Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
00001.tif
(USC Thesis Other)
00001.tif
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600
ESSAYS ON
ECONOMIC STATUS, ECONOMIC INEQUALITY, AND VARIATIONS
IN SHORT TERM INCOME
by
Russell Walker Mangum III
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL OF
THE UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Economics)
December 1995
Copyright 1995 Russell Walker Mangum III
UMI Number: 9617116
Copyright 1995 by
Mangum, Russell Walker, III
All rights reserved.
UMI Microform 9617116
Copyright 1996, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
Russell Walker Mangum III
under the direction of h M Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean o f Graduate Studies
Date
DISSERTATION COMMITTEE
Chairperson
Acknowledgements
I extend my sincere respect and gratitude to Dr. Richard Easterlin, my
graduate advisor. It is only through his comments, guidance, and patience that
I have completed this dissertation.
I give a special thanks to Dr. Morton Schapiro, who, during my education
at USC, has provided invaluable support and advice, and has been a source of
inspiration in my academic pursuits.
I also thank Dr. Eileen Crimmins, Dr. Cheng Hsiao, and Dr. Andrew
Weiss for their helpful comments and professional guidance.
Ill
TABLE OF CONTENTS
Acknowledgements ............................................................................................. ii
List of Figures........................................................................................................... vii
List of Tables................................................................................................................ix
C hapter 1 : INTRODUCTION .......................................................................... 1
C hapter 2 : RACE DIFFERENCES IN RELATIVE ECONOM IC
IM PROVEM ENTS: Com paring Baby Boomers to T heir P a r e n t s 6
I. ABSTRACT ............................................................................................. 6
II. IN TRO D U CTIO N .......................................................................................... 7
A. Earnings Trends. Baby Boomers, and Race ....................................... 8
B. Economic Status Versus Individual Earnings ...................................... 10
C. Income Per Adult Equivalent, or I A E ................................................... 14
D. Causal Factors Behind the Trends in Family Composition and
Labor Force Participation ....................................................................... 15
Household Economics Models .......................................................... 15
The Becker or New Home Economics M o d e l.......................... 17
The Easterlin M o d e l.......................................................................20
E. Specific A im s ............................................................................................. 25
III. DATA AND M ETHODOLOGY..................................................................28
A. Data 28
B. M ethodology................................................................................................29
IV. IAE OF BABY BOOMERS AND THEIR P A R E N T S ..........................34
A. IAE Profiles ................................................................................................34
B. Comparing Baby Boomers With Their Parents: IAE ..........................36
C. Summary ................................................................................................38
V. EARNINGS OF BABY BOOMERS AND THEIR PARENTS 38
IV
A. Earnings P ro file s .........................................................................................38
B. Comparing Baby Boomers With Their Parents: E a rn in g s.................40
C. Summary ...............................................................................................40
VI. HOUSEHOLD DEMOGRAPHIC AND LABOR FORCE
CHARACTERISTICS ................................................................................. 41
A. Family-Type C om position.......................................................................42
B. Own-Children Per Household ................................................................ 45
C. Household Labor Force Participation ................................................. 46
D. Changes in Women’s Status in the Household ................................... 48
E. Summary .............................................................................................. 52
VII. IAE DECOMPOSITION ............................................................................ 53
VIII. C O N C L U SIO N ............................................................................................ 56
IX. APPENDIX TO CHAPTER 2 .................................................................... 74
CHAPTER 3 : TH E IM PACT OF INEQUALITY AND EQUIVALENCY
SCALE SPECIFICATION ON RELATIVE ECONOM IC
IM PROVEM ENTS: Baby Boomers and Their P a re n ts ..................................76
I. ABSTRACT ...............................................................................................76
II. IN TRO D U CTIO N ............................................................................................ 77
A. Increasing Inequality and Economic Im provem ent............................. 77
Explanations for Recent Increases in Economic Inequality . . . 80
Earnings Inequality............................................................................... 80
Between-Group Earnings Inequality.............................................81
Within-Group Earnings Inequality............................................... 83
Household Economic In eq u ality ....................................................... 86
IAE Inequality Among Baby B oom ers.............................................88
B. Alternative Equivalency Scale S pecifications.....................................89
Four Alternative Equivalency S c a le s............................................... 90
Fuchs’ Equivalency Scale .............................................................91
The Per Capita Equivalency S c a le ............................................... 93
The Official US Equivalency S c a le .............................................93
Parametric Equivalency S c a le s .....................................................95
III. METHODOLOGY 97
A. IAE Inequality.............................................................................................97
