Race and Ethnic Inequalities by Charles Hirschman and C. Matthew Snipp.

C H A P T E R

F I V

The State of the American Dream:
Race and Ethnic Socioeconomic Inequality in the United States, 970-90
Charles Hirschman and C. Matthew Snipp

merica is a nation of immigrants, but not all

immigrant groups have experienced the same reception and opportunities or have been accorded the same influence.
American ideals and cultural values were largely shaped by the Anglo herii.i^e of the founding settlers. For most of the nation’s history, those ideals huve continued to define the American experience for subsequent waves of iiinvals. The millions of new immigrants were expected-as were native
A mcrican Indians and African slaves-to assimilate into American society and i<> conform to Anglo-American ideals and values. They were also expected to
.-iliMard their ethnic heritage as quickly as possible. Economic incentives,
UK luding hope for their own and their children’s upward mobility, motivated tin immigrants’ acceptance of cultural change and adaptation.
Despite the massive immigration during the late nineteenth and early iwcutieth centuries, the image of American society as an extension ofEngiiuthors wish to acknowledge the support provided to the second author by the Vilas
*11 usi of the University of Wisconsin, support that was used for the processing and analysis of
^.ttiril.iia presented herein. Some of the data presented in this paper also appear in an unpub^tslitlr-d manuscript, "Assimilation in American Society: Occupational Achievement and Earnlor Ethnic Minorities in the United States, 1970 to 1990,’ by the same authors.

^l"

^

’^-ttiy.-.

’
89

lish society persisted throughout the first six decades of the twentieth century. As we near the end of the twentieth century and envision the future of
American society, the patterns in the first halfof the century-the exchange of cultural conformity for a chance at upward mobility, the American
Dream-seem obsolete. There appears to be less societal pressure on new immigrants to surrender their culture, language, and traditions. In fact, group identities, including race, ethnicity, and gender, now frame claims to political power and political participation. Sensing this shift in political roles, other groups, such as American Indians, have made a concerted effort to assert the importance of their ethnic ancestry. Many traditionalists see these trends as divisive forces, while others view the emphasis on cultural diversity as the defining character of contemporary American society.
It is important to assess the state of the American Dream as the twentieth century ends-to chart where we have been and to anticipate where we are going. In this chapter, we analyze patterns and trends in social and economic inequality among the major racial and ethnic groups in American society. We track occupational and earnings attainment among men from seven major racial and ethnic groups between 1970 and 1990. The years from 1970 to 1990 represent an especially important period in American history, given the government activism of the preceding decade. In the
1960s, federal and local governments enacted civil rights laws, created affirmative action procedures, and developed equal employment opportunity programs. For the first time in American history, public policy prohibited discrimination against Americans who did not fit Anglo standards of appearance, beliefs, and behavior. From the vantage point of the present, we look back over history to see if there has been progress in the uplifting of groups that historically have been outside the economic mainstream.
We find that there has indeed been progress in the reduction of socioeconomic inequality across race and ethnic groups over the twenty years from 1970 to 1990. With the exception of Japanese Americans, however, there remain wide socioeconomic gaps between minority populations and the majority. In 1989, African Americans, Latinos, and American Indians had an income gap from the majority population that was still around
$10,000 for working men-only slightly less than the gap in 1969.

Inequality in American Society

Race and ethnic differences in occupational and economic attainment raise three key issues. First is a concern about social justice. Ethnic
(and race) differences in life chances are at odds with values that are deeply ingrained in American society: (1) that every American should have an equal opportunity to compete for the wealth of the nation; and (2) the

AA

Tka

Mam Oamnaranhv nf Durable

Ineaualitv

distribution of resources should be determined by a fair and open process influenced primarily by personal merit and achievement. The fact that some groups, despite hard work and ingenuity, receive less than others raises questions about these values.
A second issue concerns the rigidity of stratification in America. In open societies, economic well-being is determined by effort and ability in a competition among equal actors (Lenski 1966). In rigidly stratified and closed societies, economic resources and personal well-being are disproportionately determined by ascribed characteristics (such as race or family background). The measurement of changes in the structure and determinants of inequality reveals whether American society is becoming more or less open.
Finally, the persistence of inequality is linked to the ideal of assimilation. he stratification literature in sociology (and especially studies of status attainment) has traditionally viewed a reduction in economic differentials i-s an important gauge of socioeconomic assimilation
(Hirschman 1983).
Fur example, recent data from the General Social Survey show that blackwhite differences persist although their magnitude lias declined (Grusky and DiPrete 1990). The presence of black-white differences in education, income, occupational status, or other measures of economic well-being arc pliina facie evidence that socioeconomic assimilation has not occurred.
The Origins and Development of Assimilation Theory

