[Economic Report of the President (1998)]
[Administration of William J. Clinton]
[Online through the Government Printing Office, www.gpo.gov]

[DOCID: f:erp_c4._]
Economic Report of the President - - - - - - - - - - - - H. Doc. 105-176
[From the online service of the the U.S. Government Printing Office]
[wais.access.gpo.gov]



CHAPTER 4 -- Economic Inequality Among Racial and Ethnic Groups

THIRTY-FOUR YEARS AGO the signing of the Civil Rights Act of 1964 set
the Nation on a course toward racial equality. As the economy surged,
income differences narrowed for a full decade. The sharp recessions of
the mid-1970s and early 1980s hit black and Hispanic Americans
particularly hard, however. And in the expansion of the 1980s,
economic growth was accompanied by sharp increases in overall income
inequality. As a result, despite the economic growth of this period,
income differences between black and Hispanic families on the one
hand, and non-Hispanic white families on the other, did not diminish.
The recession of the early 1990s brought further economic hardship, as
the poverty rate climbed to near a 30-year high.

Since 1993, incomes have once again been rising. But the present
recovery differs from those of the 1970s and 1980s in one important
respect: economic growth has not been accompanied by sharp increases
in income inequality. Moreover, this recovery has been accompanied by
a narrowing of some measures of racial inequality. The median black
family income reached a new high, and the poverty rate for blacks fell
to a new low. After nearly 20 years of stagnation, these developments
have again raised hope for sustained progress toward economic equality
among racial and ethnic groups.

This chapter reviews statistics on the differences in economic status
among racial and ethnic groups_whites, blacks, Hispanics, Asians, and
American Indians_and evaluates various explanations for those
differences (Box 4-1). Three themes are developed in this review.
First, although some narrowing of gaps in economic status among racial
and ethnic groups has occurred, it has been uneven_faster in some
periods and for some groups than others_and substantial differences
persist. The median incomes of non-Hispanic white families and of
Asian families are nearly double those of black and Hispanic families.
The median wealth of non-Hispanic white households is 10 times that of
blacks and Hispanics. Poverty rates among Hispanics and blacks are
more than triple those of non-Hispanic whites. Unemployment rates for
blacks are twice those for whites.

Second, the sources or causes of current differences in economic
status across racial and ethnic groups are numerous and complex. The
economic status of a person, a household, or a family reflects a
mixture of current conditions, such as the state of the economy, and
more permanent characteristics, such as educational background,
occupational experience, and family background, which have antecedents
in constraints faced in childhood and by previous generations. This
commingling of short-term and long-term influences poses a challenge
for the interpretation of trends in racial inequality. For example,
current progress toward racial equality is due both to the recent
effects of the strong economy and to longer term developments such as
improvements in educational attainment and reduced discrimination over
the past half-century. The complexity of these social and economic
processes cautions against a simple explanation of trends in racial
and ethnic economic equality.

A third theme of the chapter is that racial inequality and related
policy issues are intertwined with the long-term general increase in
economic inequality that extends beyond racial differences. Lack of
progress toward racial economic equality between the early to
mid-1970s and the early 1990s coincided with marked increases in
inequality both overall and within racial and ethnic groups.

Box 4-1._Racial and Ethnic Identity and Classification

-------------------------------------------------------------------------
The identification and classification of persons by race and ethnicity
are complex and controversial issues. The concepts of race and
ethnicity lack precise and universally accepted definitions. Their
economic significance depends on a variety of factors, including how
individuals identify themselves racially or ethnically, and how others
identify and treat them. Most of the data presented in this chapter
classify persons by race or ethnicity on the basis of responses to
questions about race and Hispanic origin in the decennial Census and
other household surveys.

Whenever possible, data for five mutually exclusive racial and ethnic
groups are presented in this chapter:

 Hispanics, who may be of any race
 Non-Hispanic whites
 Blacks not of Hispanic origin
 Asians, including Pacific Islanders, not of Hispanic origin
 American Indians, including Alaska Natives (Alaskan
Eskimos and Aleuts), not of Hispanic origin.

The term ``black'' rather than ``African American'' has been used in
government statistics for more than two decades. The tables, charts,
and references to statistics in this chapter that rely on these
classifications use the term ``black.''

Hispanic identification is determined by responses to a question about
Hispanic origin. Therefore, in tables, figures and discussion of
related statistics the term ``Hispanic'' is used.
-------------------------------------------------------------------------

The increase in income inequality has two major implications. First,
since blacks, Hispanics, and American Indians are disproportionately
represented at the bottom of the income distribution, they are
affected disproportionately by developments that make all those at the
bottom worse off relative to the middle or the top. A second and more
subtle implication is that inequality within racial and ethnic groups
has grown relative to inequality between such groups. Growing income
inequality within the previously largely impoverished black population
is partly a product of black economic progress: by some measures more
than half of black families have attained middle-class incomes or
higher. Despite persistent gaps in income between blacks and whites,
the growth of the black middle class, combined with widening
inequality within the white population and the general slowdown of
economic growth in the 1970s and 1980s, may have fueled opposition to
measures or programs perceived to benefit members of minority groups
without regard to individual economic circumstances.

-------------------------------------------------------------------------
Box 4-1._continued

The terms ``American Indian'' and ``Native American'' are often used
synonymously in speech and writing. In this chapter ``American Indian''
rather than ``Native American'' is used to avoid confusion caused by the
use in some Federal programs of the term ``Native American'' to include
Native Hawaiians and Pacific Islanders.

On October 30, 1997, the Office of Management and Budget announced its
decision to revise the standards for classifying Federal data on race
and ethnicity. The new standards recognize the growing diversity of
the American population by permitting respondents to mark more than
one race on survey questionnaires. In addition, the ``Asian or Pacific
Islander'' category has been divided into two categories, ``Asian'' and
``Native Hawaiian or Other Pacific Islander,'' making a total of five
racial categories (``Hispanic'' is an ethnic category). The ``black''
category has been changed to ``black or African American.'' The
ethnicity question will include two categories: ``Hispanic or Latino''
and ``Not Hispanic or Latino.'' Federal agencies will produce data on
the number of individuals who mark only one racial category, as well
as those who mark more than one.

Published statistics are not always available for all the groups
listed above. At times statistics are lacking because survey sample
sizes are too small to yield reliable estimates for small populations
such as American Indians or Asians. Specialized surveys or samples are
required to remedy this problem.
-------------------------------------------------------------------------

The chapter begins with a brief description of recent and projected
changes in the racial and ethnic composition of the population. The
most prominent of these changes are the increase in the proportion of
the population that is Asian or Hispanic and the decrease in the
proportion that is non-Hispanic white. The chapter then provides a
detailed description of differences among racial and ethnic groups in
traditional indicators of economic status: family income, poverty, and
wealth. The next two sections of the chapter review the evidence and
the economic literature in two arenas critical to the determination of
economic status: education and the labor market. The chapter ends with
a review of evidence of contemporary racial discrimination.

Although it is difficult to quantify the precise contribution of
contemporary acts of discrimination to the wide economic disparities
across racial and ethnic groups, there is substantial evidence that
such discrimination persists in many areas of the economy. Such
evidence highlights the need for racial reconciliation, as promoted in
the President's Initiative on Race as well as the President's
proposals to strengthen enforcement of the civil rights laws (Box
4-2).

POPULATION COMPOSITION

Since 1970 the percentage of the population that is non-Hispanic and
white has fallen substantially; the percentages that are Hispanic,
American Indian, and Asian (including Pacific Islanders) have risen
rapidly, and the percentage that is black has risen slowly (Table
4-1). The large increases in the Hispanic and Asian populations are
largely due to immigration and reflect changes in immigration laws,
especially the 1965 Immigration Act, which raised the ceiling on
admissions and ended the system of national origin quotas that had
restricted immigration from the developing world. The Immigration
Reform and Control Act of 1986, which legalized a large number of
immigrants, also contributed to these changes. Under the assumption
that these trends will continue, the non-Hispanic white population,
currently the majority, is projected to fall to about half of the
total population in the middle of the next century. (These projections
assume there will be no change in rates of intermarriage, although
these rates have been increasing.)

These national population changes mask differences across and within
regions. The geographic distribution of racial and ethnic groups is
important both because it influences the potential for social and
economic interaction among them, and because it affects their economic
fortunes. For example, over this century employment has shifted from
rural to urban areas and, within urban areas, from the central cities
to the suburbs.