B. Application of Alternative Equivalency S c a le s ...................................99
IV. RESULTS .............................................................................................101
A. IAE Inequality........................................................................................... 101
IAE Gini C oefficients........................................................................101
IAE at the 25th and 75th Percentiles............................................. 102
Race Proportions at the Top and Bottom IAE D e c ile s 104
B. Stability of Results Under Alternative Equivalency Scale
Specifications .......................................................................................... 106
Trends in Alternative Measures of Equivalent In c o m e 107
Estimates of Gains in Equivalent Income Across Equivalency
Scales .............................................................................................109
Estimates of Equivalent Income Inequality Across
Equivalency S ca le s.............................................................................I l l
V. CONCLUSION .............................................................................................113
CHAPTER 4 : BACKGROUND CHARACTERISTICS AND CO LLEGE
EXPERIENCE AS DETERMINANTS OF INCOM E: A Sample of Elite
College G raduates ............................................................................................. 137
I. ABSTRACT .............................................................................................137
II. IN TRO D U CTIO N .......................................................................................... 138
III. DATA 146
A. Source ...............................................................................................146
B. Descriptive Statistics ...............................................................................148
IV. M ETH O DO LO GY ........................................................................................152
A. First Stage Estimation of Occupational C h o ic e ..................................152
B. Second Stage Income Estimation ..........................................................154
V. RESULTS 154
vi
A. Occupation Type Estimation .................................................................154
Personal Characteristics..................................................................... 155
Parent Characteristics ........................................................................156
Educational Experience Characteristics ........................................ 156
B. Income Equation Estimation by Instrumental Variables .................158
Personal Characteristics....................................... 159
Parent Characteristics ........................................................................161
Educational Experience Characteristics ........................................ 162
Occupational Characteristics............................................................. 163
C. Tests for Endogeneity of Occupational C h o ic e .................................165
VI. C O N C L U SIO N . . 166
VII. APPENDIX TO CHAPTER 4 ...................................................................178
CHAPTER 5 : SUMMARY ................................................................................180
CHAPTER 6 : BIBLIOGRAPHY 188
LIST OF FIGURES
vii
FIGURE 2.1 : SAMPLE INCOME PR O FIL E ....................................................59
FIGURE 2.2 : IAE: persons aged 15-64 by cohort, 1965-90 ....................... 60
FIGURE 2.3 : IAE: 1960-64 baby boomer and 1935-39
pre-boomer cohorts .......................................................................62
FIGURE 2.4 : IAE: 1955-59 baby boomer and 1930-34
pre-boomer cohorts .......................................................................63
FIGURE 2.5 : IAE: 1950-54 baby boomer and 1925-29
pre-boomer cohorts .......................................................................64
FIGURE 2.6 : IAE: 1945-49 baby boomer and 1920-24
pre-boomer cohorts .......................................................................65
FIGURE 2.7 : Earnings: full-time full-year working males
aged 15-64 by cohort, 1965-90 ................................................... 66
FIGURE 2.8 : Earnings: 1960-64 baby boomer and 1935-39
pre-boomer cohorts .......................................................................67
FIGURE 2.9 : Earnings: 1955-59 baby boomer and 1930-34
pre-boomer cohorts ................................................................ 68
FIGURE 2.10 : Earnings: 1950-54 baby boomer and 1925-29
pre-boomer cohorts ................................................................ 69
FIGURE 2.11 : Earnings: 1945-49 baby boomer and 1920-29
pre-boomer cohorts ................................................................ 70
FIGURE 2.12 : ACTUAL AND CONSTRAINED IAE: persons aged
15-64, 1965-90 ............................................................................. 73
FIGURE 3.1 : GINI COEFFICIENTS FOR IAE: persons aged
25-44 119
FIGURE 3.2 : 25th PERCENTILE IAE: persons aged 15-64
by cohort, 1965-90 ..................................................................... 121
FIGURE 3.3 : 25th PERCENTILE IAE: 1960-64 baby boomer
and 1935-39 pre-boomer cohorts ..............................................122
FIGURE 3.4 : 25th PERCENTILE IAE: 1955-59 baby boomer
and 1930-34 pre-boomer cohorts ........................................ 123
FIGURE 3.5 : 25th PERCENTILE IAE: 1950-54 baby boomer
and 1925-29 pre-boomer cohorts ........................................ 124
FIGURE 3.6 : 25th PERCENTILE IAE: 1945-49 baby boomer
and 1920-24 pre-boomer cohorts ........................................ 125
FIGURE 3.7 : 75th PERCENTILE IAE: persons aged 15-64 by
cohort, 1965-90 ........................................................................ 127
Vlll
FIGURE 3.8 : 75th PERCENTILE IAE: 1960-64 baby boomer
and 1935-39 pre-boomer cohorts ............................................. 128
FIGURE 3.9 : 75th PERCENTILE IAE: 1955-59 baby boomer
and 1930-34 pre-boomer cohorts ........................................ 129
FIGURE 3.10 : 75th PERCENTILE IAE: 1950-54 baby boomer
and 1925-29 pre-boomer cohorts ........................................ 130
FIGURE 3.11 : 75th PERCENTILE IAE: 1945-49 baby boomer
and 1920-24 pre-boomer cohorts ........................................ 131
FIGURE 3.12 : EQUIVALENT INCOME BY FOUR ALTERNATIVE
