Microeconomics › Section N · Labor markets and income › Topic N.5

Employment discrimination

A pay gap between groups is a fact; discrimination is a conclusion - and the distance between them is where this topic lives. Economics brings to the question a definition, a method for separating gaps into explained and unexplained parts, and a theory of why markets punish some discrimination but not all of it.

The definition and the discipline

Employment discrimination means paying or hiring people differently on the basis of gender, race or similar characteristics DESPITE comparable education, experience and expertise. That final clause carries the discipline. Raw earnings gaps between groups are facts, but they do not by themselves prove discrimination, because averages also differ in education, experience, hours and occupation. The task is to compare comparable workers: only a gap that survives the comparison is a candidate for the discrimination label, and such a gap is a serious finding.

The measured gaps

For full-time, year-round workers in 2024, women's median earnings were about 81% of men's (Census Bureau), down from about 83% in 2023 after slow narrowing since the 1980s. In 2025 Bureau of Labor Statistics quarterly data, Black full-time workers' median weekly earnings were roughly 80% of White workers'. The racial ratio rose sharply in the late 1960s and 1970s, then stalled; by some measures it is no narrower than decades ago. Figures differ with the measure (weekly or annual, median or mean), so check which one a claim uses. Why the gaps exist, and how much is discrimination, is what the rest of this topic dissects.

Becker's question

Gary Becker (Nobel memorial prize, 1992) asked the awkward question: if prejudice is costly, why does discrimination persist in competitive markets? His answer sorts prejudice by its source. Employer discrimination - refusing equally productive, cheaper workers - costs the discriminator money, so profit-seeking rivals who hire the passed-over group should outcompete it. Customer discrimination - buyers preferring to be served by a particular group - can make exclusion profitable, so the market does not punish it. Coworker discrimination - prejudiced employees demanding extra pay to work alongside a group - gives employers a reason to segregate workplaces, which competition need not remove. The market's penalty falls mainly on the first channel, and history confirms its limits: discrimination against Black Americans persisted from the abolition of slavery in 1865 to the Civil Rights Act of 1964 and beyond, much of it imposed by segregation laws rather than left to markets.

Worked example: what prejudice costs the employer

Two firms each need 100 equally productive workers. Workers of group B are willing to work for $22 an hour; the discriminator's preferred group A commands $25. 1. The colour-blind employer hires the best value: 100 workers at $22, an hourly labour bill of $2,200. 2. The prejudiced employer refuses group B and hires 100 group-A workers at $25: $2,500 an hour, a $300 hourly penalty ($2,400 over an eight-hour day) purely for refusing identical talent. 3. The competitive logic: the rival's lower costs let it price lower, win customers and expand. 4. The limits: if CUSTOMERS refuse to be served by group B, the colour-blind plan loses revenue and the penalty can flip into a reward. Assumptions: equal productivity; wages as stated.

The cost of excluded talent

Discrimination is an injustice to those excluded and also a waste the whole economy pays for. When groups are kept out of fields, the inventions, firms and ideas they would have produced never appear. Economist Lisa D. Cook has measured this for innovation: she estimates that US GDP could be as much as about 4.4% higher if women and African Americans took full part in invention. Any labour market that screens out talent for reasons unrelated to productivity leaves the economy inside its production possibilities frontier, a loss borne by everyone, including those doing the excluding.

Roots of the gender gap

Research finds layers rather than one cause. Occupational sorting: men and women still work in different occupations and industries, which explains a large share of the raw gap. Care work: women carry more unpaid household and care work, which shapes hours and experience; mothers earn less than comparable women without children. Devaluation: studies of 1950-2000 found that as occupations became more female, their relative pay fell. Norms: Kerwin Charles, Jonathan Guryan and Jessica Pan found that women born in US states with more sexist attitudes earn less throughout life, even after moving away. Claudia Goldin, awarded the 2023 Nobel memorial prize for work on women's labour-market outcomes, stresses that jobs paying a premium for long, inflexible hours penalise whoever carries more care work.