Burgess (1969, 360) defined assimilation as "a process of imerpenetration and fusion in which persons and groups are incorporated into a common cultural life." In subsequent work. Park’s race cycle
’.heoi-y described a cycle of race relations in which culturally distinct groups came into contact and eventually fused into a common culture and society.
The cycle included four distinct phases: "contact, competition, accommodation, and assimilation" (Park 1950, 150).
Park and his colleagues at the University of Chicago conducted studies in the 1920s when the city of Chicago was teeming with recent European immigrants. The process of assimilation Park described captured the experience of the European newcomers. Since Park, sociologists have tended
(u think about assimilation as an evolutionary process taking place in more
>.>r less discrete periods or phases.
Milton Gordon’s book Assimilation in American Life (1964) represents the next major milestone in assimilation theory. Gordon described seven different types of assimilation: cultural (acculturation), structural, marital
(amalgamation), identificational, attitude receptional (absence of prejudice), behavioral receptional (absence of discrimination), and civic
(absence ofvalue and power conflict). Gordon also posited that while tliese
Park and

different types of assimilation were connected to one another, one did not necessarily follow from the other. He noted, for example, that African
Americans (in the 1950s and early 1960s) had undergone cultural assimilation but had not experienced large-scale entrance into cliques, clubs, and institutions of the host society, that is, structural assimilation.
In the years since the publication of Gordon’s book, assimilation theorists have proposed a variety of schemes to describe the incorporation of ethnic minorities into the dominant society. For example, Greeley (1974) suggested that assimilation is neither linear nor unidirectional. And
Yinger (1985) contended that the subprocesses Gordon (1964) had identified could be found working independently of one another. These theories display an appreciation for the complexity of assimilation and imply that complete assimilation of some groups (e.g., blacks) is not likely in the foreseeable future.
Over the years, the ideal of assimilation and the race cycle theory have endured a hail of criticism. Some of these criticisms were anticipated, even. prior to the development of assimilation theory. Horace Kallen (1924) argued that immigrants should not be expected to surrender their culture and identity as a condition for participation in American society. Critics, in the 1960s, argued that ethnicity plays a central role in the lives of even the most acculturated groups (Glazer and Moynihan 1970). For many groups in American society, ethnicity is largely a symbolic construction and has a relatively minor role in their lives (Gans 1979; Waters 1990). Alba (1985), for example, wrote about the "twilight of ethnicity" among Americans of
Italian descent. More recently, the phenomenon of ethnic resurgence has challenged the race cycle tlieory and the idea that assimilation i.s inevitable. There has been a remarkable revival of ethnic awareness, even among groups whose cultural identities were once considered destined for extinction, such as American Indians (Cornell 1988; Nagel 1996).
Why does ethnicity seem to matter for some groups and not for others? Gordon’s theory suggests that prejudice and discrimination are critical. Acculturation is not sufficient for full participation in American society as long as "gatekeepers" continue to restrict access to neighborhoods, primary group associations, and opportunities for economic mobility. One prominent hypothesis is that "race"-physiological differences in skin color and other outward features-defines the essential difference between minority groups that are allowed to assimilate and those that are not (Cox 1948; Jordan 1974). Other perspectives suggest thai racism (the differential treatment of persons of different "races") is contingent on historical conditions and that American society holds the potential to assimilate persons of different races as well as of different ethnic groups (Myrdal 1944; Wilson 1978).

wl*** ^sfr^r-^l^l^ l*A^ll^l!^f

Assimilation Research

There is a large body of empirical researcli on various dimensions ut assimilation, including studies ofsocioeconomic assimilation, segregation in schools and housing, inter-marriage, and prejudice (Hirschman 1983).
Socioeconomic assimilation (as measured by occupational and earnings
-utainment) is widely considered an indicator of secondary-group structural
:is.similation (Gordon 1964; Lieberson 1980). Although it is not possible to present a comprehensive review of the empirical studies of ethnic assimilation in American society, there are some important themes in the literature on socioeconomic assimilation that inform and guide our study.
Our analysis compares Anglo and other "white" Americans willi
Anican Americans; American Indians; Chinese, Filipino, and Japanese
Americans; and Americans of Hispanic origin. European immigrants and
;heir descendants have generally enjoyed considerable upward mobility iiuuughout the twentieth century (Greeley 1978; Lieberson 1980). Tills u not to say that none were affected by bigotry and discrimination. On
(tic contrary, there is a long history of ethnic antagonism in this country fc-.g., anti-Semitic and anti-Catholic sentiment; see Baltzell 1964; Higham
1970). Nevertheless, the large socioeconomic differences that existed
-iinong these European ethnics in 1900 had virtually disappeared by tlie iy6Us (Duncan and Duncan 1968; Neidert and Parley 1985; Lieberson
-mcl Waters 1988).
Compared with immigrants from Europe, American Indians, African
.-iiiericans, and nonwhite immigrants from Asia and Latin America ti.-.’.e not fared as well. The American Indian population is small and
h..:-- experienced discrimination, forced isolation, and intense pressure sroin the federal government to assimilate into American society. Despite t’K.-se efforts, many (perhaps most) American Indians remain outside the mainstream (i.e., the modern, largely urban) U.S. economy (Hoxie h’S4; Fixico 1986; Gundlach and Roberts 1978; Snipp 1989; Snipp and
S-uidefur 1989). Racial discrimination and segregation have been major b;;i Tiers to the advancement of blacks in American society (Duncan
{"riy; Featherman and Hauser 1976; Parley and Alien 1987; Massey and

Demon 1993).
Hispanic and Asian groups warrant separate consideration, as they liave experienced explosive population growth since the immigration reforms w 1965 and 1986 (Chiswick and Sullivan 1995). The Latino population ip;ii licularly Mexican Americans) is an old and established American ethr-ic group. During the 1960s and 1970s they also comprised the largest and
Cisiest-growing immigrant group. It is difficult to make generalizations about the assimilation of Latinos because of their ethnic heterogeneity and

Rar^ anrl Ft-hnir

C/wirtflMM/vi<- lr*afr^li** Si .1*.