EQUIVALENCY SCALES: persons aged 25-44 ................ 133
ix
TABLE 2.1 :
TABLE 2.2 :
TABLE 2.3 :
TABLE 3.1 :
TABLE 3.2 :
TABLE 3.3 :
TABLE 3.4 :
TABLE 3.5 :
TABLE 3.6 :
TABLE 3.7 :
TABLE 3.8 :
TABLE 4.1 :
TABLE 4.2 :
TABLE 4.3 :
TABLE 4.4 :
LIST OF TABLES
IAE & EARNINGS OF BABY BOOMERS AND PARENTS,
Persons Aged 25-44 by Age G roup............................................ 61
HOUSEHOLD DEMOGRAPHIC AND LABOR FORCE
STATISTICS, Baby Boomers and Their Parents
Aged 25-34 71
IAE, EARNINGS, AND LABOR FORCE PARTICIPATION,
Baby Boomers and Their Parents Aged 25-44 ..................... 72
OFFICIAL U.S. POVERTY THRESHOLDS, 1992 ............ 117
ADULT EQUIVALENCY SCALE IMPLICIT IN OFFICIAL
U.S. POVERTY THRESHOLDS ............................................. 118
GINI COEFFICIENTS FOR IAE, Persons Aged
25-44 120
IAE LEVELS OF BABY BOOMERS AND THEIR
PARENTS, Persons aged 25-44 by Age G ro u p..................... 126
RACE PROPORTIONS AT HIGHEST AND LOWEST IAE
DECILES. Persons Aged 25-44 ............................................. 132
EQUIVALENT INCOME BY FOUR ALTERNATIVE
EQUIVALENCY SCALES, Persons Aged 25-44 ............. 134
EQUIVALENT INCOME BY FOUR ALTERNATIVE
EQUIVALENCY SCALES, Persons Aged 25-44
by Age Group .............................................................................135
GINI COEFFICIENTS FOR EQUIVALENT INCOME, FOUR
ALTERNATIVE EQUIVALENCY SCALES, Persons Aged
25-44 136
CONSORTIUM ON FINANCING HIGHER EDUCATION
MEMBER INSTITUTIONS........................................................170
DESCRIPTIVE STATISTICS ...................................................171
COEFFICIENT ESTIMATES, OCCUPATION
EQUATION .............................................................................172
COEFFICIENT ESTIMATES, INCOME EQUATION . . 176
ESSAYS ON
ECONOMIC STATUS, ECONOMIC INEQUALITY, AND VARIATIONS
IN SHORT TERM INCOME
Abstract
The first essay uses CPS data to analyze black-white differences in Income
Per Adult Equivalent (IAE) gains by baby boomers over pre-boomer “parent”
cohorts. Although median IAE of blacks was consistently lower than whites, both
black and white baby boomers have experienced substantial IAE gains, blacks to a
larger degree. Trends in earnings, household demographic factors, and household
labor force decisions exhibit salient race differences in sources of IAE gains. IAE
improvements by black baby boomers resulted more from earnings gains, while
white improvements came more from changes in demographic and labor force
characteristics.
The second essay explores the effects of increasing economic inequality,
by race, on IAE gains of baby boomers relative to their parents. Using CPS data,
analysis of baby boomer IAE across the distribution reveals that despite
increasing inequality over the period, substantial gains existed even at the 25th
and 75th percentiles. The relative proportions of blacks and whites at the upper
and lower IAE deciles were mostly unchanged between 1965-90, except a decline
in the proportion of blacks in the lowest decile. The reliability of IAE, one
measure of equivalent income, depends on the restrictiveness of its equivalency
1
scale assumption. The equivalency scale assumption is relaxed and three
competing specifications are alternatively employed. It is found that relative
levels and trends of equivalent income are robust to equivalency scale
specification, and only absolute levels are scale sensitive. These findings support
IAE as a measure for evaluating equivalent income.
The third essay investigates the effects of micro-level characteristics on
short-term income among graduates from highly selective and prestigious private
colleges and universities. A model is developed to allow for the endogeneity of
occupation to income. Using data from The Consortium On Financing Higher
Education, two-stage instrumental variables (2SIV) estimation shows the choice
of occupation is an endogenous event, determined by a wide range of personal and
family characteristics. Within the 2SIV framework, variations in short term
income are significantly influenced by various personal and family characteristics.
Failure to model occupation as endogenous furnished significantly different
results, both in the magnitude and the significance of various effects.
CHAPTER 1
1
INTRODUCTION
This dissertation is a compilation of three separate essays. The common
theme throughout the analyses is the examination of personal and
household/family characteristic effects on economic well-being. Two separate
micro-level data sets are used in the analyses, one provided by the Bureau of
the Census’ Current Population Survey (CPS), and another from The
Consortium On Financing Higher Education (COFHE).
The first essay, titled Race Differences in Relative Economic
Improvements: Comparing Baby Boomers to Their Parents, utilizes data
from the March income supplement of the Current Population Survey for
quinquennial years from 1965 to 1990. The aim in this analysis is to examine
race differences in the economic status of baby boomers relative to their
parents. Recent publications by The Congressional Budget Office (1993) and
Easterlin, Macunovich, and MacDonald (1993) show that baby boomers, not
distinguished by race, have fared quite well economically relative to their
parents. The analysis in this chapter examines race differences in the
economics status of baby boomers relative to the parents, and explores race
differences in the influence of household demographic and labor force factors.
Economic status is represented by Total Real Money Income per Adult
Equivalent (IAE), a household based measure of equivalent income. IAE is a
representation of total household income that has been adjusted for age-size
compositions of the household. This captures not only variations in resources
across households but also differences in household demands. Additionally,
this analysis isolates trends in earnings and household demographic and labor
market factors to identify race differences in the relative sources of changes in
economic status.
The second essay is titled The Importance of Inequality and
Equivalency Scale Specification on Relative Economic Improvements: Baby
Boomers and Their Parents. This chapter examines two issues: first, the
effect of increasing inequality on baby boomer gains in IAE relative to their
parents for both blacks and whites, and second, the stability of economic status
estimates across competing adult equivalency scale specifications.