Hispanics and American Indians are heavily concentrated in the West
and, to some extent, the South. Asians are concentrated in the West.
Within the South, Hispanics are concentrated in Florida, Texas,

-------------------------------------------------------------------------
Box 4-2._The President's Initiative on Race

On June 14, 1997, the President announced a new Initiative on Race.
The President envisions an America based on opportunity for all,
responsibility from all, and one community of all Americans. Race
relations remains an issue that too often divides our Nation. The
President's vision is to have a diverse, democratic community in which
all Americans respect and even celebrate their differences while
embracing the shared values that unite them. To reach this goal the
President has launched a national effort to deal openly and honestly
with our racial differences. The effort includes study, dialogue, and
action to address the continuing challenge of how to live and work
more productively together.
To further the goals of expanded opportunity and fairness for all
Americans, and in conjunction with the President's Initiative on Race,
the Vice President announced on January 19, 1998, in a Martin Luther
King, Jr. Day address at the Ebenezer Baptist Church in Atlanta, a
package of new civil rights enforcement initiatives. These proposed
initiatives place an emphasis on prevention and nonlitigation remedies
for discrimination, and on strengthening the ability of the Federal
civil rights agencies to enforce antidiscrimination law. The
Administration's plan increases resources for compliance reviews and
technical assistance, and offers alternatives to litigation by funding
expansion of alternative dispute resolution mechanisms. The plan would
set performance goals for the Equal Employment Opportunity Commission
to speed the processing of complaints and reduce case backlogs, and
would provide for better coordination across Federal agencies and
offices. The Administration's 1999 budget proposal contains $602
million for civil rights enforcement agencies and offices_an in-crease
of $86 million, or more than 16 percent, over 1998 funding.
-------------------------------------------------------------------------



Table 4-1.--Racial and Ethnic Composition of the U.S. Population
[Percent of population]
------------------------------------------------------------------------------
Year          American    Asian /1/     Black    Hispanic     Non-Hispanic
Indian                                             white
------------------------------------------------------------------------------
1970..............     0.4         0.7       10.9         4.5             83.5

1997 (estimated)..      .9         3.8       12.1        10.3             72.9

2050 (projected)..     1.1         8.7       13.6        23.8             52.8

1990 by region/2/:

Northeast.......      .2         2.5       10.3         7.4             79.4
Midwest.........      .5         1.3        9.5         2.9             85.8
South...........      .6         1.3       18.3         7.9             71.8
West............     1.6         7.3        5.1        19.1             66.7
------------------------------------------------------------------------------
/1/ Includes Pacific Islanders.
/2/ Detail may not add to 100 percent because data for the category
``other'' are not shown.

Source: Department of Commerce (Bureau of the Census).


and Washington, D.C. And despite massive outmigration over much of the
20th century, the majority of blacks continue to live in the South. In
fact, net black migration from the South to the North ended some time
in the 1960s.

There are also differences within regions in the racial and ethnic
distribution of populations. In 1990 Hispanics, Asians, and blacks
were much more likely than whites or American Indians to live in the
central cities of metropolitan areas. Hispanics, Asians, and whites
were much more likely than blacks or American Indians to live in the
parts of metropolitan areas outside the central city. Nearly half of
American Indians lived in rural areas; 37 percent lived on
reservations or other American Indian and Alaska Native areas.

ECONOMIC STATUS

FAMILY INCOME

Annual income is the most widely accepted indicator of current
economic status. This section reports incomes for families, where a
family is defined as two or more persons related by birth, marriage,
or adoption who reside together. In 1996 the median income of Asian
families was about $49,100, the highest among the groups considered in
this chapter. Asians are followed closely by non-Hispanic whites
($47,100) and, with a $20,000 gap, by blacks ($26,500) and Hispanics
($26,200; Chart 4-1). Because of the smaller size of the American
Indian population, reliable national data on their incomes are not
available for every year. However, according to the most recent data
(from the 1990 Census), American Indians had the lowest median family
income (and the highest poverty rate) of the five racial and ethnic
groups. With few exceptions these rankings have been stable over the
past 25 years.

Black and non-Hispanic white real median family incomes are somewhat
higher than they were 25 years ago, and Hispanic incomes are somewhat
lower. Since 1972, when data for Hispanics first became available on
an annual basis, real median family income has increased 14 percent
among non-Hispanic whites and 9 percent among blacks, but has fallen 9
percent among Hispanics.

As a result of faster income growth for non-Hispanic whites, the
Hispanic median family income has dropped sharply relative to
non-Hispanic white income over the past 25 years, and the relative
incomes of blacks has also dropped somewhat over the same period.
However, the Hispanic population has grown tremendously over this
period, primarily because of immigration. The relative decline in the
Hispanic median income reflects, at least in part, compositional
changes in the Hispanic population resulting from the immigration of
persons with relatively little education. The median incomes of both



black and Hispanic families are about 56 percent of the
non-Hispanic white median, lower than in 1972. Because these ratios
vary by a fair amount from year to year, it is difficult to identify
turning points precisely. But it is clear that, between the early to
mid-1970s and the early 1990s, black and Hispanic family incomes
declined relative to non-Hispanic white family incomes. Since 1993,
however, black family incomes have increased faster than those of
non-Hispanic white families.

Inequality Within Groups and the Growth of the Middle Class

Although a useful summary measure, median family income is an
incomplete indicator of the economic status of entire groups. For
example, trends in median income do not reveal the dramatic increases
in overall income inequality between the early 1970s and the early
1990s, nor do they speak to inequality within groups. Consideration of
other indicators of economic status may alter conclusions about the
nature of economic inequality among racial and ethnic groups. For
example, despite their higher median family income, the poverty rate
for Asians exceeds the rate for non-Hispanic whites by nearly 6
percentage points, indicating that this population is economically
heterogeneous.

Definitions of ``middle class'' are necessarily arbitrary. By one
indicator_household income between two and five times the poverty
line_a large middle class emerged among both blacks and whites between
1940 and 1970 (Charts 4-2 and 4-3). The poverty line used here to
adjust income corresponds to a 1960s' standard, since the poverty line
was developed in the early 1960s and reflects societal standards of
economic need at that time.



According to this measure, the white middle class expanded
considerably in each decade from 1940 to 1970, whereas the expansion
of the black middle class was greatest in the 1960s. Some scholars
have pointed to figures such as these as evidence of tremendous black
economic progress since 1940. However, that progress has not been
steady. Progress clearly slowed in the 1970s and 1980s. Furthermore,
although Chart 4-3 suggests that moderate growth of the black middle
class continued over the 1970s, annual data show little growth between
the early to mid-1970s and the early 1990s. In sum, a substantial
economic expansion of the black middle class between the 1940s and the
early 1970s was followed by 15 to 20 years of stagnation between the
mid-1970s and the early 1990s, with perhaps a resumption of growth in
the mid-1990s.



Since the early 1970s, income inequality has increased not only
overall but also within racial groups (Chart 4-4). However, only among
Hispanics has increased inequality taken the form of growth in the
proportions of both upper income and poor families at the expense of
the middle. Although both whites and Hispanics experienced declines
in the proportion of middle-income families, among whites there was
rapid growth in the proportion at the top, and a small decline in the
proportion at the bottom. The proportion of black families in the
middle- and upper income groups combined has changed little since the
mid-1970s, but by some measures there has been movement of families
from the middle of the income distribution to the top.

Poverty

Gaps in poverty rates between non-Hispanic whites and Asians on the
one hand, and blacks and Hispanics on the other, remain substantial
(Chart 4-5). However, the gaps in poverty rates between blacks and
whites have decreased since 1993, after remaining largely stagnant
from the mid-1970s to the early 1990s. In 1996 the black poverty rate
reached its lowest level ever, as did the difference in poverty rates
between blacks and whites. The decline in the black poverty rate in
the current recovery exceeds slightly the declines recorded in the
recoveries of the 1970s and 1980s. The poverty rate for Hispanics
fell slightly from 1993 to 1996, although it is still high, exceeding
the rate for blacks. The poverty rate for Asians has been flat since
1994.



