.
  • Overall, the raw data show that in the past 50 years the ratios of the median full-time earnings of Black men and women have increased relative to White men, particularly from 1963 to 1973.
    • Increased educational attainment levels for Black men may account for some of this increase.
    • Tight labor markets and anti-discrimination laws in the 1960s also might have narrowed the wage gap.
    • However, Black economic progress may be deceiving due to self-selection bias.
      • Selective withdrawal is a source of self-selection bias in which low-skilled low-wage Black workers remove themselves from employment. Ceteris paribus selective withdrawal would increase the observed relative wages of those Black men who remain employed.
        • ep = employment-to-population ratio
        • u = unemployment rates
        • Between 1972 and 1975 the Black ep ratio decreased as the relative wages of Black men appeared to be increasing, indicating that selective withdrawal may have been taking place.
          • Since the unemployed are largely made up of low-skill low-wage workers, the remaining employed Black men have relatively high abilities and high earnings when compared to White men.
      • Since the 1980s a shift in earning distributions increased the wages of skilled workers and lowered the wages of unskilled workers.
        • Since Black men are, on average, less skilled than White men, this shift is again predicted to lead to selective withdrawal of low-wage Black workers, inflating the observed relative earnings of Black workers who remain employed.
    • The systematic difference in earnings maybe due to the differing occupational distributions of Black and White workers.
      • The Duncan index of dissimilarity for Black and White men was 0.28 in 1998
      • However, Hirsch and Macpherson (2004) found that low earnings in predominantly Black occupations are explained by the low abilities of the both Black and White workers employed there.
    • The wage gap may also be the product of human-capital differences and labor-market discrimination.
      • Empirical research and many Oaxaca transformations have revealed that policies to curtail labor-market discrimination are misplaced.
      • Instead, policies should address the root causes of identified productivity differences, which include unequal schooling quality for Black children and extreme family poverty.
  • Though they are subject to both racial and gender discrimination, Black women appear to have made considerable economic progress.
    • The ep ratios of Black and White women suggest that the Black-White female wage gap has narrowed to 15% in 2009 and that the remaining gap is due to productivity differences rather than discrimination.
    • Neal (2004) found that the observed Black-White female wage gap may be 60% smaller than its true value due to selectivity bias.
      • The observed wage distribution of White women understates their true average wage because a disproportionate number of high-skilled women withdrew from the labor force to raise children.
      • The observed wage distribution of Black women overstates their true average wage because a disproportionate number of women who would have earned less withdrew from employment.
      • The result is an understated assessment of the wage differential.

12.3 Gender

  • There is intense disagreement over the gender-based wage differential.
    • One side sees the differential as the outcome of prejudicial discrimination.
    • The other views the differential as the result of voluntary choices in a nondiscriminatory labor market.
  • There is considerable gender-based occupational segregation in the United States, amounting to a Duncan index of 0.68, 0.59, and 0.59 in 1970, 1980, and 1990, respectively.
    • Though this segregation has declined, it remains high.
    • A disproportional number of women are employed in low-wage occupations.
    • Today the median full-time weekly earnings of women are about 80 % of men’s.
  • Women may be overrepresented in low-wage occupations because of occupational discrimination or because of their own choices.
    • The human-capital model suggests that women select occupations best suited to their lifetime plans. Since they periodically withdraw from the labor force to raise children, they are predicted to choose lower and more durable levels of human capital investments than men.
      • Though the costs of human capital acquisition are the same, women accrue fewer benefits because of tenuous labor-market attachment.
    • Audit studies seek to uncover discrimination in employers’ hiring practices by looking for barriers to entry in certain occupations. The studies match up pairs of individuals who differ only in a nonproductive trait of interest and send them to apply for job openings.
  • Some Oaxaca transformations have revealed that observable worker characteristics explain only about a quarter of the gender wage gap.
    • Today, a growing body of evidence points to differences in actual experience levels as the cause of the gender wage gap.
    • The human capital model predicts that women will have less labor-market experience because of their decision to raise children.
      • Because of difficult accounting for atrophy of skills, economists often measure experience using potential experience as a proxy.
      • The increase in the overall level of female work experience may account for the narrowing of the observed male-female wage gap.
        • Women’s increased attachment to the labor force encourages them to make human capital investments.
        • An increase in the amount of time that women spend employed protects their human capital from atrophy.
    • In a study controlling for productivity and wage differences, Hellerstein and Neumark (1999) find no gender discrimination.