the disparate circumstances of their immigration, internal migration, and settlement patterns (Bean and Tienda 1987). Even within a small group like Cuban Americans, there are substantial differences with respect to the circumstances of their immigration and the extent to which they have become assimilated (Portes and Bach 1985).
Although there is considerable diversity in the Latino population, there are also commonalties such as language. However, compared with Latinos, the so-called Asian and Pacific Islander population has spectacular diversity. Except for the continent of origin, there are few visible similarities.
Many of these groups are exceedingly small in number, making it difficult to obtain reliable data on their socioeconomic characteristics. However, in the 1980s, Asian and Pacific Islanders were the fastest growing segment of the American population (Barringer, Gardner, and Levin 1993).
Asians, especially those ofJapanese and Chinese descent, pose an anomaly for assimilationist thinking about racial and ethnic inequality. Despite long histories of discrimination and persecution, Americans of Japanese and Chinese descent have, in fact, attained even higher levels of social and economic well-being than the white population. Only part of this success can be linked to their high levels of schooling and concentration in urban areas (Hirschman and Wong 1984; Nee and Sanders 1985). Some observers who attribute their success to hard work and ingenuity describe Asians as
"model minorities." But the evidence of the successful assimilation of all groups of Asians is mixed. Some recent Asian immigrants, especially those who were forced to flee in the aftermath of the Vietnam War, have had a much more difficult experience. Whether these disadvantages will persist into the future is an open question.
Finally, there is compelling evidence to suggest that there is growing socioeconomic diversity luithin each of these groups-African Americans,
Latinos, American Indians, and Asians. For example, studies have shown that in recent years, some segments of the African American population have enjoyed greater opportunities and attained a middle-class lifestyle, while many more are left behind in inner city ghettos (Landry 1987; Witson 1987; Parley and Alien 1987). The increasing socioeconomic inequality within minority group populations is a crucial issue (but is not addressed

directly in our subsequent analysis).
In this research, we measure the socioeconomic differences between groups and estimate how much of the interethnic gap can be "explained" by antecedent characteristics. Interethnic inequality is measured as the differences between the mean occupational and earnings attainment of each minority group (African Americans, American Indians, Chinese,
Japanese, Filipinos, and Latinos) and those of non-Hispanic whites (Hirschman 1980).

04

The Now riomooranhv

of Durable Ineoualitv

Data and Methods
Data
We use data from the Public Use Microdata Sample (PUMS) files t the decennial censuses of 1970, 1980, and 1990. These data permit us to examine recent trends in the socioeconomic achievements of seven racial and ethnic minorities. Census data have virtues and liabilities. Large samples are a primary virtue. They allow us to study relatively small populations that are usually not represented in national surveys. Our analyses include comparisons of the socioeconomic characteristics of whites, African Americans, Latinos, Japanese, Chinese, Filipinos, and American Indians. The primary liability of census data is the paucity of background variables dial might explain relative levels and trends in race and ethnic inequality.
We restrict the sample to men between the ages of twenty-five to sixtyfour who were working at the time of the census and received positive earnings the previous year. This sample does not completely represent these ethnic populations or even all adult men in these ethnic groups. But important theoretical and practical considerations led us to restrict the analysis to this subset of the population.
First, women’s roles, and particularly their participation in the economy, have changed dramatically since 1970. The question of gender stratification is intertwined with changes in racial and ethnic stratification in such complex ways that a full study of the topic would exceed the scope of this chapter. Second, the age range was dictated by the fact that we are most interested in persons who are economically active. Many persons younger than twenl.ytive years old are either still in school or just entering the workforce. Persons age sixty-five and older are likely to be retired or very close to retirement.
Third, we selected men who were employed in order to facilitate comparisons witli earlier studies of race and ethnic assimilation (Duncan 1969;
Hirschman and Wong 1984). Tliese restrictions limit our results to a select segment of the population, but other work leads us to believe that our estimates of ethnic differences will be conservative. In our prior study, we found that the restricted sample (only those in the labor force and with earnings last year) excluded 15 percent of white men in the working years, but excluded 25 percent of black men (Hirschman and Wong 1984, 589). Ethnic differences reported below would in all likelihood be larger if we included persons who were not active members of the workforce.

Variables

Our study focuses on ethnic differences in two indicators of economic well-being: occupational status and earnings, and changes in these indicators

Race and Ethnic Socioeconomic Inequality in the United States

95

Table 5.1.

Definition and measurement of demographic and socioeconomic characteristics

Ethnicity

Self-reported racial identification. Hispanic is self-identified in a separate
Item in the 1980 and 1990 censuses and based on a composite measure in the 1970 census. Whites are non-Hispanic whites; black Hispanics are coded black.

Age
Birthplace/length
of U.S. residence
Place of residence

Years of schooling
Occupational SEI
Sector

Weeks worked

last year

Hours worked last week

Earnings

Age at last birthday: 25-34; 35-44; 45-64.
A composite measure based on country of birth and place of residence five years ago. Coded native born; foreign born, in U.S. five years ago; foreign born, not in U.S. five years ago.
State or region of residence April 1, 1970, 1980, 1990. Coded for
California; New York; Hawaii; South, metropolitan area; South, nonmetropolitan area; rest of U.S., metropolitan; rest of U.S.. nonmetropolitan.
Number of years of formal schooling completed. Coded as 0-8; 9-11; 12;
13-15; 16 or more.

