CHAPTER 7: THE WAGE STRUCTURE — EXAM-READY

SUMMARY
Borjas, Labor Economics (McGraw-Hill, 2016)

CHAPTER OVERVIEW
This chapter examines the factors that determine the shape of the wage distribution. Two key
"fundamentals" drive wage dispersion: (1) productivity differences among workers, and (2) the
rate of return to skills, which varies across labor markets and over time. The U.S. wage
distribution changed dramatically in the 1980s, with sizable increases in inequality both across
and within skill groups. The chapter also explains the "superstar phenomenon" and how wage
inequality transmits across generations via intergenerational investments in human capital.
Chapter structure:
Section 7-1: The Earnings Distribution (shape and skewness)
Section 7-2: Measuring Inequality (Lorenz curve, Gini coefficient)
Section 7-3: The Wage Structure: Basic Facts (trends in U.S. inequality)
Section 7-4: Policy Application: Why Did Wage Inequality Increase?
Section 7-5: The Earnings of Superstars
Section 7-6: Inequality across Generations (intergenerational mobility)

SECTION 7-1: THE EARNINGS DISTRIBUTION

Short Summary
The U.S. wage distribution (2012 data) shows two key features: substantial wage dispersion, and
positive skewness (a long right tail). The mean weekly wage ($959) exceeds the median ($769),
indicating most workers earn relatively low wages while a small number at the top earn
disproportionately high wages. International data confirm that the U.S. is more unequal than
most developed nations. Human capital theory explains positive skewness through the
interaction of ability and investment decisions.

Bullet-Point Notes
Key Facts about the Wage Distribution:
Mean weekly wage (2012) = $959; Median = $769 — the gap confirms positive skewness
Positively skewed = bulk of workers earn below average; a small number in the upper tail
earns very large shares
Top 10% of U.S. households receive 30% of total income; bottom 10% receive only 2%
 For comparison: Top 10% in Germany = 22%; in Mexico = 41%; in Guatemala = 43%
Why is the Wage Distribution Positively Skewed? (Human Capital Explanation — Figure 7-2):
Workers invest in human capital until the marginal rate of return (MRR) equals the
discount rate (r)
Diagram: Three downward-sloping MRR curves — MRRL (low-ability), MRR* (medium-
ability), MRRH (high-ability) — all meeting the same horizontal discount rate line (r)
Low-ability workers acquire HL units of human capital; medium acquire H*; high-ability
acquire HH
High-ability workers earn more for two reasons: (a) higher innate ability raises
productivity directly, AND (b) they acquire more human capital (higher investment)
This positive correlation between ability and human capital investment "stretches out"
wages, producing positive skewness even if the ability distribution itself is symmetric

International Comparison (Table 7-1):

Nordic countries (Norway, Sweden) most equal: top 10% receive only 22–23%
Latin American countries (Mexico, Guatemala, Chile) most unequal: top 10% receive 41–
43%
The U.S. is more unequal than most rich nations but less unequal than developing countries

MUST-REMEMBER POINTS
Positive skewness = long right tail = mean > median
Two reasons high-ability workers earn more: innate ability + more human capital
acquired
The stopping rule (invest until MRR = r) generates skewness even from a symmetric
ability distribution

Common Exam Questions:

Why is the wage distribution positively skewed? Explain using the human capital model.
What does Figure 7-2 show about the relationship between ability and human capital
investment?
Why does the mean wage exceed the median wage?

SECTION 7-2: MEASURING INEQUALITY

Short Summary
There are several ways to measure income inequality. The most common graphical tool is the
Lorenz curve, and the most common numerical summary is the Gini coefficient. Additional
measures include the 90-10 wage gap and 50-10 wage gap, which capture different parts of the
distribution. Each measure has limitations, and no single number fully captures the shape of an
income distribution.