Child Poverty

Differences across racial and ethnic groups in the prevalence of child
poverty not only indicate inequality in the current well-being of
children, but also represent differences in economic opportunity that
contribute to future inequality among adults and in subsequent
generations. Although child poverty is associated with health,
develop-mental, and educational disadvantages, the importance of low
family income per se as compared with parental education, family
structure, or other characteristics associated with poverty remains in
dispute (see Chapter 3).

Since 1993, child poverty rates have generally fallen, but they remain
too high, and differences in child poverty rates across racial and
ethnic groups are stark. Between 1993 and 1996 the poverty rate for
white children fell 1.5 percentage points to 16.3 percent. The rate
for black children fell even more, from 46.1 percent to 39.9 percent,
the lowest rate in more than 20 years but still very high. The rate
for Hispanic children fell marginally after 1993 and stood at 40.3
percent in 1996, higher than the rate for black children. The poverty
rate for Asian children rose 1.3 percentage points, to 19.5 percent,
between 1993 and 1996.

HOUSEHOLD WEALTH

Household wealth_the total value of a household's material and
financial assets, minus its liabilities_contributes to economic
well-being independently of income. Greater wealth allows a household
to maintain its standard of living when income falls because of job
loss, family changes such as divorce or widowhood, or retirement.
Financial wealth may also be particularly important in the presence of
borrowing constraints. For example, evidence that the receipt of an
inheritance increases entry into self-employment suggests that a lack
of personal financial capital limits small business ownership.

Wealth has been measured less frequently than income in government
statistics. There are two major Federal sources of data on household
wealth for the population: the Survey of Income and Program
Participation (SIPP) and the Survey of Consumer Finances (SCF).
Figures are not comparable across the two surveys for many reasons:
for example, the SCF and the SIPP employ different definitions of
``family'' and ``household.''

Measures of wealth show even greater disparities across racial and
ethnic groups than do measures of income. For example, according to
data from the 1993 SIPP, the median net worth of white households
($47,740) was over 10 times that of black or Hispanic households
($4,418 and $4,656, in 1993 dollars, respectively). Figures from the
1995 SCF are $73,900 for non-Hispanic whites and $16,500 for all other
groups combined (in 1995 dollars). Very substantial wealth gaps
between whites on the one hand and blacks and Hispanics on the other
are found even among families with similar incomes.

Differences in wealth result primarily from differences in lifetime
labor market compensation, differences in saving rates and the return
on those savings (including appreciation of the value of assets), and
differences in inheritances or other transfers from relatives.
Holdings among non-Hispanic whites in all major categories of wealth
exceed those of blacks and Hispanics. Three important components of
wealth for families are housing equity, holdings of stocks and mutual
funds, and private pension wealth.

Home Equity

The most important asset for most households is the equity in their
home. Differences in home equity arise from differences in
homeownership rates, in home values, and, among homes of a given
value, in the level of equity accumulated. Since 1993 there have been
increases in homeownership among all groups, but the homeownership
rate among non-Hispanic whites is more than 50 percent higher than
that of blacks or Hispanics.

Some evidence suggests that gaps among racial groups in home values,
although large, are narrowing. For example, between 1992 and 1995 the
median value of the primary residence was unchanged at about $92,000
for non-Hispanic whites but increased from $54,200 to $70,000 for all
other groups combined. In 1993 the median equity among homeowners was
about $50,000 for whites (in 1993 dollars), $29,000 for blacks, and
$36,000 for Hispanics. These values were $3,000 to $5,000 higher in
1993 than in 1991 (in 1993 dollars).

This Administration's efforts may have contributed to recent increases
in homeownership and home values among blacks and Hispanics. The
Administration has strengthened regulations under the Community
Reinvestment Act and has stepped up enforcement of fair lending laws.
Data collected under the Home Mortgage Disclosure Act show that,
between 1993 and 1996, conventional home mortgage lending to blacks
has increased 67 percent; such lending to Hispanics has increased 49
percent. These increases are much larger than the percentage increase
in conventional home mortgage lending overall in this period.

Discrimination in Mortgage Lending

There are a variety of possible explanations for differences in
homeownership rates among racial and ethnic groups. Research has
documented substantially higher denial rates in applications for home
mortgages among blacks and Hispanics than among whites. An analysis of
lending practices in Boston found that applications from blacks and
Hispanics were rejected about 28 percent of the time, compared with 10
percent for whites. However, applications from whites, blacks, and
Hispanics differed along many economic dimensions_including income,
loan-to-value ratios, and the presence of private mortgage insurance,
as well as other characteristics of properties and applicants_which
together explained about two-thirds of the difference in rejection
rates. Still, about one-third of the gap remained unexplained by these
factors.

The remaining gap has three possible explanations. The first is that
some relevant economic characteristics correlated with race are
observed by the lender but not by the analyst, and average differences
in those characteristics across racial and ethnic groups account for
the higher denial rate among minorities. However, the Boston study was
careful to incorporate extensive controls, including all factors that
lenders, underwriters, and others reported to be important in making
lending decisions. The second explanation is that the higher denial
rate reflects lenders' expectations of higher default rates among
minorities with similar qualifications and other characteristics. This
practice_rejecting applications on the basis of group
characteristics_is known as statistical discrimination and is illegal.
The third possible explanation, ``noneconomic'' or prejudice-based
discrimination, in which lenders discriminate against minorities and
lower their profits as a result, is also illegal.

The authors of the Boston study argue that no clear-cut evidence
exists of differences by race in default rates, after adjusting for
other characteristics of applicants and properties such as those
measured in the study. However, this argument and the study itself
have been challenged in subsequent studies, which claim to find
evidence of higher default rates among minorities. Other researchers
have argued in response that differences in default rates between
minorities and whites may not be a good indication of their
creditworthiness because, for example, whites might be treated more
favorably in foreclosure proceedings. As discussed in the concluding
section of this chapter, audit studies provide additional evidence of
discrimination in home mortgage lending, although continued research
is needed on the extent and nature of discrimination in this area.

Holdings of Major Financial Assets

Whites have higher rates of ownership of every kind of major
financial asset than do blacks or Hispanics, and among those holding
each kind of asset, holdings by whites are much more valuable. This is
not surprising given whites' greater median wealth. But some gaps are
particularly striking. For example, as of 1993 nearly 95 percent of
black households owned no stocks or mutual funds, and 95 percent
reported owning no private pension wealth (the corresponding figure
for whites is about 75 percent in each category). Differences in stock
ownership in 1993 are particularly important because between 1993 and
1997 the value of common stock appreciated enormously: for example,
the Standard and Poor's 500 index roughly doubled in value. Another
striking difference is in transaction accounts (such as checking
accounts), which are held by the vast majority (92 percent) of
non-Hispanic white families but by only 69 percent of all other racial
and ethnic groups combined.

THE ROLE OF FAMILY STRUCTURE IN INCOME AND POVERTY

Increases in family income and decreases in poverty rates for both
blacks and whites were rapid in the postwar period, especially in the
1960s. Blacks also made progress relative to whites in the 1960s. But
black family income was flat from the early to mid-1970s to the early
1990s, and the ratio of black to white family income generally fell
over this period. For example, since 1967 the ratio of black to white
average income for all families has fallen slightly, from 0.65 to
0.62. However, black-white ratios of income within family types have
increased, from 0.71 to 0.80 among married-couple families, and from
0.63 to 0.73 among female-headed families. (The overall ratio of
income is lower than the ratios among these subgroups because a larger
proportion of black families are female headed, a group with much
lower average income than other family types.) During this period the
shift toward female-headed families was faster for blacks than for
whites (Chart 4-6). Some observers have suggested that these
trends_particularly the rise of female-headed families_may largely
explain the persistence of differences in family income and poverty
rates among racial and ethnic groups. However, an adjustment for
changes in family structure since 1967 suggests that such changes
explain only about one-fifth of the income and poverty gaps between
blacks and whites observed today. Moreover, this adjustment may
overstate, perhaps greatly, the adverse effects of family structure on
income if those with lower income or lower expected income are less
likely to marry or to stay married.