Findings in the first essay lead to the conclusion that baby boomers,
across race, have made substantial gains in IAE over their parents at the same
age. However, these conclusions were based on median measures of economic
status, and may not be relevant across the IAE distribution. It is plausible that
a sufficient widening o f the IAE distribution between 1965 and 1990 could lead
to dramatic gains at the upper tail while preventing gains at the lower tail. It
may be the case that changes in IAE inequality between 1965 and 1990 have
worsened the relative economic position of certain baby boomers in the
distribution, resulting in smaller and larger proportions at the upper and lower
ends, respectively. In this case, the results of chapter 2 become less applicable
to the baby boomer population in general. This essay will examine whether
increases in IAE inequality have undermined previous median-based
conclusions of substantial gains by baby boomers over their parents for
individuals at lower ends of the IAE distribution.
To investigate the impact of changing IAE inequality on the economic
status of baby boomers, three separate procedures are employed. First, Gini
coefficients are calculated by year and race as an overall measure of IAE
inequality. Second, race specific levels of IAE for baby boomers are compared
to that of their parents at the 75th and 25th percentiles.1 These comparisons
establish relative IAE gains o f baby boomers at upper and lower tails o f the
IAE distribution. Third, the proportions of blacks and whites with IAE in the
top and bottom 10% of the overall distribution are determined for the years
1965 to 1990.
The first essay of this dissertation examines the economic status of baby
boomers by use of IAE, one measure of equivalent income. The concept of
'The main interest here is really at the lower ends of the distribution, but we
also examine the upper end for comparison.
equivalent income relates to the adjustment of total household income for the
effects of variations in age/size composition on household demand structure.
Any equivalent income measure requires the application of a specific
equivalency scale. An equivalency scale represents the particular structural
assumptions made about the connection between age/size composition and
household demand. Over the years many different equivalency scale
specifications have appeared in economic literature, but no single equivalency
scale has emerged as superior. Earlier estimates2 of IAE and IAE inequality
are reexamined by applying three alternative equivalency scale specifications.
This reexamination will identify instability of estimates resulting from narrow
equivalency scale assumptions.
The third essay, titled Background Characteristics and College
Experience as Determinants of Income: a Sample of Elite College
Graduates, is an econometric analysis which uses longitudinal micro-level data
on graduates from a set of highly selective and prestigious private colleges and
universities. The data are on personal and family background characteristics, as
well as information about education, work experience, and short-term earnings.
The individuals in the sample have survived a highly rigorous and
competitive selection process. All individuals in the sample have high levels of
2"Earlier estimates" refers to estimates in the first essay and the beginning of
the second essay.
education, which is arguably of the highest quality. Based on the relative
homogeneity and the "elite" nature of this group, two important issues are
explored. The first issue examined is the extent to which personal and family
background characteristics, even in this "elite" sample, are able to explain
variations in short-term income. The second issue relates to the endogeneity of
occupation. Occupation type is a variable commonly used in income
determination models, but it is most often treated as exogenous. This essay
examines the effect of treating occupation as endogenous by comparing
estimates from a two-stage instrumental variables technique to ordinary least
squares estimates. By allowing occupation to be endogenous, we allow many
personal and family background characteristics to affect income indirectly
through occupation, rather than affecting income directly. Results in the third
essay underline the importance of theoretical causal pathways in modelling
variations in income.
All three essays underline the importance of personal and
household/family characteristics on levels and trends in economic status. This
dissertation aims to present evidence in support of the connection between
decisions about education, occupation, family type, and number of children and
levels of economic status.
6
CHAPTER 2
RACE DIFFERENCES IN RELATIVE ECONOMIC IMPROVEMENTS:
Comparing Baby Boomers to Their Parents
I. ABSTRACT
This study uses Current Population Survey Data to analyze black-white
differences in household based equivalent income of baby boomer versus pre
boomer cohorts. The cohort comparisons are performed in a manner that
matches cohorts of baby boomers to cohorts representing their parents.
Although equivalent income o f black baby boomers was consistently lower than
that of white baby boomers, it is shown that both black and white baby
boomers have experienced levels of equivalent income substantially higher than
their pre-boom "parent" cohorts at the same age. Additionally, the percentage
gains in equivalent income realized by black baby boomers exceeded those of
white baby boomers. Inspection of trends in earnings outcomes, household
demographic factors, and household labor force decisions points to salient race
differences in the sources of increases in equivalent income. The economic
improvements of black baby boomers arose more from gains in earnings, while
white baby boomer improvements stemmed more from changes in demographic
factors and labor force decisions.
II. INTRODUCTION
Over the past few decades, the relative economic conditions of
individuals has been an increasingly popular topic in economic literature. The
issues investigated have included, but are not limited to, the growth in earnings
or income of individuals and household units, levels and trends in the
distribution of earnings or income, the role of the household in affecting
economic status, the connection between observed trends and the baby boom,
as well as race differences in each of these areas. This study adds another
topic to the existing literature by examining race differences in the economic
status of baby boomers relative to their parents. After establishing what the
race specific changes in baby boomer economic status have been, we then
explore race differences in the components that effect overall economic status.
Although it is no surprise to find that in 1990 the economic status of black
baby boomers falls short of white baby boomers, it is interesting to find that
gains in economic status of black baby boomers over their parents are larger
than gains by white baby boomers.