Duncan’s Socioeconomic Index, updated for 1980 and 1990.
Composite variable based on class of worker and industry classification.
Coded as self-employed; government employed; retail trade, not selfemployed; other periphery, not self-employed, not government employed, in retail trade; other core, not self-employed, not government employed.

not

Number of weeks worked in 1969, 1979, and 1989. Coded as less than

fifty weeks, fifty or more weeks.
Hours worked during the week before the census (April I). Coded as less than forty, forty, more than forty.
Total income received from wages and salaries, self-employment income from farm and nonfarm sources. Earners with zero or negative incomes were excluded from each sample.

from 1970 to 1990. We estimate a series of regression models that include age, education, residence, immigration status, and other variables as covariates. A complete list and description of these variables is presented in Table 5.1. Our measure of occupational status is the Duncan Socioeconomic Index (SEI), which has been updated to reflect changes in the occupational structures reflected in the 1980 and 1990 censuses (Duncan 1961; Hauser and Warren
1997). Earnings are defined as wage and salary income as well as income received from farm or nonfarm self-employment. We use the consumer price index to adjust for inflation and express income in constant 1989 dollars.
Many of our independent variables are conventionally scaled, such as education, which is measured in years of completed schooling. However, there are several others which merit clarification. We use self-reported responses to the "race" question on the census form to create ethnic categories. We added an additional category for Hispanic. In 1970, this item was a composite based on Spanish surname, Puerto Rican birthplace or parentage, and Spanish language. For 1980 and 1990, the Hispanic category was

06

The New Demograohv

of Durable Inequality

based on a self-identification question on Spanish or Hispanic origin. The white category is the residual and properly refers to non-Hispanic whites.
We combined the variables for place of birth and "place of residence 5 years ago" to construct a measure of immigrant status. Foreign-born persons who were not living in the United States five years before the decennial census (1965, 1975, and 1985) are presumed to be recent immigrants. Place of residence is measured with a classification that uses information on region, state, and metropolitan location. This coding was used previously by
Hirschman andWong (1984) in an attempt to identify geographical locations of ethnic concentrations and of greater and lesser economic opportunity.
The economic sector variable identifies those who are self-employed:i common means of economic adaptation for immigrants-and lliose in retail trade (a low-wage sector). Tlie balance are classified according to a widely used classification for core and periphery industries (Tolbert,
Horan, and Beck 1980). The measures for residence and sector are admittedly crude approximations, but we contend that they reflect differential access to opportunities in the American stratification system.
Any empirical analysis requires compromises based on the quality of data and measurement. We acknowledge that the Hispanic category covers a broad array of ethnic heritages, but inconsistent measurement across tlie three censuses and small numbers in particular groups do not permit disaggregation. An equally thorny problem concerns the difficulty of separating
(jut the impact of immigration and ethnicity. Major fractions of the Hispanic and Asian populations are foreign born whereas almost all blacks, American
Indians, and a majority of whites are native born. One solution would he to limit our analysis to native-born ethnic populations, but this would exclude significant shares (sometime a majority) of some ethnic groups.
Analytical Strategy

We present descriptive data in Table 5.2. Equations in Tables 5.3 and 5.4 model the economic well-being of ethnic minorities, estimating three sets of ethnic effects: (1) gross, (2) total, and (3) direct and indirect effects. The gross effect is the level of ethnic inequality between whites and ethnic minorities on either Duncan SEI points or 1990 dollars. The total vjfectis the level of ethnic inequality net of age and birthplace. The remaining effects are based on a detailed decomposition of the total effect ofeilinicity on attainment. Direct effects; measure the level of ethnic inequality on occupational attainment and earnings after the effects of the remaining covariates (e.g., residence, education, and industrial sector) are removed.
Indirect effects are the effects of ethnicity on occupational attainment and earnings that are mediated through the covariates in our models. We catRace and Ethnic Sncinecnnomic Incoualitv in the United States

07

Ttte 5.2. Indicators of socioeconomic attainment among men aged twenty-five to sixty-four by ethnicity: lS70.1980.and 1990

Mean SEI
1970
White
Black

Am. Ind.

Japanese
Chinese

Filipino
Hispanic

36.5
24.0
27.4
38.4
42.9
33.7
28.6

Mean Annual Earnings*

1980

1990

39.4
28.3
31.9
44.7
46.7
39.1
29.7

1970

40.1
30.2
31.2
47.4

$35,125
$20,436
$22,149
$35,429
$32,243
$25,044
$27,304

46.1

38.2
32.7

Standard

White
Black

Am. Ind.
Japanese
Chinese
Filipino
Hispanic

1980

N

1990

$35,278
$23,061
$25,925
$37,205
$31,424
$30.463
$23,797

dilate indirect effects by estimating successive regression equations and subtracting the ethnic coefficients with the intervening variable from the ethnic coefficients in the preceding equation without the intervening variable (Alwin and Hauser 1975).