The adjustment amounts to taking a weighted average in which the
average income or poverty rate specific to a racial group and family
type in 1996 is weighted by the corresponding percentage of families
of that racial group and family type in 1967. The adjustment shows
that if family structure for blacks and whites had not changed since
1967, in 1996 the black-white ratio of family income would have been
0.70 rather than 0.62, and the ratio of poverty rates would have been
2.6 rather than 3.0. Thus, these ratios indicate that roughly
one-fifth of both the income gap and the poverty gap in 1996 is
explained by changes in family structure after 1967. These are
surprisingly modest effects when one considers that since 1967 the
proportion of female-headed families increased from 28 percent to 47
percent among black families and from 9 percent to 14 percent among
white families. (Results are similar if the difference in family
incomes rather than their ratio is used to measure the income gap
between blacks and whites; differences in poverty rates rather than
ratios suggest a somewhat larger effect of family structure changes
since 1967 on the poverty gap. Also, similar adjustments demonstrate
that family structure can account for only a small portion of the
difference in income and poverty between Hispanics and non-Hispanic
whites.)

If the dramatic changes in family structure since the 1960s account
for only a modest portion of current income gaps among whites, blacks,
and Hispanics, what accounts for the remainder? Since the labor market
is the most important source of family income, a later section of this
chapter investigates gaps among racial and ethnic groups in labor
market outcomes such as earnings and employment. However, such
outcomes are linked to the skills that workers bring to the labor
market, many of which are developed prior to labor market entry. The
next section therefore discusses differences in education across
racial and ethnic groups.

EDUCATION

Education is one of the most powerful predictors of economic status.
Many dimensions of education are important, including the quality of
schooling, the quantity of schooling (often called ``attainment,'' for
example the number of years completed), and student achievement or
learning. The link between educational attainment and earnings has
been well established, in part because data on attainment have been
collected in the Census and in labor market surveys over a number of
years. There is less agreement on the measurement and economic
importance of other dimensions of education. Furthermore, the economic
importance of a college education has increased dramatically over the
past 20 years, as the relative demand for highly educated workers has
risen sharply. The focus of this section is on secondary and
postsecondary educational attainment. Of course, differences in later
educational attainment among racial and ethnic groups can result from
effects of discrimination and social and economic disadvantages
experienced in early childhood or in elementary education. (Chapter 3
discusses early childhood and elementary education.)

DIFFERENCES AND TRENDS IN EDUCATIONAL ATTAINMENT

Differences

Substantial gaps in educational attainment persist among racial and
ethnic groups. The most recent year for which comparable national data
are available for all groups discussed in this chapter is 1990. Asians
had the highest average attainment: in 1990, 40 percent of Asians 25
years and older had completed 4 or more years of college, compared
with 22 percent of whites, 11 percent of blacks, and about 9 percent
of Hispanics and American Indians. About 80 percent of whites and
Asians had at least completed high school, versus two-thirds of
American Indians and blacks and about half of Hispanics. For
Hispanics, attainment also varies considerably between immigrants and
the native-born. For example, Hispanic immigrants have much lower
rates of high school completion than native-born Hispanics. Asian
immigrants, on the other hand, have educational attainment similar to
that of their native-born counterparts.

Trends

To provide an indication of recent changes in educational attainment
across racial and ethnic groups, this section examines attainment for
younger persons (those aged 25-29 years).
High school. High school completion rates have increased steadily over
the 20th century. As educational attainment has increased, gaps in
high school completion among racial and ethnic groups have generally
narrowed, at least among the native-born. In 1967 the gap between
blacks and whites in high school completion rates was 20 percentage
points. This gap has narrowed considerably, but a 7-percentage-point
difference remains between blacks and non-Hispanic whites (Chart 4-7).
And although their high school completion rate has risen since the
early 1970s, Hispanics lag far behind and have not gained ground
relative to non-Hispanic whites. In interpreting these trends,
however, it is important to recall that the composition of the
Hispanic population has changed rapidly. The Hispanic population has
roughly doubled in size between 1980 and 1996, and the fraction that
is foreign-born has been growing. In fact, the slow progress in high
school attainment among Hispanics is in large part explained by the
increasing representation of immigrants with less education. For
example, between 1980 and 1990 the proportion of 18- to 21-year-old
dropouts (those who were neither enrolled in nor had completed high
school) fell from 30 percent to 23 percent among native-born
Hispanics, but remained at 47 percent for foreign-born Hispanics.
Still, as of 1990 a substantial gap in high school completion rates
remained between native-born Hispanics and non-Hispanic whites.

Postsecondary education. Educational attainment beyond high school has
increased dramatically for blacks, Hispanics, and whites over the past
30 years, although Hispanics have shown little increase in the 1980s
and 1990s. The percentage of non-Hispanic whites with a bachelor's
degree or higher is more than twice that of their black and Hispanic
counterparts. High school completion rates, college enrollment rates
among high school graduates, and college completion rates among
college enrollees combine to determine rates of college completion.
Some of the gaps in college completion rates reflect differences in
high school completion rates. For example, the gap between blacks and
Hispanics in completing 1 or more years of college is explained almost
entirely by lower high school completion rates among Hispanics. But
even among those who have completed high school, non-Hispanic whites
are more likely to enter and to complete college than blacks or
Hispanics. Again, Hispanics' low college attainment rates appear to be
due partly to low rates among immigrants: between 1980 and 1990 the
proportion of 18- to 24-year-olds enrolled in college increased from
18 percent to 28 percent among native-born Hispanics, but remained at
about 16 percent for foreign-born Hispanics.

Among women aged 25-29, college completion gaps widened between whites
on the one hand, and blacks and Hispanics on the other, over the
1980s. In fact, except among white women, there was



relatively little increase in college completion rates over the 1980s
for men or women of these ages (Charts 4-8 and 4-9). However, in the
1990s rates of college completion among black men and women began to
pick up, reflecting an increase in college enrollment rates of black
high school graduates in the mid-1980s. College completion also
increased among white men in the early 1990s.

EXPLAINING EDUCATIONAL ATTAINMENT GAPS

High school completion rates increased sharply in the postwar period.
Compared with the rather steady increase in high school completion,
college attendance and completion have fluctuated, especially for
males, although they have increased steadily since the mid-1980s.
Increases in college attainment have been attributed to two
developments. First, since the late 1970s growth in demand for highly
educated workers has raised the relative wages of college graduates.
Second, because educational attainment has generally increased over
time, the parents of recent high school graduates tend to be better
educated than the parents of high school graduates some years ago.
This is important because parents' and children's education levels are
highly correlated. Federal financial aid has also expanded
dramatically in the 1990s, doubling in real terms since 1993. This
expansion is expected to increase college enrollment and attainment
among low-income students, but it is too early to assess the magnitude
of this effect.

Levels

Most studies in the economics literature of gaps in college-level
educational attainment among racial and ethnic groups have focused on
college entry. Parental education and family income are important
determinants of gaps in college entry among racial and ethnic groups.
Both factors affect high school completion as well. For example, one
detailed recent study concluded that differences among blacks, whites,
and Hispanics in family background (primarily parental education and
income) can account for all the gaps in rates of high school
completion and college entry among racial and ethnic groups. The study
found that among young people with similar family income and parental
education, rates of college entry appear to be higher among blacks and
Hispanics than among whites. The importance of family background and
income differences is reduced when achievement test scores are
controlled for, but the interpretation of this finding is the subject
of great controversy. For example, low test scores result at least
partly from disadvantages relating to family background and may
therefore be a mechanism whereby such disadvantages are translated
into low educational attainment.





Trends

More attention has been paid to explaining differences in educational
attainment among racial and ethnic groups than to explaining their
trends. Large inflows of less educated immigrants have kept the
average educational attainment of Hispanics relatively flat. As noted
above, high school graduation rates have increased for native-born
Hispanics but continue to be much lower among immigrants. The
narrowing of differences in high school attainment between blacks and
whites over the past 30 years can be largely explained by increases
relative to whites in black parental educational attainment.

As high school completion gaps between blacks and whites were
decreasing steadily, differences in earnings between college and high
school graduates of all races were increasing markedly. Naturally,
attention has turned to explaining differences among racial and ethnic
groups in college enrollment and completion. College attendance among
high school graduates has increased for all groups. However, the
enrollment rate among recent graduates began to increase for whites
around 1980, about 5 years before the rate for blacks began to
increase. Therefore, the disparity in college enrollment rates widened
in the early 1980s and translated into wider differences in college
completion among racial and ethnic groups in the late 1980s or early
1990s (Charts 4-8 and 4-9).