Bullet-Point Notes
Quintile Analysis:
Rank all households by income; divide into 5 equal groups (quintiles)
U.S. data (2010 — Table 7-2):

Quintile Share of Income Cumulative Share

First (bottom 20%) 3.4% 3.4%

Second 8.6% 12.0%

Third 14.7% 26.7%

Fourth 23.3% 50.0%

Fifth (top 20%) 50.0% 100.0%

The Lorenz Curve (Figure 7-3):

Plots cumulative share of income (y-axis) against cumulative share of households (x-
axis)
Perfect equality line (AB): A 45° straight line — 20% of households earn 20% of income,
etc.
Actual Lorenz curve: Lies below the perfect-equality line; the more unequal the
distribution, the further it bows away from the 45° line
"Perfect inequality" Lorenz curve = an inverted-L shape (all income goes to the top quintile)
Lorenz curves from two different distributions can intersect, making comparison
ambiguous

Gini Coefficient (Formula 7-1):

Area between perfect-equality Lorenz curve and actual Lorenz curve

Gini coefficient =
Area under perfect-equality Lorenz curve (triangle ABC)
​

Ranges from 0 (perfect equality) to 1 (perfect inequality)

U.S. Gini coefficient for household income ≈ 0.43
Note: Area of triangle ABC = 0.5 (denominator in the formula)
Limitation: The same Gini value can be produced by different redistributions (e.g., taking
from the very bottom vs. from the middle), so it doesn't fully describe the shape
Additional Inequality Measures:
90-10 wage gap: Percent wage differential between the worker at the 90th percentile and
the 10th percentile → measures the range of the distribution
50-10 wage gap: Percent differential between the 50th percentile and the 10th percentile →
measures inequality between the middle class and low-income workers
The 80-50 gap vs. the 50-20 gap can also reveal where in the distribution inequality is
concentrated

MUST-REMEMBER POINTS
Perfect equality Lorenz curve = 45° line; actual curve bows below it
Gini = 0 means perfect equality; Gini = 1 means perfect inequality
U.S. Gini ≈ 0.43
Gini is a single-number summary and misses details about where inequality occurs in
the distribution

Common Exam Questions:

Draw and explain the Lorenz curve. What does the area between the curves represent?
Define and calculate the Gini coefficient. What are its limitations?
What is the difference between the 90-10 and 50-10 wage gaps?

SECTION 7-3: THE WAGE STRUCTURE — BASIC FACTS

Short Summary
Beginning around 1980, the U.S. labor market saw a historic increase in wage inequality. This
happened both across skill groups (college vs. high school graduates) and within narrowly
defined skill groups. The Gini coefficient rose sharply from the 1970s onward, driven mainly by
the widening of the upper end of the distribution. The college–high school wage premium made
a "great U-turn" after 1979, rising from 47% to 90% by 2001.

Bullet-Point Notes
Key Documented Changes in the U.S. Wage Structure (post-1979):
The wage gap between those at the top and bottom of the distribution widened
dramatically
Wage differentials widened among education groups, among experience groups, and
among age groups
Wage differentials also widened within demographic and skill groups (same education,
age, sex, occupation)
Trend in the Gini Coefficient (Figure 7-4):
Gini declined from the 1930s through 1950 (compression during wartime/post-war)
Relatively stable 1950–1970
Dramatic rise after 1970, especially post-1980
Most of the rise is due to the widening 80-50 wage gap — inequality at the upper end is the
primary driver

The College–High School Wage Gap (Figure 7-5):

Rose slightly in the 1960s, then declined through most of the 1970s
Made a "great U-turn" around 1979 and rose very rapidly
1979: College graduates earned 47% more than high school graduates
2001: College graduates earned 90% more than high school graduates
Concurrent rise in the experience premium (returns to experience also rose)

Residual ("Within-Group") Inequality (Figure 7-6):