12.4 Public Policy

  • Since the mid-1950s the U.S. government has initiated many policies to eliminate discrimination in the workplace.
    • These policies emanated from U.S. Supreme court rulings and through the creation of new federal, state, and local laws.
  • In the early 1950s there was deep racial, ethnic, and gender discrimination.
    • Jim Crow laws led to de jure segregation of Blacks and Whites.
    • Plessy v. Ferguson (1896) supported segregation by ruling “separate but equal” accommodations constitutional.
    • Protective legislation that had been created in the 1880s to shield women and children from the effects of the Industrial Revolution were badly out of date by the early 1950s. These laws contributed to gender discrimination.
  • Brown v. The Board of Education overturned the decision made in Plessy v. Ferguson and eliminated the separate but equal foundation of Jim Crow laws.
    • The landmark decision provided the basis for legal attacks to break down state-sponsored racial segregation.
  • The Equal Pay Act of 1963 neutralized protective legislation by prohibiting wage discrimination based on sex.
    • The law specified equal pay for equal work.
    • Since it only specified equality of the wage rate, the law provided a large loophole for prejudiced employers to discriminate against women in other ways, such as hiring decisions.
  • The Civil Rights Act of 1964 was a very broad piece of legislation that aimed to end discrimination of all kinds based on race, color, religion, sex, and national origin.
    • The law also created the Equal Employment Opportunity Commission (EEOC) to enforce the legislation.
    • Since its passage the act has been modified to prevent discrimination based on age or disabilities.
    • Though the law renders discrimination illegal, it makes it no less difficult to define discrimination. Two very different definitions have emerged over the years.
      • Disparate treatment: an employer, motivated by discriminatory intent, treats applicants or employees differently because of their membership in a protected class.
      • Disparate impact: there is a systematic pattern revealing that an employer eschews hiring suitably qualified women and minorities.
  • Griggs v. Duke Power (1971) ruled in favor of the disparate impact interpretation of the laws.
    • In 1989 Ward’s Cove Packing Co. v. Antonio reduced the employer’s burden from establishing business necessity to the weaker standard of business justification.
    • The Civil Rights Act of 1991 reestablished business necessity.
  • Executive Order 11246 ruled that any employer with government contracts whose value exceeds $50K must implement an affirmative action plan with five elements:
      • A detailed analysis of the employment of women and minorities within the firm
      • An explanation of any underutilization that is detected
      • Specific goals and timetables for increasing their employment
      • Steps to meet these goals
      • Timetables describing internal auditing procedures to ensure the plan’s success
    • Regents of the University of California v. Bakke (1978) ruled against rigid college admissions quota systems favoring minorities.
  • The Civil Rights Act of 1991 sought to reverse recent Supreme Court decisions. It reestablished the disparate impact standard by forcing employers to prove that workplace standards are job related and consistent with business necessity.

12.5 Antidiscrimination Policies: The Evidence

  • In addition to political reforms, economic changes since the 1950s were somewhat responsible for the improvement in the relative wages of black men.
  • Studies find that between 1960 and 1975 at least 35% of the decline in the black-white wage gap may be attributable to policy changes.
    • Black economic progress occurred in jumps that reflect the discontinuous nature of the political process.
      • The greatest improvement was in the South, where political efforts were most concentrated.
    • Holzer and Neumark (2000a) survey the efforts to assess affirmative action and find that the policies have been effective.
    • Heckman and Payner (1989) also find that black progress cannot be attributed to economic changes alone. They argue that political changes are the only explanation.
  • The Americans with Disabilities Act (1990) may have had unintended adverse consequences for disabled workers.
    • Employers may have been less inclined to hire disabled workers due to the potentially high legal costs of firing a disabled worker.