1980

1970

$35,523
$23,172
$21,511

3,139

$42,750
$32,017
$28.945
$25,760

1.130

Deviation

Coefficient of Variation

Standard
Deviation

(SEI)

(SEI)

(earnings)

1990

---------V

2,726
939

911
611
1,394

3,063
2,540

3,013

2.272

2.585

1,647

3.037

1.859

2,977

1,379
2,231

3,029

’1 yST
,i F

2.482

of Variation

1980

1990

1970

1980

1990

1970

1980

1990

197C0

1980

1993

20.0
13.9
16.4
21.9
24.1
23.2
17.5

20.5
16.4
18.3
21.3
23.5
22.7
17.5

20.9
17.7
17.6
20.9
23.2
21.2
18.8

0.55
0.58
0.60
0.57

0.52
0.58
0.57

23,726
12,500
16,054
22,189

0.68
0.61
0.72
0.63
0.71
0.70
0.72

0.65
0.65
0.75
0.62
0.77
0.78
0.72

O.OT
0.76

0.50
0.58
0.59

23,042
14,898
19,334
23,245
24,307
23,807
17,163

31,477
18,127

0.56
0.69
0.61

0.52
0.59
0.56
0.44
0.50
0.55
0.57

22.959
17,473

19,561

33,909

30.035
26,332
23,803

earnings for each ethnic group for the years 1970, 1980, and 1990. The SEI and earnings ofJapanese workers and the SEI of the Chinese exceed those of the white sample. On the other hand, the earnings and occupational stains of Hispanics, blacks, American Indians, and, to a lesser degree, Filipinos are well below the earnings and occupational status of whites.
The results are remarkably stable. The rankings of groups on SEI and earnings are virtually unchanged between 1970 and 1990. The average SEI of all groups increased between 1970 and 1990. Most of the gains were in ilie four- to six-point range except for Japanese workers who gained nine points. There was little net change in the earnings hierarchy. However, the pattern is more mixed than that for occupational status. In constant dollars, white earnings were stagnant from 1970 to l990. Blacks, on the oilier hand, enjoyed modest gains in each decade since 1970. The same is true forjapanese workers. Other groups experienced modest gains in one decade and declines in another. This instability might be the result of compositional differences due to immigration, changing racial self-identification (in the case of American Indians), or reporting errors in the earnings data.

%

fc
W

(earnings)

1970

0.48

Table 5.2’ shows the means and standard deviations for SEI und

it:

|

2.841

--------1
Coefficient

17.293

|.

^1 ^

,)

0.80
0.75
o.^
0.91
0.92

^^,
8S--

’i

ii’;’.

fc
’’.1,’’

!-’^1’

W

Ifi’..

y
Ratio of SEI to White Mean

White
Black
Am. Ind.

Japanese
Chinese

Filipino
Hispanic

1970

1980

1990

100
66
75
105
118
92

100
72
81
113
119
99
75

100
75
78
118
115
95
82

78

Ratio of Mean
Earnings to
White Mean

White Minority

Gap (SEI)
1970

-12.5
-9.1

1.9
6.4

-2.8
-7.9

1980

-11.1
-7.5
5.3
7.3
-0.3
-9.7

1990

-9.9
-8.9

7.3
6.0

-1.9
-7.4

White Minority

Gap (earnings In

1970

1980

1990

100

100
65
73

100
65
61
120
90
81
73

58
63
101
92
71
78

105
89

86
67

is;

thousands of dollars;

1970

1980

19%;

^’1: r -$14.7
-$13.0
$0.3
-$2.9
-$10.1
-$7.8

-$12.2
-$9.4
$1.9
-$3.9
-$4.8
-$11.5

-$12:*;
-$IU;

s^
-$3-’.’;

t^.

-M d’

-^ ^

Am. Ind. American Indian; SEI socioeconomic index.
*Adjusted to constant 1989 dollars.
Sources: Public Use Microdaia Sample (PUMS) files of the 1970, 1980, and 1990 Census of Population, U.S. Bureau of;
Census.

Results: Ethnic Differences in Socioeconomic Attainments

Models of Ethnic Stratification, 1970-90
The patterns in Table 5.2 reveal the relative status (and change in siatus) of ethnic minorities in American society. They do not lake into account ethnic differences in education, place of residence, or other attributes that may affect socioeconomic attainment entirely apart from any consideration of ethnic relations. For example, some ethnic groups may have higher earnings because they are more heavily concentrated in higher paying urban labor markets, not because they receive different rewards for their work. We use regression analysis to decompose the ethnic effects on occupational and earnings achievements.

To present descriptive statistics, we selected representative sample ot whiles mid blacks in llie public use files and included all PUMS observations of tlie smaller populations sncli as
American Indians and Asian Americans.

98

The New Demoeraohv of Durable Inpnualitv

Occupational Attainment indirect The coefficients in Table 5.3 show the gross, total, direct, and
1990.
and
1980,
effects of ethnicity on occupational SEI for the years of 1970, comThe first panel shows the gmss differences in SEI for each ethnic group pared to white men. In spite of the overall upgrading of average occupational little change m ethstatus (reported in Table 5.2), there has been remarkably
The
average employed here. observed nic inequality over the twenty years white man in employed the average black man was thirteen SEI points behind
Indian and
American
of situation 1970 and ten points behind in 1990. The were both groups Hispanic men was similar to that of black men, although of three
All
men. white generally two to three SEI points closer to the status of occupations these established minority groups held substantially lower-status that than white men. In contrast, Asian American men were in occupations
FilWhereas
white men. were, on average, as good as or better than those of
Amerand Japanese ipino men held slightly lower statusjobs in 1990, Chinese men. white did than positions ican men worked at higher-status effects of age and the of net differences ethnic The second panel shows the gross effects. to identical almost are immigrant status. The total effects of men aged twenty-five to sixty-four
Table 5.3. Effects of ethnicity on occupational attainment points) SEI in the labor force 1970, 1980, and 1990 (average