One possible explanation of these differences is the increasing direct
costs of college. A recent study found that the schooling decisions of
blacks are more sensitive than those of whites with similar incomes to
tuition and other direct costs, perhaps because of lower wealth among
blacks than among whites with similar incomes. It also found that the
rise in the direct cost of higher education explains some, but no more
than one-third, of the lower propensity of blacks to enter college in
the 1980s. However, college tuition and other costs continued to
increase in the late 1980s, a time when black college enrollment began
to increase. The study concluded that the positive effects of rising
parental education appear to have more than offset the negative
effects of rising costs.

AFFIRMATIVE ACTION IN HIGHER EDUCATION ADMISSIONS

The term ``affirmative action'' encompasses a variety of activities and
programs, ranging from outreach and recruitment efforts to programs
that consider race as a factor in an evaluation process, which are
intended to increase minority representation in employment, education,
or contracting. Under the law, and as reflected in Department of
Education guidelines, colleges and universities may not establish
quotas for admission or set aside a certain number or percentage of
admissions slots based on race. However, they may consider race or
national origin as one factor in making admissions decisions, for the
purpose of remedying the effects of past discrimination or achieving a
diverse student body.

Affirmative action in admissions has been the subject of recent
contention. The Board of Regents of the University of California voted
in 1995 to prohibit universities within its system from considering
race in admissions. The California Civil Rights Initiative, known as
Proposition 209, prohibits the State from utilizing race- or
gender-based affirmative action programs in State employment, public
contracting, and education. In Texas et al. v. Hopwood the Court of
Appeals for the Fifth Circuit held that the admissions procedure used
by the University of Texas Law School in 1992 was unconstitutional.
However, this Administration strongly supports affirmative action in
higher education, and the practice remains widespread.

Such programs are intended to serve a variety of societal purposes,
including to remedy past or present discrimination, to secure the
educational benefits of a diverse campus community, to compensate for
educational or other disadvantages faced by promising applicants, to
prepare students for an increasingly diverse society, and to train
students to serve the needs of diverse communities. But what are the
more narrow economic effects of affirmative action in higher education
admissions?

A recent study found that black and Hispanic students are more likely
to be admitted to ``elite'' institutions of higher education (that is,
those with average Scholastic Aptitude Test, or SAT, scores in the top
20 percent of 4-year institutions) than non-Hispanic white or Asian
students with similar grade point averages (GPAs) and test scores. Of
course, in assessing student merit and making admissions decisions,
universities consider many criteria, such as letters of
recommendation, extracurricular activities, region of residence, and
adverse personal circumstances. The study also found no evidence of
differences by race, after controlling for test scores and grades, in
admissions to the less elite institutions where 80 percent of college
students are educated. Nonetheless, admission to elite institutions is
of interest because of the strong link between college selectivity and
later earnings.

Critics of affirmative action programs in higher education admissions
argue that some of the intended beneficiaries may actually be harmed
by such policies. (The same criticism could also be made of programs
for children of alumni or faculty.) They contend that affirmative
action programs impede the academic performance of minority students
and increase their college dropout rates by encouraging them to enter
colleges for which they may not be well prepared. However, the study
discussed above found little evidence of economic harm to these
students, as measured by graduation rates and earnings. The key
question this criticism raises is whether students admitted to elite
institutions because of affirmative action would have fared better had
they instead attended less selective institutions. In fact,  attending
an elite institution is associated with a lower college GPA, but a
higher graduation rate and higher earnings, for all students, after
controlling for SAT scores and high school GPA. The relationship
between college selectivity and both college completion and earnings
is similar for blacks and Hispanics and others.

The higher graduation rate among similar students attending more elite
institutions raises questions about which practices at elite
institutions increase graduation rates. Possibilities range from more
engaging professors or classes to better support services. It is also
possible that students expect a higher economic return to additional
investment in education at an elite college and are therefore more
highly motivated to obtain a degree.

The authors of the study argue that the number of applicants denied
admission because of affirmative action programs is small. But many
other students who are rejected may erroneously conclude that they
would have been admitted in the absence of such programs. As a result,
affirmative action in admissions may generate resentment far in excess
of its actual aggregate effects. Nonetheless, individuals denied
admission as a result of these policies may bear some costs_even if
those individuals are difficult to identify and are few in number.

As an alternative to race-conscious admissions policies, some have
called for ``color-blind'' policies that might target low parental
income or education. Blacks and Hispanics are, of course, a minority
of the population and account for a small minority of the population
of youths with high SAT scores. As a result, although blacks and
Hispanics are much more likely than whites to be poor, they make up a
relatively small share of the low-income population with the SAT
scores or GPA needed to gain admission to elite colleges. Therefore,
targeting low-income applicants alone would very likely result in a
dramatic reduction in minority representation at elite colleges.
Class-based, color-blind admissions standards would not yield
substantial numbers of blacks and Hispanics at most top-ranked
institutions at present. Some commentators have therefore concluded
that race-conscious admissions policies are needed to retain a
semblance of racial diversity on elite college campuses.

LABOR MARKETS

The largest share of most families' income is derived from earnings
from labor. Changes in labor markets can therefore have considerable
effects on economic inequality across racial groups. Differences in
labor market outcomes among racial and ethnic groups are intertwined
with general developments in labor markets. Among the most important
recent developments are technological changes that have increased the
demand for highly educated labor, growing immigration and
international trade, declining trade union membership, increased
participation of women in the labor market, and, most recently,
increases in the minimum wage and expansions of the earned income tax
credit. (See Chapter 7 for a discussion of the effects of
international trade on labor markets.) Developments that appear
race-neutral may nonetheless affect racial and ethnic groups
differently. For example, since Hispanics, on average, have much lower
educational attainment than whites and blacks, they are more likely to
be harmed by falling demand for less educated workers. However, lower
demand for less skilled workers would not necessarily be expected to
increase wage gaps among racial and ethnic groups for workers with
similar levels of education.

In analyzing changes in racial inequality in labor markets it is
important to bear in mind the growing economic diversity within racial
groups that began to be observed in the mid-1960s. For example, the
growing income inequality among blacks described above is mirrored in
the labor market, with college-educated professionals at one extreme
and labor force dropouts at the other. Although both groups face
substantial barriers in the labor market related to race, the nature
of these barriers could be quite different. The growing labor market
diversity within racial groups cautions against the search for a
single explanation for changes over time in differences among racial
groups.

Three periods mark changes in black-white inequality in the labor
market since 1960: a period of rapid progress from 1965 to the
mid-1970s; a period of stagnation or erosion of gains between the
mid-1970s and the early 1990s; and a period of mixed results since the
early 1990s. The beginnings and ends of these periods are difficult to
determine precisely because focusing on different data series and
different subgroups can yield somewhat different results.

TRENDS IN LABOR MARKET OUTCOMES

Unemployment and Employment Gaps

The current economic recovery has reduced unemployment substantially
for all groups. The overall unemployment rate has been below 6 percent
for over 3 years and has been at 5 percent or below since April 1997.
Improvement in the employment situation overall has been accompanied
by a reduction in the difference in the unemployment rate between
blacks and Hispanics on the one hand, and whites on the other. The
proportion of black women employed has risen above that for white
women in recent months. However, unemployment rates for blacks are
more than double those for whites and fluctuate more sharply over the
business cycle (Chart 4-10).

Men. In 1997 the unemployment rate for black men 20 years and older
was 8.4 percent, its lowest annual average since 1974. At 3.6 percent,
the white male unemployment rate for 1997 was also near a 20-year low.
Although the ratio of black unemployment to white unemployment is thus
more than 2 to 1, as it has been for many years, for the past 3 years
the difference in rates has been roughly 4 to 5 percentage points,
smaller than the gaps that prevailed from 1975 to 1994.  Among men
aged 25-54, the labor force participation rate for blacks is about 84
percent, about 9 percentage points lower than the rate for whites.
These rates have fallen in the past 25 years for both blacks and
whites, although the decline has been somewhat larger among blacks.