The 90-10 wage gap within groups of workers with the same age, education, gender, and
race also rose sharply from the late 1970s to the late 1990s
This shows inequality increased even among workers with similar observable
characteristics
Key implication: Changes in the wage structure were pervasive throughout the entire labor
market, not just between broad skill categories

MUST-REMEMBER POINTS
Wage inequality increased BOTH across skill groups AND within skill groups
College–HS wage premium: 47% (1979) → 90% (2001)
The Gini declined through 1950, was stable until ~1970, then rose dramatically
"Residual" inequality = within-group inequality = unexplained by measurable
characteristics

Common Exam Questions:

What happened to U.S. wage inequality after 1979? Describe using at least two measures.
What is "residual wage inequality" and why is it important?
Describe the trend in the college–high school wage premium from 1963 to 2005.
SECTION 7-4: POLICY APPLICATION — WHY DID WAGE INEQUALITY
INCREASE?
Short Summary
Despite extensive research, no single factor explains the rise in U.S. wage inequality. The
dominant framework models the labor market with two types of workers (skilled/unskilled) and
examines how shifts in relative supply and relative demand affect the wage ratio. The key
finding: the relative demand for skilled workers must have risen faster than the relative supply.
The main "usual suspects" — supply shifts, immigration, international trade, skill-biased
technological change, declining unions, and the minimum wage — each explain part, but no
single one dominates.

Bullet-Point Notes
The Supply–Demand Framework (Figure 7-7):
Two worker types: skilled and unskilled. Let:
r = wage ratio (skilled wage / unskilled wage)
p = ratio of skilled workers to unskilled workers
Demand curve: Downward sloping — as r rises, employers hire relatively fewer skilled
workers
Supply curve: Treated as perfectly inelastic (p is fixed in the short run)
Initial equilibrium at Point A (S₀ and D₀): wage ratio r₀, fraction p₀ skilled
Supply increased (S₀ → S₁) = more college graduates; alone this would push to Point B
(lower r)
Demand also increased (D₀ → D₁) = final equilibrium at Point C (higher r AND higher p)
Conclusion: Demand must have outpaced supply to explain the observed rise in the
skilled wage premium

1. Supply Shifts:
Education levels rose dramatically (Table 7-3): High school dropouts fell from 50.1% (1960)
to 8.5% (2010); college grads rose from 9.3% to 30.2%
Baby boom cohort entered the 1970s labor market → shifted college graduate supply
outward → depressed the college wage premium in the 1970s
Immigration: Between 1979–1995, immigration raised high school dropout supply by
20.7% but high school graduate supply by only 4.1% → increased unskilled labor supply
Immigration accounts for about one-third of the decline in relative wages of high school
dropouts between 1980 and 1995
Verdict: Supply shifts alone cannot explain overall inequality rise because supply of skilled
workers was growing at the same time as their wages rose

2. International Trade:
 Exports-to-GDP ratio rose from 8% (1970) to 19% (1996); 40% of imports by 1996 came
from less-developed countries
U.S. exports use skilled labor; U.S. imports from LDCs are produced by unskilled labor
Effect: Trade increases demand for skilled workers and reduces demand for unskilled
workers → rightward shift in relative demand curve (Figure 7-7)
Highly unionized, concentrated industries (autos, steel) hit hardest by imports → less-
skilled workers displaced to competitive, lower-wage sectors
Chinese import competition reduced manufacturing wages in the typical U.S. locality by
about 1 percent
Verdict: Trade explains some of the change, especially for unskilled workers, but not all of
the rise in within-group inequality

3. Skill-Biased Technological Change (SBTC):

Technology (especially computers/IT) substitutes for unskilled labor and complements
skilled labor → raises demand for skilled workers relative to unskilled
In 1984: 25% of workers used computers; by 1997: 50% — but 75% of college graduates vs.
only 11% of high school dropouts used computers
Workers using computers earned ~18% more (1989) — interpreted as returns to computer
use
BUT: "Pencils" problem — German workers who use pencils earn 14% more than those who
don't; clearly not technological change → suggests the wage premium may reflect pre-
existing productivity differences, not computer use itself
Major critique: SBTC hypothesis doesn't reconcile with the timing — inequality rose
sharply in the 1980s but the information revolution accelerated in the 1990s, yet within-
group inequality may have slightly declined in the 1990s
Verdict: Likely an important contributor but the evidence is not conclusive; no widely
accepted measure of SBTC exists; "residual" methodology attributes unexplained variation
to SBTC by default