Black

Japanese

Chinese

Filipino

Hispanic

-3

-8
-10
-7

-9
-8
-9

2
S

6
7

7

6

0
-2

2

-10

-10
-8
-9

5
8
7

-4
0
-1

-9
-10
-7

1970
1980
1990

0

-1

-1

-3
-2

-1

0

-1

-1
-3
-6

-4

-S

-1

Schooling

1970
1980
1990

-7
-6
-5

-6
-4
-5

6
7
10

5
6
7

2
5
7

-6
-7
-5

Sector

1970
1980
1990

0

0
0

0
0

0
0
0

0
0
0

0
0
0

0
0
0

-7

-3
-3
-3

-2
2
4

3
4

-3
-3
-3

-2
-2

Gross

Total

Indirect via:
Residence

Direct

SEI

American
Indian

1970
1980
1990

-13
-11

1970
1980
1990

-13

1970
1980
1990

-10

-6

-5

6
8

Because there are important ethnic differences in immigrant status
(Asians and Hispanics are more likely to be foreign born), one could conclude that there is little occupational handicap for immigrants.
The third panel presents indirect effects (through which ethnic differences are mediated) in residence, schooling, and industrial sector. The parameter estimates are dependent on the order in which the variables are entered in successive equations. Although the temporal sequence of these variables cannot be specified with any assurance, we have estimated three equations by adding each variable in a sequential and cumulative order.
First, place of current of residence is added to the model with ethnicity, age, and immigrant status as independent variables. In turn, the same exercise is repeated for years of schooling and sector-adding the variable to ilie prior equation and measuring the change in the ethnic coefficients.
Die final equation shows the "direct effects" of ethnicity that remain after all these other considerations are held constant.
Current place of residence has an unexpected role as a mediator of ethnic advantage/disadvantage in occupational attainment. There is a slight disadvantage for American Indians and Hispanics because they live in geographical areas with a lower average SEI. Tlie effect on the black-white differential is negligible. Residence appears to be a more serious handicap for Asian Americans, especially in 1990. Japanese, Filipino, and Chinese
Americans have an average occupation that is four to six SEI points lower in 1990 than they would have if their geographical distribution was the same as for the population as a whole.
Educational composition plays two quite different roles for the six ethnic minorities in our samples. For the three disadvantaged minorities
(blacks, American Indians, and Hispanics), having less education is the single most important reason for their lower occupational attainment. The absolute levels of occupational disadvantage associated witli education are four to seven SEI points, and this accounts for about half or more of their total SEI ethnic-disadvantaged status in 1990.
In contrast, Asian Americans have much higher average levels of education than the general population, and all other things being equal, this boosts tlieir occupational attainments. Indeed the occupational advantage associated with above-average levels of schooling for Asians has risen a bit over (lie
(wo decades, and there was about a seven-point SEI advantage for Chinese and Filipinos and a ten-point SEI advantage forJapanese Americans in 1990.
In spite of the great attention given to "sector" or industry in the sociological literature, this variable plays no role in mediating ethnic differences in occupational attainment for any of the ethnic groups across the three lime periods represented here.

socioeconomic index.

...^--nku nf niirahio Inoniiallru

Race and Ethnic Snrinernnnmir Inonnalit" i" <+*>

"!*’"< C-*"

The last panel in Table 5.3 shows the direct effects of ethnicity on occupational attainment after all the relevant variables from the census files are included in the equations. The coefficients are modest, at least relative to gross (or total) differences in occupational attainment. In 1990, holding all measured variables constant, the average employed black man held an occupation five SEI points lower than the average white man; American
Indian and Filipino men were three points lower; and Hispanics were one point lower. The comparable direct effect of being Japanese or Chinese was four SEI points above the white level.
Some interpret the net or direct effect as a measure of ethnic discrimination, but all we can really say is that the relationship is unexplained. Discrimination (e.g., negative and positive preferences within certain labor markets) may play a role, but other unmeasured factors, including family background and social networks, may also account for some of the differences. Moreover, discrimination may also be operating indirectly through the intervening variables (e.g., in funding for inner city or reservation schools, funding that affects the quantity and quality of schooling).

Earnings Attainment
Table 5.4 presents comparable results for earnings attainment.
The twenty years from 1970 to 1990 was a period of stagnant wages and rising inequality in the American economy (Karoly 1993; Levy 1995). How did the state of the economy play out in terms of wage differences among various subgroups of workers? The First panel of Table 5.4 shows that ethnic differentials in male earnings were largely unchanged over the same period. The black-white differential actually dropped a bit (from $14,700 in 1969 to $12,200 in 1979) and remained at $12,400 by 1989. American
Indian earnings fluctuated over the period, but their disadvantage relative to whites of almost $13,000 in 1969 grew to $14,000 by 1989. Unlike all other ethnic groups, Japanese men reached parity with the earnings of whites in 1969 and had pulled ahead (by $7,000) in 1989. Chinese, Filipino, and Hispanic men earned less than whites in 1969 and continued to do so in 1989, although the size of the differences peaked in 1979.
Holding constant age composition and immigrant status reveals the total effects of ethnicity on earnings in the second panel of Table 5.4. By and large, the pattern of ethnic advantage and disadvantage is similar to the gross effects. Black and American Indian men are the most disadvantagedabout $12,000 to $13,000 below the average for white men. Hispanic men are only slightly better off with a $9,300 earnings gap in 1989. Next are Filipinos and Chinese with deficits of $7,100 and $3,500 below white men.
Japanese Americans earned $7,600 more than white men in 1989.