Women. Labor market outcomes for women are important for understanding
differences in economic well-being among racial groups, for two
reasons. First, women's earnings have historically made up a larger
proportion of two-parent family income among blacks than among whites.
Second, because of their much higher rate of single parenthood, black
families rely more heavily on female earnings than do white families.
For women aged 20 and above the most striking employment trend is a
long-term increase in labor force participation. Participation rates
for black women have long exceeded rates for white women, but the
difference has narrowed considerably and nearly disappeared by the
early 1990s. However, beginning in 1995, participation rates of black
women accelerated, reaching 64 percent in 1997. The rate for white
women appears to have reached a plateau at about 60 percent. But
because black women also have higher unemployment rates than white
women, their employment-to-population ratios are much more similar
than are their participation rates. Still, the black female
employment-to-population ratio surpassed the white ratio in 1996.
Labor force participation rates for Hispanic women are lower than
those for either blacks or whites.

Gaps (both ratios and differences) among racial groups in unemployment
rates for women are similar to those for men. The black-white
unemployment ratio for women has remained above 2, but the difference
has been somewhat smaller in the 1990s than in the 1980s.

Occupations

Like educational attainment, occupation is regarded as an indicator of
more permanent economic and social status than are wages or income in
a single year or month. Workers in different occupations are affected
differently by changes in the economy. For example, workers in
blue-collar occupations are more likely than white-collar workers to
be laid off in recessions.

Over the postwar period black men and women have both experienced
tremendous change in the occupations in which they work. Some of these
changes were experienced by all workers, black and white, but some are
specific to blacks, due, for example, to reduction in the most overt
forms of discrimination and to large migration flows out of the rural
South.

Women. In 1940, 60 percent of employed black women worked in domestic
service occupations, more than triple the percentage among all
employed women. The proportion of black women employed in domestic
service fell to 35 percent by 1960 and to 2 percent by 1996. Over the
same period, black (and white) women moved in large numbers into other
service occupations, as well as into clerical and sales jobs. The
proportion of black women in managerial and professional occupations
increased slowly between 1940 and 1960, then jumped in the 1960s and
1970s, reaching about 19 percent in 1980.

A major revision of the occupational classification system,
implemented after 1982, makes tracking changes over the entire 1980s
difficult. Since 1983 the fraction of black women employed in
managerial and professional occupations grew steadily, but increased
less than that of white women. As a result, the gap between white and
black women in the percentage working in managerial and professional
occupations widened by more than 2 percentage points over the past 15
years. Hispanic women are less likely than black or white women to be
employed in managerial and professional occupations, and more likely
to be employed in private household service and in the relatively low
skill blue-collar occupations of operators and fabricators.

Men. In 1940, 41 percent of black men worked as farmers or farm
laborers; that share had fallen to only 14 percent in 1960. (The
corresponding percentages for all men were 22 percent and 8 percent,
respectively.) By 1970 employed black men were more likely than other
employed men to work in blue-collar occupations (60 percent compared
with 48 percent). Black men were therefore concentrated in those
occupations that were the most affected by the severe cyclical
downturns of the 1970s and early 1980s, and at least until recently by
the long-term decline in manufacturing employment. By 1996 only about
45 percent of employed black men and 38 percent of all employed men
worked in blue-collar jobs.

In the period between 1960 and 1980 the percentages of black men and
black women who worked in professional and managerial occupations were
roughly equal, and both increased by about 10 percentage points. But
since 1980 black men have not moved into professional and managerial
occupations as rapidly as black or white women. In 1996 the share of
black men working in managerial and professional occupations stood 6
percentage points behind that of black women, 11 percentage points
behind that of white men, and 15 percentage points behind that of
white women.

Hispanic men are the least likely of all the groups considered here to
work in managerial and professional occupations. They are far more
likely than black or white men to work in farming and related
occupations, more likely than black men to work in precision
production (``craft'') occupations, and slightly less likely than black
men to work in the lower skill blue-collar occupations.

Earnings Gaps

Black-white earnings gaps. By all available measures, the wages of
blacks increased rapidly relative to those of whites in the 1960s and
early 1970s, but progress slowed or reversed between the mid-1970s and
the early 1990s (Charts 4-11 and 4-12). Trends in earnings inequality
among racial groups in the 1990s are less clear. Most wage series show
little progress in the pay of blacks relative to that of whites.
However, one wage series_median annual earnings for full-time,
year-round male workers_does show substantial recent progress among
black men relative to white men, with the black-white ratio reaching a
new high of about 0.8 in 1996. Firm conclusions about black-white pay
gaps for men in the 1990s are therefore difficult to reach.
Explanations for the narrowing of the pay gap in the 1960s, as well as
the widening between the mid-1970s and the early 1990s, are discussed
below. Researchers have just begun to examine the record of the 1990s.

Wage growth in the 1960s and early 1970s was faster for black women
than for black men, both relative to white women (Chart 4-12) and
relative to white men. Between 1967 and 1975 the gap in median wages
between white and black women fell from about 20 to about 5 percentage
points. Among younger women the differential disappeared, and there is
even evidence that young, college-educated black women were paid more
than comparable white women in the 1970s. But the earnings gap
increased starting in the mid-1970s and stood at about 17 percentage
points in 1997. Black and white women have both gained relative to
white men.





As noted above, whites on average have higher educational attainment
than blacks. But sizable pay gaps among racial and ethnic groups
remain for workers with similar educational attainment (Table 4-2). At
least until the 1990s, these trends in black-white pay gaps were more
pronounced for younger workers, who tend to bear the brunt of labor
market adjustment. For example, the pay gap between blacks and whites
narrowed most among young college graduates in the 1960s and early
1970s, and then widened most among this group after 1975.



TABLE 4-2.--Ratios of Black and Hispanic to White Median
Weekly Earnings, 1997
------------------------------------------------------------------------------
Black-white ratio                   Hispanic-white ratio

Sex             Workers with Workers with         Workers with Workers with
All workers high school  bachelor's   All      high school  bachelor's
diploma only degree only  workers diploma only degree only
------------------------------------------------------------------------------ Men. .   0.74          0.75       0.74       0.63         0.78          0.86
Women. .~  .83           .85        .90        .71          .86           .94~
------------------------------------------------------------------------------
Note--Data are for full time workers a~ged 25 and over.

Source: Department of Labor (Bureau of Labor Statistics).
~~

Earnings gaps for other groups. Less information is available about
differences in pay between whites and other minority groups. The pay
of Hispanic men and women fell relative to that of whites over the
1970s and 1980s. Much of the deterioration in the pay of Hispanics is
linked to educational differences and immigration. For example,
differences in pay between Hispanics and whites with similar
educational attainment are much smaller than the overall differences
(Table 4-2). In fact, a recent study reported that, between 1980 and
1990, differences in pay between whites and minorities living in the
same metropolitan areas, with comparable levels of schooling and
working in similar occupations, widened by 2.5 percentage points for
blacks and 4.1 percentage points for American Indians, but by less
than 1 percentage point for Hispanics and Asians.

EXPLAINING EARNINGS GAPS

Differences in pay among racial and ethnic groups can result from
differences in the average quantities of factors related to labor
market success, such as educational attainment, and from differences
in the ``prices'' of such factors, that is, their value in the labor
market. Differences among racial and ethnic groups in the prices these
factors command have been attributed to labor market discrimination.
But differences in the quantities of these factors may also reflect
discrimination outside the labor market or even within it. For
example, if blacks with higher educational attainment are
discriminated against in the labor market, their returns to investing
in education may be artificially reduced. Facing a lower return,
blacks may invest less in higher education.

Historically, blacks have received less schooling and attended schools
with larger class sizes and smaller budgets than those attended by
whites. Largely as a result of the 1954 Supreme Court decision in
Brown v. Board of Education, the Civil Rights Act of 1964, and the
1968 decision in Green v. County School Board, which required active
integration of schools, schools became increasingly integrated in the
late 1960s and early 1970s. Schooling gains can account for perhaps 20
percent of the gains in black workers' relative earnings in the 1960s
and early 1970s.

Other factors that explain trends in wage gaps among various racial
groups include migration (especially before the 1960s), regional and
industry demand conditions, macroeconomic shocks, and government
intervention. Government intervention to increase economic
opportunities for disadvantaged minorities has taken many forms,
including education and training programs, the enactment and
enforcement of civil rights and equal opportunity laws, requirements
(under Executive Order 11246) that Federal contractors engage in
affirmative action programs, and court-monitored affirmative action
programs intended to remedy past practices of discrimination.