A. Earnings Trends, Baby Boomers, and Race
Between 1965 and 1990, average earnings trends in the U.S. have
exhibited periods of moderate, slow, and stagnant growth. The period up to
1973 was generally unremarkable to previous periods, but since 1973, growth
slowed to a halt, and at times even became negative (Blackburn, Bloom, and
Freeman, 1990; Levy and Michel, 1991; Levy and Mumane, 1992). Recently
proposed explanations for the falling off of earnings growth include demand
side effects such as decreased productivity growth and changing industrial
sector composition, as well as supply side effects such as changing
skill/education distributions'1 and relative cohort size (baby boom) effects
(Freeman, 1979; Easterlin, 1980; Dooley, 1984; Connoly, 1986; Berger, 1984,
1989; Blackburn, 1990; Levy and Michel, 1991; Bound and Johnson, 1992;
Levy and Mumane, 1992; Berman, Bound, and Griliches, 1994). These
findings have spumed many researchers to conclude that with regard to mean
individual earnings, current Americans are no better off (and possibly worse
off) than Americans in the early 1970’s. Indeed, recent research on earnings
levels has most commonly been tied to studies of earnings inequality, a subject
which is addressed in the next chapter.
3 Part of the skill/education "mismatch" could arguably come from changes in
industry sector demand.
Expanding on a supply side factor above, we note that the American
work force in 1990 was represented largely by baby boomers.4 It has been
asserted that baby boomers, whose relative size in the US population is
considerably larger than other population groups, face a dubious and unique
economic future. The general idea is that the relatively large baby boomer
cohort has aged into a relatively large supply of labor which, it is argued, has
led to decreased earnings levels for baby boomers. On the other hand, in a
period when changing skill/education distributions increasingly affected group
specific earnings growth, baby boomers were characterized by relatively higher
levels of education. Thus, although the effects of the baby boom have certainly
spread to the population in general, it is not clear what the specific effect has
been on baby boomers themselves. While most economic research into the
supply side effects of the baby boom relates to the effects of all individuals (or
all earners), this study focuses on the relative experiences of baby boomers
alone.
Along with decreases in earnings growth another concurrent trend has
emerged: a decreasing black-white earnings gap over the 1965-90 period. Most
of this decrease has been attributed to the 1st half of the period, and the black-
white gap in 1990 remained substantial (Freeman, 1976, 1990; Bound and
4 In 1990, labor force participants between the ages of 25-45 were baby
boomers.
Freeman, 1989; Smith, 1989; Bound and Freeman, 1989, 1991,; Card and
Kreuger, 1993; Card and Lemieux, 1994). The relative gains in black earnings
have been credited to several factors, including decreases in the black-white
completed education gap (Smith and Welch, 1986), increased school quality for
blacks (Card and Kreuger, 1992), and the effects of government anti-
discrimination and affirmative action efforts (Heckman, 1989; Donohue and
Heckman, 1991). Hence, the earnings experiences of black Americans have
been influenced by factors above and beyond those affecting all Americans.
B. Economic Status Versus Individual Earnings
The above issues relate to trends in individual earnings'. If the primary
interest lies in the level of a persons’ economic "status", other additional factors
are important. Indeed, individual earnings/income plays a vital role in
determining household economic living levels, but it fails to tell the whole
story. For instance, other things equal, a decrease in average household fertility
results in fewer demands on available income, more disposable income per
household member, and higher economic status. Similarly, an increase in
household labor force participation, other things equal, results in increased
disposable income per household member, and an increase in economic status.
individual earnings is a large (sometimes the sole) component of income, so
discussion here can loosely refer to earnings or income. Future references to
"earnings/income" reflects this fact and is not intended as a mathematical division.
Recent research has called for the use of the family or household as the
appropriate context for constructing measures of individual economic status.6
The critical point is that using individual earnings/income as a measure of
economic status ignores a wide range of salient intra-household dynamics that
partially determine a persons' true economic status. To expound this point, we
turn to a discussion of household resources and household demand.7
In an effort to capture intra-household effects on individual economic
status, we first consolidate all individual income into total household income.
This aggregate measure not only accounts for the levels of individuals’ earnings
(given they are working), but also accounts for decisions relating to household
labor force participation. We argue that total household income has an
advantage over individual income in that it better represents the total resources
available to the household.
In addition to examining the effects of household resources on individual
economic status, we also consider the demands on those resources. Envision a
household in which the level of resources is constant. This, of course, does not
6 See Lehrer and Nerlove, 1984; Easterlin, Macdonald, and Macunovich, 1990;
Macunovich and Easterlin, 1990; Hanratty and Blank, 1992; Karoly, 1992;
Burtless; 1993; Easterlin, Schaeffer, and Macunovich, 1993; Havemanand Buron,
1993; Mayer and Jencks, 1993.
7 The aim in this section is to incorporate the effects of household labor force
and demographic decisions. Thus, one can assume that prices and preferences are
not active causal factors.
imply constant economic status. Economic status in this scenario is not
constant because of potential variation in the demand on resources. Increased
(decreased) household demand could easily be introduced by the addition
(departure) of persons from the household.8 Holding resources constant, an
increase in the number of people in the household will cause those resources to
be "spread thin", thus decreasing the economic status of each person previously
in the household.9 Therefore, an appropriate measure of economic status will
account for differences in household resources and household demand.
Incidentally, the arrival (departure) of a child may alter household demand
differently than the arrival (departure) of an adult. If this is the case, then
tabulations of marginal household demand should be conditioned on age to
account for these differences. Put alternatively, household demand is
influenced by both household size and the adult/child ratio.
Another issue related to household demand comes into play here. Larger
households may benefit from efficiency gains through economies of scale. We
have discussed how household size and the adult/child ratio are important in
determining levels o f household demand. It is also probable that when an adult
8 This could come from births, deaths, doubling up, or separations/divorces.
These topics are discussed more formally in a subsequent section.