Roots of the racial gap

Before 1964, refusing to hire a qualified Black applicant was lawful across much of the country. Legacies persist through education: gaps in college completion, rooted in unequal schools, wealth and neighbourhoods, explain a substantial share of today's earnings gap, so many economists see narrowing the education gap as a major lever. Education is not the whole story: in a 2004 field experiment, Marianne Bertrand and Sendhil Mullainathan found that CVs with White-sounding names received about 50% more interview callbacks than identical CVs with Black-sounding names. Housing matters too: a 2012 federal housing study found Black and Asian homebuyers were shown fewer homes than White buyers, shaping which schools and job networks families can reach.

Worked example: reading a pay-gap claim

An analyst finds that group A earns $30 an hour on average, group B $24 - a raw gap of 20 percent. 1. Compare the groups' measured characteristics: education, experience, occupation, hours. Group A averages more of each. 2. Statistically adjust - compare workers as if identical in those measured characteristics: the gap falls to about $26 versus $24, roughly 8 percent. 3. Read both numbers honestly: most of the raw gap tracks measured differences, themselves partly the product of past discrimination in schools and housing. The residual 8 percent is the candidate for current labour-market discrimination, though it may also contain unmeasured factors. 4. The verdict: the unexplained part justifies investigation, such as matched-CV testing, rather than dismissal or instant verdict. Assumptions: adjustment done on the stated characteristics.

The policy toolkit

Legal prohibition comes first. The Equal Pay Act of 1963 requires equal pay for equal work regardless of sex. Title VII of the Civil Rights Act of 1964 bars employment discrimination based on race, colour, religion, sex or national origin, and in Bostock v. Clayton County (2020) the Supreme Court held that 'sex' covers sexual orientation and gender identity. The Equal Employment Opportunity Commission (EEOC) enforces these laws. Enforcement is hard: in Wal-Mart v. Dukes (2011) the Court refused to let a nationwide sex-discrimination claim proceed as one class action. Affirmative action has narrowed sharply: the Court ended race-conscious university admissions in 2023, and a January 2025 executive order revoked the 1965 order requiring federal contractors to keep affirmative-action plans. Court-ordered remedies after proven discrimination remain.

Common mistakes to avoid

First, treating a raw earnings gap as proof of discrimination - the whole method is comparing comparable workers. Second, treating an adjusted gap as proof of innocence - the adjustment's own inputs (education, occupation) are shaped by past discrimination elsewhere. Third, expecting markets to fix everything: Becker's penalty falls mainly on employer discrimination; customer and coworker prejudice can survive. Fourth, expecting one cause: the gaps are layered - sorting, care burdens, devaluation, education, housing, hiring bias. Fifth, quoting a gap without its measure and year: 81% and 83% can both be right for different measures or years.

Key takeaways

Discrimination means less pay or fewer jobs for comparable workers: a raw gap (20 percent in the worked claim) partly tracks education, experience, occupation and hours, while the adjusted gap (8 percent) is the candidate for discrimination. Becker: employer discrimination costs the discriminator ($300 an hour in the worked example), but customer and coworker prejudice pay no such penalty. The gaps are layered, from sorting and care work to schooling, housing and biased hiring, and exclusion costs everyone. Policy rests on the 1963 and 1964 Acts and the EEOC, while affirmative action has narrowed sharply since 2023.

Flashcards

Flashcard

Employment discrimination

Less pay or fewer jobs for comparable workers, based on group identity

The 'comparable' clause is the discipline.

Test your understanding

Question 1 of 5

Why does a raw earnings gap between two groups not, by itself, prove discrimination?

Question 2 of 5

In Becker's analysis, why does CUSTOMER discrimination persist in competitive markets?

Question 3 of 5

In the worked example, 100 equally productive workers cost $22 an hour from group B or $25 from group A. What does refusing group B cost the prejudiced employer?

Question 4 of 5

What did the 2004 CV field experiment by Bertrand and Mullainathan find?

Question 5 of 5

How does housing discrimination connect to employment discrimination?

← Previous topicNext topic →