Table 5.4. Effects of ethnicity on the earnings attainment of men aged twenty-five to sixty-four the labor force 1969, 1979, and 1989

in

Black

Gross

American
Indian

Japanese

Chinese

Filipino

Hispanic

$301
$1,930
$7,227

-$2,885
-$3,852
-$3,507

-$10,083
-$4,815
-$6,578

-$7,820
-$11,481
-$9,763

$547

-$1,858

-$8,232

-$7.398

$2,524

-$3,339
-$7,138

-$10,665
-$9,329

1969
1979
1989

-$14,691
-$12,216
-$12,351

-$12,965
-$9,351
-$14,012

1969
1989

-$14,664
-$11,934
-$12,062

-$12,772
-$9,017
-$13,326

$7,634

-$2.582
-$3.502

1969
1979
1989

-$115
$67
-$465

-$1,307
-$488
-$843

$2,466
-$266
-$3,911

$1,682
$12
-$2,451

$2,017
$5
-$3,398

-$1.307
$174
-$561

1969

-$3,476
-$2,571
-$3,777

$3,138
$3,758
$6,822

$2,689
$3,020
$5.092

$166
$2,726
$4,742

-$4,165
-$4,516

1989

-$4,084
-$3,360
-$3,276

Sector

1969
1979
1989

-$1,000
-$1,081
-$1,355

-$1,118
-$849
-$912

$456
-$145
-$327

$220
-$921
-$1,023

-$1,118
-$541
-$1,178

-$730
-$629
-$539

Occupation

1969
1979
1989

-$2,023
-$1,638
-$1,635

-$1,030
-$835
-$992

-$571
$601
$1,405

$348
$953
$1,361

-$959
-$798
-$877

-$763

1969
1979
1989

-$980

-$1,353
-$1,618

-$1,118
-$1,344
-$2,011

$20
-$46
$162

-$318
-$514
-$594

-$652
-$1.071
-$1.114

-$763
-$721
-$747

1969
1979
1989

-$9,036
-$6,342
-$3,713

-$5,793
-$8,183
-$4,791

-$2,723
$5,951
$3,484

-$10,151
-$10,055
-$5,887

-$7,329
-$8,183
-$5,313

-$8.506
-$6.051
-$3,543

Total

1979
Indirect via:
Residence

Schooling

1979

Weeks and

Hours
Worked
Direct

-$3,518

-$673
-$421

We decompose the total effects of each ethnic group on earnings (aclu;tlly the ethnic-white earnings gap) in the third panel. Place of residence plays an unexpected role for Asian Americans. In 1969, the geographical concentration of Japanese, Chinese, and Filipinos in Hawaii and California was a distinct economic advantage-about $1,700 to $2,500-in comparison to whites. By 1989, these locations were a liability, with Asians suffering a $2,500 to $3,900 deficit because of their concentrations in these same areas. The geographic distribution of black, American Indian, and
Hispanic men was a modest liability of several hundred dollars.
As with occupational attainment, education is the single most important variable mediating ethnic differentials in earnings. In 1989, about 25 to 33 percent of the economic gap between black, American Indian, and Hispanic

attainment men and white men can be explained by the lower educational contrast, In of these minority groups; the difference is about $3,500 dollars. and Chinese, the higher education (compared to white men) ofJapanese,
Filfor
reason for their higher earnings ($4,700 is the

major
Filipino men men). ipino men, $5,100 for Chinese men, and $6,800 forjapanese including variables, measured background the of
Holding constant all that sectors and in occupations found education, minorities were usually the of
The
sum men. white by those than occupied paid substantially less was sector and both via occupation on earnings indirect effects of ethnicity for about $3,000 for black men, $2,000 for American Indians, and $1,000 ecoan reveals men for
Filipino
Hispanic men. The job sorting process other nomic disadvantage comparable to American Indian men. On the occuhand, Chinese and Japanese do relatively well in finding well-paying a have men, pations. However, all groups, with the exception ofJapanese and week problem finding jobs that employ them for enough hours per American and black
For
men. white with the gap weeks per year to close cost Indian men, the problems of unemployment and underemployment the for economic penalty
The
men. white to them about $2,000 relative to $1,100.
$600
other groups is less, ranging from

after
The final panel of Table 5.4 shows the direct effects of ethnicity
The
magnimodels. in controlling for all of the covariates included prior of $5,000 tude of the deficits due to direct effects is substantial: in the range have There
1969.
in to $10,000 for all ethnic groups (except Japanese) the twenty-year deficit, for been changes, mainly in reducing the size of the
(with the groups minority all 1989,
However,
by period measured here: direct deficits relexception ofJapanese) still experienced very substantial ative to whites-in the range of $3,500 to $5,900. net effects, it may
Rather than speculate on the reasons for the residual and patterns of ethmagnitude overall the be more instructive to consider inequality. of earnings nic inequality. There are three ethnic "patterns"
These
and
Hispamcs.
Indians,
The first is represented by blacks, American per $10,000
(about
whites groups suffer substantial inequality relative to
33
to per25 About year) and there has been little moderation over time. levels, but the remainder is due cent of the gap is due to lower educational industry, weeks and to differences on labor market factors: occupation, second type or pattern hours worked, and other unmeasured factors. The of the deficit for these is illustrated by Chinese and Filipinos. The size and American groups is about half the size of that for blacks, Hispanics, are current disadvantage and
Filipino
Indians. The sources of the Chinese
Their
potenfactors. unmeasured and residence labor market positions, fact, In schooling. of levels tial disadvantage is reduced by their higher equal) else being (all tlieir educational advantage over whites generates