9 This does not imply that a person(s) entering the household would have lower
economic status than before. In fact, the person moving in would most likely be
better off than before; this relates to the impetus for the doubling up in the first
place.
is added to a household, the increase in household demand is less than a single
person; the same is true for children. This differential in demand comes from
inherent sharing in the household. Individuals, when in the same household,
share many products and services that otherwise go unshared in a one person
household.1 0 Individuals can, by living together in one household, decrease
average demand required to maintain some fixed level of economic status.
Hence, economies of scale are assumed to exist among adults and among
children.
If the level of household demand hinges on the specific household
composition, what determines household composition? The answer lies in a
series of household demographic and socio-economic decisions. Examples of
these decisions include but are not limited to: levels and timing of childbearing,
timing of and decisions on marriage, prevalence of doubling-up of families into
conglomerate households, and prevalence of unmarried couple unions. Recall
that with reference to household resources, another type of socio-economic
decision came into play: that relating to labor force participation. For the
purposes of this study, the determinants of economic status can be separated
l0Consider the capacity to share services such as public utilities and insurance,
and products such as furniture and appliances, not to mention residence space such
as the kitchen and living room. Average expenditures on these types of products
and services are almost always less than the analogous expenditures for a person
living alone.
14
into two categories: those related to labor market outcomes and those related to
household demographic and socio-economic decisions." 1 2
C. Income Per Adult Equivalent, or IAE
Given that a primary aim of this study is to explore the relative
economic well-being of baby boomers, an examination of individual
earnings/income alone is insufficient; in truth, it goes only part of the way. As
an alternative, the measure of economic status used in this study is Total Real
Money Income Per Adult Equivalent, or IAE. IAE was developed as a
measure of individual economic status derived from individual earnings/income
as well as the effects of household demographic and socio-economic decisions.
IAE is obtained by dividing the sum of individual income in the household by
the number of adult equivalents in the household, thus considering both
resources and demands." Scales used to calculate the number of adult
equivalents adjust for age differences in demand and economies of scale (see
the appendix to chapter 2 for a specific description of the scale used here).
"Factors that effect income other than earnings are beyond the scope of this
study.
"Specific variables used to capture these decisions are discussed in subsection
D.
"Individual income that has been adjusted to reflect the effects of household
composition is commonly referred to as "equivalent income". In this sense, IAE
is one form of equivalent income.
D. Causal Factors behind the Trends in Family Composition and Labor
Force Participation
One component of IAE, earnings levels, is clearly driven by economic
conditions and labor market outcomes. While the other components of IAE
(household composition and labor force participation) may seem less structured
and open to individual subjective response, they too are in fact influenced by
personal economic evaluations.1 4
This study examines three factors that are representative of the
demographic and socio-economic decisions discussed above. These factors are:
familv-type composition, own-children per household, and labor force
participation per household.1 5 Classification of family-type identifies persons
as being either childless, in a couple with children, or a single parent.
Household Economics Models
In the last three and a half decades, two competing theoretical economic
models have been developed to explain individual decisions and actions related
to the "house and home" (Gary Becker, 1960, and Richard Easterlin, 1966).
1 4 Where IAE is endogenously determined by household resources and
demands, the levels of resources and demands are themselves endogenous to other
factors. This endogeneity is discussed in subsequent sections.
1 5 Children per household could be considered a subtopic under the rubric of
family type composition.
16
Much of the motivation for the development of these models seems related to
the onset of the baby boom and subsequent baby bust, but the scope of these
theories is not limited to any single event or time frame. Before discussing
these theories, we identify players and actions in the model as they apply
here.1 6
Both the Becker model (commonly referred to as the New Home
Economics Model) and the Easterlin model are first and foremost
microeconomic models explaining certain decisions and actions of individuals.
The models are often categorized as "household" models because they address
decisions and actions seen as rooted in the family or home.1 7
One topic addressed by the models is household labor supply: whether or
not a household member will work and how much. Labor supply is not always
included as a household topic but one advance offered by the Becker and
Easterlin models is the labor supply dynamic brought about by the formation of
the "household". A second topic included in the Becker and Easterlin models
is that of marriage and divorce: if and when to marry, and given marriage, if
and when to divorce. This is quite clearly a "household" topic, as the issue at
1 6 Both the Becker and Easterlin models address issues and topics not
mentioned here. The models are addressed here only in as much as they relate to
the present topic. This is by no means a exposition or survey of either model.
l7This characterization is somewhat misleading as the models apply equally to
single individuals living alone.
hand in the decision is whether or not to form/dissolve a household. Still
another subject covered in the Becker and Easterlin models is childbearing: if
and when, and at what level.1 8 Below, the Becker and Easterlin Models will
be summarized in as much as they relate to family-type composition, own-
children per household, and household labor force participation.
The Becker or New Home Economics Model. Becker's model treats
children as a type of "good" that enters into parents’ demand function.1 9 As
with other goods and services, children have a related price; here the price of
children is represented best by the opportunity cost of the time needed to raise
and care for them. The time needed to raise and care for children has
traditionally been attributed to women,2 0 thus the price of children is proxied
by the wife’s wage (whatever that is or could be). Thus, an increase in the
wife’s wage has both an income effect, which tends to increase expenditure on
children and all other goods, and a children’s price effect (children now have
higher opportunity cost) which decreases expenditure in children relative to
other goods. It is generally accepted that this price effect is larger than the
lsThe "level" here can relates to the quality/quantity issue, but does not relate
directly to this study.