Changes to the Mid-1970s

Between 1920 and 1990 blacks experienced two periods of rapid progress
relative to whites in the labor market: the first was during the
wartime economy of the 1940s, and the second was the period from 1965
to 1975. Migration from the South was substantial in the 1940s, 1950s,
and into the 1960s: 10 to 15 percent of all blacks and roughly 20 to
25 percent of young black men migrated in each of these decades. Wage
gaps between blacks and whites were much larger in the South than in
other regions. For example, in 1960 the black-white gap in wages was
about twice as large in the South (50 to 60 percent compared with 20
to 30 percent outside the South).

Following passage of the Civil Rights Act of 1964, the relative wages
of black workers increased sharply_more than can be explained by
macroeconomic factors such as growth in gross domestic product. The
improvement in relative wages was by far the greatest in the South,
where State fair-employment laws were weakest, where institutional
discrimination was greatest, and where Federal antidiscrimination
efforts were focused. Although there was some progress in the relative
earnings of blacks before 1964, the evidence is overwhelming that
progress accelerated substantially in the period from 1964 to 1975,
and that Federal attacks on racial exclusion in the South were
critical to this acceleration.

As noted above, gains in years of schooling and school quality explain
perhaps 20 percent of the gain in relative wages for blacks in this
period. There were large increases in the economic returns to
schooling for blacks. In principle, these could result from either
increased quality of schooling or decreased discrimination in the
labor market. However, decreased discrimination is the more compelling
explanation, since returns to education increased even among older
cohorts whose education had been completed prior to 1965.

But part of the improvement in schooling and school quality is also
attributable to Federal actions. The Supreme Court ruled in the Brown
decision that segregated schools are unconstitutional. Yet despite the
Brown decision and provisions of the Civil Rights Act that threatened
to cut off Federal aid to segregated schools, in the mid-1960s black
children in the South still overwhelmingly went to segregated schools.
The dramatic changes came after the 1968 and 1969 Supreme Court
decisions that required immediate integration. Therefore, improvements
in school quality that resulted from school desegregation do not
explain improvements in black wages in the South between 1965 and
1975.

Demand forces seem responsible for much of the improvement in relative
wages between 1964 and 1975. Partly because Federal actions coincided
with a strong economy, the precise role of Federal action, including
the associated voluntary compliance, has been difficult to establish
statistically. However, the observation that the most rapid progress
came in the South, where Federal efforts were concentrated, supports
the importance of the Federal role. Detailed studies show that blacks
moved into industries in the South from which they had previously been
excluded. For example, after 55 years of near-total exclusion, black
employment advanced rapidly in South Carolina's textile industry from
1965 to 1975.

A recent evaluation of the impact of the Equal Employment Opportunity
Act of 1972 confirms earlier findings of the importance of Federal
equal opportunity law to the labor market progress of blacks. The act
expanded civil rights coverage of Title VII of the Civil Rights Act to
employers with 15 to 24 employees (previously only larger
establishments were covered), as well as to State and local
governments. Blacks employed in the newly covered small establishments
in States where small employers were not already covered by State
fair-employment practice laws, largely in the South, were most
affected by this legal change. Blacks gained in relative employment,
earnings, and occupational status in small establishments in Southern
States after 1972.

Changes Since the Mid-1970s

Men. In the mid-1970s and 1980s, wages for less educated workers and
for black and Hispanic workers deteriorated. Wage differences between
blacks and whites grew fastest in the North Central region, where
employment and earnings declined more generally. On the demand side,
the heavy concentration of blacks in central-city manufacturing jobs
in the Midwest in the 1970s made them particularly vulnerable to
recessions and the decline of manufacturing employment. Ironically,
then, the movement out of the South and into manufacturing employment
that had contributed so much to black economic progress in the 1960s
and early 1970s also contributed to the deterioration of the late
1970s and 1980s.

Labor supply responses such as migration and training can help offset
the effects of reductions in labor demand. Lower mobility will produce
larger wage and employment declines in response to demand shocks.
There appears to have been slower adjustment out of declining areas
and industries among blacks and less educated workers, on average,
although it is unclear whether this supply adjustment was slower for
minorities than for whites with similar educational attainment.

Perhaps the most important change in the labor market over the past 25
years has been the increase in the demand for more educated workers.
But wage inequality has generally increased even for workers with the
same educational attainment. Although growing wage differences between
blacks and whites could be a symptom of increased discrimination, the
increase in general wage inequality makes this inference more
difficult. The increase in general wage inequality for workers of the
same age and educational attainment could lead to widening differences
in wages between blacks and whites, as the following example
illustrates. Suppose that in 1975 the median wage for black men aged
30 with a high school degree stood at the 35th percentile of the
distribution of wages for the corresponding group of white men.
Suppose further that wage inequality increased generally after 1975,
so that by 1990, wages at the 35th percentile of the white wage
distribution had fallen 10 percent relative to the white median (for
this group). Then, even if the black median wage remained at the 35th
percentile of the white wage distribution, the general growth of wage
inequality would have resulted in a 10-percent decline in the
black-white ratio of median wages.

Scholars have recently attempted to quantify these effects. Estimates
vary, however, regarding the extent to which the widening of pay gaps
between blacks and whites is accounted for by increasing general wage
inequality. One early study concluded that such effects could account
for the entire increase in black-white wage differences among young
workers in the 1980s. But this conclusion has been challenged. For
example, the increase in wage gaps between blacks and whites has been
greatest among young, college-educated workers. But the median wages
of black and white workers for this group were similar in the
mid-1970s. Therefore, a general decline at the bottom of the wage
distribution relative to the median cannot account for the fall of the
black median relative to the white median for this group. For other
groups of workers, however, increases in general wage inequality
appear to be more important.

Researchers have also hypothesized that the increase in general wage
inequality among workers of similar ages and education levels is due
to the growing value in the labor market of ``unmeasured skills''
(skills not measured by years of schooling or age). Some have
hypothesized further that the growth in wage differences between
blacks and whites is related to differences in unmeasured skills
between blacks and whites. For example, skills differences between
blacks and whites with the same years of schooling might result from
differences in the quality of the schools that blacks and whites
attend. Some studies have attempted to explore this issue by directly
examining school quality or measures of ``skill'' such as performance on
tests of cognitive achievement or ability. However, important aspects
of school quality may be difficult to measure. Studies find that
differences in test scores can explain a substantial portion of
black-white differences in wages in a given year, but have not been
able empirically to account for the reversal in black-white wage
convergence since the mid-1970s.

In addition, a recent study concludes that growing returns to
unmeasured skills are simply not large enough to account for the
stagnation of black economic progress after the mid-1970s. First,
changes in school quality cannot explain the widening of pay gaps over
time within cohorts whose schooling is of fixed quality over their
lifetimes. In principle, an increase in the labor market return to
school quality could lead to a widening of pay gaps between blacks and
whites even within cohorts, if blacks attended lower quality schools.
But second, the study found that even after differences in schooling,
age, location, and unmeasured skills are taken into account, young,
college-educated black men experienced at least a 13-percent drop in
wages relative to their white counterparts in the 1980s.

In sum, black men's earnings fell relative to those of white men of
similar age and educational attainment in the late 1970s and 1980s.
The evidence available indicates that increasing overall wage
inequality may have contributed to this deterioration and may be
linked to unmeasured skill differences, but these explanations are
incomplete. For example, this explanation does a poor job with young,
college-educated black men, for whom the erosion of relative pay was
substantial. These investigations therefore provide indirect evidence
that discrimination also contributed to widening pay gaps across
racial groups.

Women. Less attention has been paid to recent increases in the wage
gap between black and white women. Since the early 1970s, working
women have made substantial gains in earnings relative to men. The
narrowing of the gender pay gap has been attributed to greater
lifetime labor force participation among women and the dramatic
increase in the value of education and work force experience.

As noted above, black women reached virtual pay parity with white
women in the early 1970s, after a long period of steady improvement
(Chart 4-12). Since the mid-1970s, however, the wages of young black
women have fallen about 10 percentage points relative to those of
young white women. The relative decline was more rapid among young
college graduates. Chart 4-12 shows only ratios of weekly earnings of
full-time workers, but the trends in pay gaps among racial and ethnic
groups for women are similar in other data series (such as annual
earnings of full-time, year-round workers) and for workers of similar
ages and educational attainment.

Both labor force participation rates and attainment of a college
degree rose more for white women than for black women in the 1980s.
Over the 1980s the returns to education also increased. Changes in
demand for specific occupations and the decline in unionization rates
appear to have hurt black women relative to white women. Black women
were also more likely to be employed in declining industries than
white women.