about a $5,000 gain. Without this educational "boost," their economic siinations would be similar to the level of blacks, American Indians, and Hispunics. Finally, the third pattern is represented by the Japanese. Japanese have higher incomes than whites in 1979 and 1989. In large part, this is due to their higher educational attainment, but they also experience an
"advantage" that is not measured by the variables in the model.

Conclusions
There was much talk in the 1990s about rffvmcdiscrimination, the problems of white men who are unable to find jobs or obtain promotions because of preferences given to minorities. These perceptions seem quite ai odds with the findings of this chapter, findings which show only modest
’-lianges in race and ethnic stratification among male workers over tlie two decades from 1970 to 1990.
However, we show that the patterns of race and ethnic inequality are
(Oinplex. Looking first at occupational attainment, black, American
Indian, and Hispanic men suffer about a ten-point SEI occupational deficit relative to whites, while Asian American men are equal to or above wliite men in their

occupational positions.
For the groups that are behind whites, tlie "problem" is primarily educational deficits. If black, American Indian, and Hispanic men had educational attainments equal to whites, tlie results liere suggest that they would only experience modest occupational disadvantages. There would still be a five-point SEI deficit for black men, but this would be half of tlieir curicnt handicap. The reason for the higher occupational attainment of Asian
American men is simply their educational level. If the Asian American men had the same education as white men, there would be only modest ethnic
)ccupational differences.
Turning to differences in income, earnings inequality is a much deeper problem for racial and ethnic minorities in America. All minorities, witli
(lie exception ofjapanese, earn less than whites. For blacks, American Indians, and Hispanics, the financial shortfall relative to whites is huge-about
510,000 per male worker for the period under study-and there lias been little sign of progress. Chinese and Filipino men are also behind, but the gap is somewhat less.
These results-tlie persistence of race and ethnic differentials in late twentieth-century America-challenge conventional theories about tlie declining role of ascribed factors in the American stratification system. There are three
"big" hypotheses in the stratification literature that need to be reviewed in light of these findings. The first is the "inheritance of poverty" thesis-some groups are disadvantaged because of their social origins. Without any mea-

sures of die characteristics of the families of origin, census data do not permit a direct test of this hypothesis. Although not the most important factor, some studies have demonstrated thatfamily-of-origin characteristics are a partial explanation of black-white socioeconomic inequality (Duncan 1969). It is plausible that differential social origins play an important role in the generation of socioeconomic differentials for other ethnic groups.
The second hypothesis is the relative role of qualifications versus discrimination in accounting for socioeconomic attainment. If the economy worked according to meritocratic precepts, then measures such as education and other personal attributes that make workers more productive would be the major determinants of socioeconomic success. Of course, the economy is not an impersonal machine; it is human beings who in fact recruit and promote workers. This leaves open the possibility that employers and supervisors hire and reward workers on the basis of invidious dis-

The economy of the latter half of the 1990s has been characterized by ising opportunity, with a record low unemployment rate. Will this reduce ihe kinds of persistent racial and ethnic inequalities we have documented tor the 1970s and 1980s? Perhaps. Tight labor markets make it more costly lor employers to make invidious distinctions among workers. But the slow pace of change during the 1970s and 1980s suggests that racial inequality i-< woven deeply into the fabric of American society.

tinctions, not only according to their qualifications. By measuring the an impact of education on occupation and earnings, it is possible to get factors. other versus of role "qualifications" approximate reading on the
Our results show that education plays a critical role in reproducing inequalattainity across race and ethnic lines. In other words, one path to economic is unexplained ment is educational attainment. But most ethnic inequality
Does
by education and other measures of the relative quality of workers. markets, labor this mean that discrimination is operating? Perhaps. In slack employers can select among many qualified applicants. It would be surto prising if gatekeepers ignored ethnic characteristics as irrelevant hiring barcolor of a is one discrimination of decisions. The traditional image avoid to is modern
The
style excluded. are simply where racial minorities be unreto more are prone "they" because certain minority groups

hiring

liable, cause discipline problems, or not fit in with other workers.
A third hypothesis in the stratification literature that should be reviewed is that ethnic inequality is generated by the structure of employing organizations. In addition to large public institutions and private corporations, and manthere are tens of thousands of small businesses whose ownership ethnic boundaries. agement is closely lied to geographical, kinship, and of enterconcentration a geographical not is simply The ethnic enclave for businesses among relationships informal prises, but loosely structured serve ties informal These trust. of the presumption whom ethnicity signals industries and firms to organize the recruitment of labor throughout many found in particare groups minority some in the economy, meaning that not particularly are data census ular sectors and not others. Although results suggest that our appropriate for examining this kind of hypothesis, the high behind the informal economy may be an important dimension market. levels of racial and ethnic inequality in the American labor

Race and Ethnic Socioeconomic Ineniialitv in the Unit-pd States

101