1 9 Becker considers "child services" the actual good. Increased expenditures on
children may or may not result in more children, just as increased expenditures on
vacations may or may not mean more vacations.
:oThis relates to household formation as a means of gaining efficiency through
specialization.
18
income effect, thus higher wife’s wages decreases expenditure on children. On
the other hand, an increase in the father’s wage (considered to have only an
income effect and not a price effect since he does not typically care for
children) causes an increase in consumption of all goods including children.
The Becker theory has been employed in explaining household behavior
that resulted in the baby boom. Butz and Ward (1977, 1979a) devised a
Becker type model to explain the baby boom. The causal force behind the
beginning of the baby boom was increased male earnings, which had a positive
income effect and led to increases in expenditures on children (as well as other
goods). The reversal of the baby boom, sometimes referred to as the baby
bust, was spumed by an exogenous increase in female wages which led to an
increase in female labor force participation.2 1 The assumption that the
negative price effect of increased womens’ wages on household fertility
outweighed the positive income effect insures that both increased female wages
and increased female labor force participation negatively affected expenditures
on children. Following the Becker model, these combined negative effects on
household fertility surpassed the positive male income effect and led to an
overall decrease in expenditures on children (relative to other goods). The
growth in female labor force participation and the relatively high' cost of
2‘Increases in labor force participation came from increases in the number of
hours worked from those already in the labor force, as well as the entry of new
labor force members.
19
children translate into other household-level behavioral changes, some of which
are discussed below.
Another integral component of Becker’s theory related to this study is
marriage and divorce. Becker sees marriage as rising out of efficiency gains in
household specialization; that is to say, women and men marry due to the
prospects of gains from specialization. Traditionally, men specialized in the
labor market and in acquiring resources, whereas women specialized in
childbirth, child raising, and household production. Any changes in the factors
that bring about this specialization will lead to new equilibrium in rates of
marriage and divorce.
Again, the Becker model can be applied to economic events occurring at
or about the time of the baby boom. The increases in female wages and labor
force participation which played a vital role in the reversal of the baby boom
also reduced the gains from marriage. As women became more prevalent in
the labor force, mens’ relative advantage in the labor market, the returns from
marriage, and the relative cost of divorce all decreased. This resulted in
delayed and decreased marriage and in increased divorce, both of which led to
increased childlessness and single parenthood.
The Becker model has offered explanations for several significant
economic and demographic trends over the past half century. Note however,
that the driving force behind increases in female labor force participation and
20
household fertility was the increase in the female wage. It induced more entry
by women into the labor force, it increased the relative price of children, and it
decreased the returns to marriage. But, for this explanation to be causal, the
increase in the female wage must be assumed exogenous. If the female wage
increase was a result of other exogenous factor shifts, then this application of
Becker's model is not causal but merely descriptive. The alternative model by
Richard Easterlin, discussed next, addresses the same issues addressed here but
does not restrict the female wage to be exogenous.
The Easterlin Model. Easterlin postulates different causal mechanisms
to the same events addressed above. Easterlin's explanations to the events fall
under his Relative Income Hypothesis. Under the Relative Income Hypothesis,
individuals form subjective economic aspirations based on their personal socio
economic history. Decisions on lifecycle events (such as marriage and
childbearing) depend on each persons economic status relative to their
aspirations. Stated alternatively, a person's behavior depends not on some
standard measure of their economic status, but on their perceived level of
economic status.
In Easterlin’s model, individuals marry when they feel they are
economically "ready"; the same goes for childbearing. Individuals who surpass
their material aspirations will marry as planned (if not sooner) and have higher
expenditures on children (as well as other goods). On the other hand,
individuals who fall short of their material aspirations will delay marriage and
childbearing until they feel ready. Women who perceive their economic status
to be lower than expected turn to alternative activities aimed at improving
economic status. In this case, some women would marry later and work in the
mean time, and women who married would delay and/or reduce childbearing;
both cases increase labor force participation.
Easterlin’s model applies directly to the onset and decline of the baby
boom. Individuals who were 20-30 years old in 1945 (those of primary
childbearing age at the beginning of the baby boom) were bom between 1915-
1925 and grew up during the Great Depression. Following Easterlin’s theory,
the economic aspirations of these individuals were formed during the
depression. By 1945, these individuals had come of age and had entered the
marriage/childbearing decision phase of their life in a time when the economic
conditions were far different. For individuals who were children during the
Great Depression, the post-WWII boom was a period of unprecedented
prosperity. Economic realities were far above economic aspirations; perceived
economic status was on average very high. The baby boom was the result of a
household level response of large increases in childbearing driven by higher
perceived economic status. This reasoning parallels some of the Becker
model, in that higher perceived economic status has a positive income effect.
22
The positive income effect results in higher expenditure on children as well as
other goods.
The baby bust is a sort of mirror image of the boom, statistically and
causally. Individuals who were 20-30 at the peak of the baby boom (around
1955) were the last to grow up during the depression. All persons bom after
this time did not have childhood experiences of the depression with which to
compare at the time of marriage/childbearing decisions. By 1964, when
completed fertility returned to levels equal or below those just before the baby
boom, individuals aged 20-30 were making marriage/childbearing decisions
based on material aspirations formed largely during the post-WWII boom; the
economic outlook was relatively disappointing. Additionally, the relatively
large baby boom cohorts had entered the work force depressing relative wages
and increasing relative unemployment. The household response was delayed
marriage, decreased childbearing, and increased female labor force
participation. The baby boom was over.