Studies document a widening of pay gaps among racial groups for women
of similar ages and educational attainment. But since white women's
labor force participation rates have increased relative to those of
black women (at least until the mid-1990s), their labor market
experience at any age may also have increased relative to that of
black women. And pay tends to rise with greater labor market
experience. Thus, a possible yet unexplored explanation for the
decrease in the pay of black women relative to white women since the
mid-1970s is the increasing relative attachment of white women to the
labor force. Discrimination could also have contributed to the decline
in the black-white earnings ratio among women.

Affirmative Action in Employment

Aside from labor market changes that increased the demand for more
skilled labor, weaker enforcement of antidiscrimination laws during
the 1980s may have contributed to the decline in black workers'
relative earnings between the mid-1970s and the late 1980s. There is
evidence that enforcement of equal opportunity and affirmative action
laws has an effect on hiring decisions.

Affirmative action programs have proved controversial, but their
aggregate effects remain unclear. Because a variety of civil rights
and antidiscrimination measures were undertaken in a relatively short
time, it has been difficult to distinguish the effects of affirmative
action from those of broader civil rights enforcement. The Office of
Federal Contract Compliance Programs (OFCCP) is responsible for
monitoring the hiring and promotion practices of Federal contractors.
Large government contractors (those with 50 or more employees and
$50,000 or more in Federal contracts) must develop an affirmative
action program to remedy any underutilization of minorities and women
and must make good faith efforts to implement the program. One
approach to assessing the effects of affirmative action on employment,
therefore, is to compare government contractors (who are covered by
OFCCP enforcement) with firms that are not government contractors
(noncontractors). This approach, however, is subject to biases that
can lead it to overstate or understate the effects of affirmative
action plans. On the one hand, noncontractors often take steps to
ensure diversity and compliance with equal opportunity laws, even
though they are not covered by OFCCP rules. This would lead the method
to understate the effects of affirmative action. On the other hand,
increased employment at contractor firms could also result from a
shift of employment from noncontractors to contractors. In this case
the difference between contractor and noncontractor hiring could
overstate the employment effects of affirmative action.

According to these studies, active enforcement by OFCCP during the
1970s appears to be related to government contractors' increasing
their hiring of minority workers, although the effect is relatively
modest. For example, one study found that the employment share of
black males in contractor firms increased from 5.8 percent to 6.7
percent between 1974 and 1980. In noncontractor firms the share
increased from 5.3 percent to 5.9 percent. The literature also finds
that OFCCP had a significantly positive effect on the employment of
black females and a smaller but still positive effect on white
females.

A 1996 study concluded that, in contrast to findings for the 1970s,
there was no consistent evidence of the success of government
antidiscrimination efforts in the 1980s. As noted, in the 1980s OFCCP
enforcement was greatly weakened. Debarments of contractors found to
be noncompliant, awards of back pay to affected employees, and
conciliation agreements following violations all decreased during the
decade. Enforcement has apparently increased in the 1990s as new
initiatives have been adopted that focus enforcement on the worst
offenders, target areas of obvious noncompliance, and strengthen
sanctions.

DISCRIMINATION

No discussion of differences in economic status among racial and
ethnic groups would be complete without a consideration of the ongoing
importance of discrimination. Two statements appear to be true. First,
discrimination is far less pervasive and overt today than it was
before the Civil Rights Act of 1964. Second, audit studies and
significant judgments in favor of victims of discrimination make it
clear that discrimination against members of racial and ethnic
minority groups persists in many areas of the economy. However, there
is far less agreement about the degree to which current acts of
discrimination are responsible for differences in economic status
among racial and ethnic groups.

Many States' laws dictated a system of race-based classifications that
placed blacks at a disadvantage in the economy, in education, and
before the law. As late as the early 1960s overt racial discrimination
was common. For example, newspaper advertisements clearly stated
employer preferences for whites or blacks for specific jobs. The
practice was common even in States like New York, where
antidiscrimination legislation predated national civil rights
legislation.

Evidence of continued racial discrimination takes a variety of forms.
Perhaps the most convincing evidence comes from audit studies, in
which similar white and minority candidates are sent to the same
sources to seek jobs, rent apartments, or apply for loans for home
mortgages. For example, a white and a black job seeker may be given
similar rsums and sent to the same set of firms to apply for a job.
These studies typically find that employers are less likely to
interview or offer a job to minority applicants, and that minority
applicants are treated less favorably by real estate agents and
lenders and in some types of consumer purchases (such as automobiles
and meals in restaurants). For example, one national study found that
the incidence of unfavorable treatment in the housing market was 23 to
30 percentage points higher for a black or Hispanic auditor than for
his or her ``matched'' white counterpart. In the area of housing
discrimination the Department of Justice recently launched a national
program to test housing developments, seeking evidence of
discriminatory practices. Pairs of black and white persons are trained
to pose as prospective tenants and sent to ask about the availability
of units. In a case brought using evidence developed with this
technique, the Department of Justice obtained a consent decree against
housing providers in suburban Detroit that resulted in a $125,000
civil penalty paid to the Treasury and required the defendants to make
$225,000 available to the victims of their discrimination.

Various Federal agencies also receive and resolve thousands of
discrimination complaints each year. On the one hand, although a
settlement of charges does not always involve admission of
discriminatory practice, at a minimum the bringing of a charge
indicates the perception that discrimination has occurred. On the
other hand, only a portion of employees who experience discrimination
actually bring charges. In fiscal 1996 alone, the Equal Employment
Opportunity Commission, which is responsible for enforcing the
principal Federal statutes prohibiting employment discrimination
including Title VII of the Civil Rights Act of 1964, obtained $145
million in monetary benefits (excluding litigation awards) for parties
bringing discrimination charges, through settlement and conciliation.
From 1993 to 1997 the OFCCP conducted 19,852 compliance reviews and
3,192 complaint investigations and obtained over $158 million in
financial settlements, including over $60 million in back pay for
30,171 victims of employment discrimination by Federal contractors.
During the first term of this Administration, the Department of
Housing and Urban Development (HUD) reached out-of-court settlements
on 6,517 housing discrimination cases. HUD took enforcement action on
1,085 cases, either issuing housing discrimination charges or
referring cases to the Department of Justice. During this period HUD
obtained $17.8 million in compensation for victims of housing
discrimination. The Department of Justice settled major mortgage
lending discrimination suits in the 1990s, including suits against
large lenders in the Atlanta and Boston areas. In fiscal 1997 the
Department of Education's Office of Civil Rights received 1,422
complaints alleging discrimination based on race, color, or national
origin in access to equal educational opportunities. The office
facilitated a change in 249 of these cases.

Less direct evidence of discrimination comes from earnings comparisons
such as those described earlier in this chapter. As noted there, even
after adjusting for many characteristics that affect earnings, these
studies typically find that blacks are paid less than their white
counterparts. The traditional interpretation is that the unexplained
differential reflects discrimination in pay. However, these studies
are not uniformly accepted as providing evidence of discrimination in
the labor market: some researchers have argued that the studies fail
to control adequately for differences in average characteristics
between groups. Others argue that controlling for such characteristics
may not be appropriate if differences in characteristics such as
education and labor market experience are themselves partly the result
of discrimination both outside and within the labor market.

More direct evidence of labor market discrimination, in addition to
that from audit studies, comes from lawsuits that prove in a court of
law a pattern and practice of discriminatory behavior. But these
narrow, albeit powerful, pieces of evidence do not translate easily
into estimates of the aggregate economic impacts on employment or
economic well-being of discriminatory behavior. Significant analytical
challenges, requiring a combination of approaches, remain in assessing
the contribution of current acts of discrimination to current
differences in economic status among racial and ethnic groups. For
example, minorities who face discrimination by one employer may be
able to find employment with another, nondiscriminatory employer. (But
even in this case, discrimination imposes psychological costs and
additional job search costs on minorities.) This example also suggests
sthat, especially where discrimination is prevalent, reducing
discrimination can yield substantial economic benefits, by increasing
the number of nondiscriminatory employers.

It is an important goal of social and economic policy to ensure that
discrimination does not limit the economic opportunities available to
members of racial and ethnic minority groups. This Administration
remains committed to ensuring equal opportunity for all Americans.