4. Declining Unions:
Union membership fell from 24% (1973) to 12% (2010) of the workforce
Unions traditionally raised wages of less-skilled (non-college) workers and compressed the
wage distribution
De-unionization effectively shifts the relative demand curve for skilled workers upward
Estimated contribution: About 10% of the rising college–high school wage gap can be
attributed to union decline
Verdict: Explains part of the story but not the within-group inequality rise

5. Minimum Wage:
 Nominal minimum wage was frozen at $3.35/hour from 1981 to 1989; in real terms fell from
$5.62 (1981) to $4.12 (1990) in 1995 dollars
Declining real minimum wage compressed wages at the very bottom → raised wage gap
between low-skill and other workers
Verdict: Explains some compression at the bottom but cannot explain the rise in the
college–HS gap or within-group inequality among the educated

Problems with Existing Explanations:

No single story explains the entire picture:
Immigration and trade explain the unskilled–skilled gap but not within-group
inequality
Minimum wage explains the bottom of the distribution only
SBTC has timing problems
De-unionization contributes modestly
Any complete theory must explain both timing and structure of changes throughout the
entire distribution

International Comparison Puzzle (Table 7-4):

The 90-10 wage gap rose significantly in the U.S. (267% → 326%, 1984–1994) and U.K. (177%
→ 222%)
But fell in Germany (139% → 125%) and barely changed in Japan and Norway
If SBTC was global (same information revolution everywhere), why did countries diverge?
Hypothesis: Countries differ in labor market institutions (safety nets, unions, minimum
wages):
In the U.S., the same economic shocks produced wider wages (flexible institutions)
In Europe, stronger safety nets prevented wage compression, manifesting instead as
higher unemployment
"Prices vs. quantities" trade-off: labor markets adjust through wages OR employment,
not both equally

MUST-REMEMBER POINTS
In Figure 7-7: supply (S₀→S₁) alone would lower the skill premium; demand must also
shift (D₀→D₁) to reach Point C
"The demand curve won the race" — relative demand for skilled workers grew faster
than relative supply
No single factor explains the full increase; it is multi-causal
Union decline explains ~10% of the college-HS wage gap increase
The international puzzle: same technology shock, different wage outcomes — due to
institutional differences
 Same economic shock → wage dispersion (U.S.) vs. unemployment (Europe)

Common Exam Questions:

Use Figure 7-7 to explain why relative wages of skilled workers rose despite an increase in
their relative supply.
What role did skill-biased technological change play? What are the main criticisms of this
explanation?
Explain why immigration alone cannot explain the overall rise in U.S. wage inequality.
Why did wage inequality rise in the U.S. but not uniformly across other OECD countries?

SECTION 7-5: THE EARNINGS OF SUPERSTARS

Short Summary
In certain occupations, a tiny number of workers earn astronomically high wages while most
workers in the same field earn very little. This is the superstar phenomenon. It requires two
conditions: (1) sellers are not perfect substitutes (talent differences matter enormously to
consumers), and (2) technology allows the very talented to reach very large markets at low cost.
The phenomenon does not arise in all occupations — only those where technology enables mass
distribution.

Bullet-Point Notes
What is the Superstar Phenomenon?
A few persons in certain professions dominate the field and earn salaries vastly exceeding
those of average workers in the same occupation
Examples: Oprah Winfrey ($315M, 2010), entertainers earning $50M+, top baseball players
($28M/year)
Contrast: Even the best economics professors earn only $400K maximum; talented grocery
clerks earn at most 2–3× the typical salary

Two Conditions Required for Superstardom:

Condition 1 — Imperfect substitutability: Sellers differ in quality in ways consumers
strongly prefer. A single trip to the plate by Hank Aaron produces far more excitement than
1,000 trips by an average player. Consumers will pay a large premium for the best.
Condition 2 — Technology of mass distribution: The cost of distributing the product does
not rise proportionally with market size. Madonna records a song once; technology
distributes it to millions at near-zero marginal cost. In contrast, a heart surgeon must
personally attend each patient → the market size for her services is physically constrained.

Why Superstars Exist in Some Occupations but Not Others:

 Rock musicians, athletes, actors → technology enables mass reach → superstar
phenomenon arises
Heart surgeons, professors, grocery clerks → personal contact required → market size is
limited → no superstar phenomenon (or only a muted version)
Rock concert evidence: Each additional 5 inches of attention in Rolling Stone Encyclopedia
allowed 3% higher ticket prices in early 1980s; 7% by late 1990s (returns to superstardom
increased as technology changed)
The shift to digital music reduced album revenue → concerts became the main revenue
source → concert prices rose sharply (typical Paul McCartney ticket: $288 in 2010)

The Key Insight:

Small differences in talent can generate enormous differences in pay when talent is not
substitutable and reach is unlimited
The superstar phenomenon is a demand-side story: consumers prefer the very best and
technology allows them to access it globally

MUST-REMEMBER POINTS
Two conditions: (1) imperfect substitutability, (2) low-cost mass distribution
technology
Superstar phenomenon does NOT arise in all occupations — only where technology
allows mass reach
Theoretical foundation: Sherwin Rosen (1981), "The Economics of Superstars"
The phenomenon has intensified over time as distribution technology improved

Common Exam Questions:

What two conditions are required for the superstar phenomenon to arise?
Why do athletes and entertainers become superstars but not heart surgeons or professors?
How has changing technology in the music industry affected the superstar phenomenon?

SECTION 7-6: INEQUALITY ACROSS GENERATIONS

Short Summary
Wage inequality is not just a within-generation phenomenon — it transmits across generations.
Parents invest in their children's human capital, creating a positive correlation between parental
and children's earnings. The strength of this link is measured by the intergenerational
correlation (slope of the regression line linking parent and child earnings). Early research
estimated this at ~0.2, but later work corrected for measurement error and revised the estimate
upward to 0.3–0.4, implying greater persistence of inequality across generations.
Bullet-Point Notes
The Intergenerational Framework:
Parents care about their own well-being AND their children's well-being
High-income parents invest more in children's human capital → positive correlation
between parent and child socioeconomic outcomes
This investment-based link creates persistence of inequality across generations

The Intergenerational Correlation (Figure 7-8):

Regression line linking earnings of children (y-axis) to earnings of parents (x-axis)
Slope = intergenerational correlation
Three cases:
Line A (slope = 1): A $1,000 parental earnings gap produces a $1,000 children's
earnings gap — no regression toward the mean; complete transmission of inequality
Line C (slope between 0 and 1): Partial transmission; earnings differences shrink
across generations
Line B (slope = 0): Children's earnings are independent of parental earnings —
complete regression toward the mean

Regression toward the Mean:

The tendency for income differences to shrink across generations as families move toward
the mean
Causes:
Parents consume some wealth rather than investing all of it in children
Diminishing returns to investing in children's human capital
Ability exhibits its own regression toward the mean (children of very bright parents
aren't necessarily brighter)
The closer the correlation is to 0, the faster the regression toward the mean

Empirical Estimates:
Early estimates: Intergenerational correlation ≈ 0.2 (Becker and Tomes)
Implication: A 30% parental wage gap → 6% gap for children → 1.2% gap for
grandchildren → very high social mobility
Revised estimates: ≈ 0.3 to 0.4 (Solon 1992; Zimmerman 1992)
Reason for revision: Earlier studies suffered from measurement error in parental
income (self-reported data); correcting for this roughly doubles the estimated
correlation
Implication: A 30% parental wage gap → 12% gap for children → ~5% gap for
grandchildren → considerably less social mobility than originally thought
 Conclusion: Skill and income differentials are more persistent across generations than
originally believed
Slavery Study:
Examined grandchildren of U.S. slaves vs. grandchildren of free blacks (Sacerdote 2005)
Having a slave mother reduced the probability of school enrollment in 1880 by 36%
By 1920, having a slave grandmother reduced school enrollment by only 8.8%
It took approximately two generations for the descendants of slaves to catch up with
descendants of free blacks (in terms of this measure)
This does NOT imply catch-up between the black and white populations overall

Nature vs. Nurture (Theory at Work — Korean Adoption Study):

Swedish data on Korean children adopted by Swedish families: since assignment to families
was essentially random, it separates prebirth (genetic/nature) factors from postbirth
(environment/nurture) factors
Allows estimation of how much of intergenerational transmission is due to genes vs.
environment

MUST-REMEMBER POINTS
Intergenerational correlation = slope of regression line linking parent earnings to child
earnings
Early estimate: 0.2; revised estimate: 0.3–0.4 (measurement error correction)
Higher correlation = less social mobility = more persistent inequality
Slope = 1 → no regression to the mean; Slope = 0 → complete regression to the mean
Revised estimates suggest the U.S. has less social mobility than previously thought

Common Exam Questions:

What is the intergenerational correlation and what does it measure?
Why was the early estimate of 0.2 revised upward to 0.3–0.4?
Explain "regression toward the mean." Under what conditions would it be fastest?
What are three reasons why income differences shrink across generations?

CHAPTER 7 MASTER SUMMARY — MUST-REMEMBER POINTS

Topic Key Fact / Formula

Wage distribution shape Positively skewed; mean ($959) > median ($769) in 2012
Topic Key Fact / Formula

Reason for positive Ability and human capital investment are positively correlated; high-ability
skewness workers invest more

Lorenz curve Plots cumulative income share vs. cumulative household share; bows below
45° line

Gini coefficient formula Area between perfect-equality Lorenz and actual Lorenz ÷ area of triangle
ABC

U.S. Gini (household ≈ 0.43

income)

Gini range 0 (perfect equality) to 1 (perfect inequality)

90-10 wage gap % differential between 90th and 10th percentile worker; measures range of
distribution

50-10 wage gap % differential between 50th and 10th percentile; measures middle-class vs.
low-income inequality

College-HS wage gap 47% (1979) → 90% (2001); "great U-turn" after 1979

Post-1979 inequality Rose BOTH across and within skill groups

Supply-demand model r = skill wage ratio; p = fraction skilled; demand outpaced supply → Point C
(Fig 7-7)

Immigration effect Raised dropout supply by 20.7% but graduate supply only 4.1% (1979–1995)

Immigration's share ~1/3 of decline in relative wages of high school dropouts

Trade effect Imports hurt unskilled; exports help skilled; Chinese imports reduced
manufacturing wages ~1%

Union decline 24% (1973) → 12% (2010) unionized; explains ~10% of college-HS gap
increase

Real minimum wage Fell from $5.62 (1981) to $4.12 (1990) in 1995 dollars

Superstar conditions (1) Imperfect substitutability; (2) technology enables low-cost mass
distribution

Intergenerational Early estimate: 0.2; revised: 0.3–0.4 (correcting for measurement error)
correlation

Regression toward the Income differences shrink across generations; faster when correlation is
mean closer to 0
 Topic Key Fact / Formula

International puzzle Same technology shock → wider wages in U.S. (flexible institutions) vs.
higher unemployment in Europe (rigid institutions)

End of Chapter 7 Summary