Easterlin’s model portrays the increase in female labor force participation
as occurring before the increase in female’s wages (which need not be assumed
exogenous). The female wage, which may have decreased slightly at first due
to the influx of low skilled female workers, soon grew considerably as women
gained experience and were further integrated into the labor force.
The Easterlin model offers alternative explanations to observed increases
in divorce and single parenthood. Becker points to decreased gains from
marriage as the cause for increased divorce and single parenthood, but Easterlin
cites the relative stress individuals faced as they fell short of their aspirations.
This increased stress related, among other things, to increased marital strains
and thus higher divorce.2 2 Increases in single parenthood are directly related
to trends in marriage and divorce, but have two distinct sources. Single
parenthood occurs either from out of wedlock births or from divorces of
families with children. Increased in the first source of single parenthood, out
of wedlock births, has resulted largely from delays in marriage and the
unrealized expectation of birth legitimation. Increases in the second source of
single parenthood comes directly from increases in divorce. In both cases, the
proximate determinant of single parenthood in the Easterlin model is unmet
material aspirations.
The major difference between the Becker and Easterlin models lies in the
assumed original causation. The beck model relies heavily on an exogenous
increase in the female wage. Easterlin’s model need not assume any exogenous
shift(s) in supply or demand factors to initiate the observed trends. The relative
2 2 0nce again, Easterlin’s model does not rely on exogenous womens’ wages
as a "jump start" to the entire process.
24
income hypothesis relies on subjective responses influenced by the relative
position in economic cycles.
Although the two models are distinctly different in their causal structure
and endogeneity assumptions, they also have similarities in structure and
predictions. For example, although Becker believes increased women’s wages
pulled up labor force participation and Easterlin believes increased labor force
participation of women eventually pushed up women’s wages, both agree on
endogenous shifts of female labor force participation. Also, even though
Becker assumes women’s wages increased exogenously and Easterlin believes
the increase is endogenous to material aspirations and cohort effects, both agree
that women leaving home and entering the work force relates to decreased
childbearing.2 ' Additionally, Both models point to increases in divorce and
single parenthood. Most importantly, both models clearly verify the existence
economically driven household level decisions. In both models, household
decisions are made in response to economic conditions in an attempt to
maximize utility. The resulting economic status that an individual attains is
therefore endogenous to perceived economic and the associated economic and
demographic responses.
2 ‘Both Becker and Easterlin cite other factors that initially decrease
childbearing. For Becker, it was increased cost o f children for Easterlin it was
unfulfilled economic aspirations.
25
How do these models relate to the objectives of this study, that is, how
do they relate to trends in economic status? Both models posit that the three
household decision factors to be examined, namely family-type composition,
own-children per household, and labor force participation, are economically
driven events. Decisions on these factors are made with respect to their
subjective economic condition. Exploration into race differences in economic
status should include an examination of race differences in these adjustment
"responses". Just as race differences in earnings trends shed light on race
differences in economic status, so too will race differences in demographic and
labor supply response effect race differences in economic status.
E. Specific Aims
This study differs from most other research on individual economic
status in two ways. First, the primary interest lies with the IAE of baby
boomers alone, rather than a larger population such as "all workers" or "all
Americans". Second, the measuring stick used to compare the IAE of baby
boomers is the IAE of pre-boomer "parents" at the same age. More
specifically, IAE levels of baby boomers is compared to IAE levels of pre
boomers that were determined to best represent their parents.2 4 This type of
2 4 The method by which this is accomplished is discussed in the data and
methods section.
26
comparison differs from other literature, where no ex ante explicit comparative
base is specified.
Three recent analyses have examined baby boomer economic status
relative to their parents using the household as context; these studies have
reported results contrary to common assumptions of a generally poor outlook
(Congressional Budget Office. 1993; Easterlin, Macdonald, and Macunovich,
1990, 1993). The CBO report found that baby boomers had higher inflation-
adjusted incomes than their parents at the same age, and that these gains were
even larger if incomes are adjusted for the reduction in household size. Both
works by Easterlin et al. found that median levels of economic status of baby
boomers exceeded that of their parents, part of the gains being made by way of
demographic adjustments." While both the CBO and Easterlin et al. report
extensively on various factors influencing economic status (e.g. savings rates,
home ownership, pension coverage and health care costs), neither fully explore
race differences in baby boomer economic attainment. In fact, both the CBO
and Easterlin et al. caution that conclusions as to economic gains by baby
boomers based on means or medians may not be indicative of individuals or
households at the lower end of the distribution.
2 5 As in this study, both the CBO and Easterlin et. al. actually make
comparisons to "parent" generations rather than micro-level parents.
27
These caveats contributed to the main motivation for this study. If
certain racial groups were over-represented in the lower end of the economic
status distribution, then a comparable analysis by race may provide more
insight into the relative economic condition of baby boomers by race. To this
authors knowledge, this study is the first to explore race differences in the
economic status of baby boomers relative to their "parent" generations.
In the sections that follow, we look into race differences in baby boomer
economic status in two ways. First, we examine median levels of IAE. This
will provide insight into how black and baby boomers have fared relative to
pre-boomers in parent cohorts. The levels of IAE are indicative of the sum
effect of both individual earnings levels and household behavioral decisions,
without regard to the relative strength of the two types of effects.2 6 Second,
we explore race differences in the relative sources of changes in IAE between
baby boomer and parent cohorts. By examining race specific trends in family-
type composition, own-children per household, and household labor force
participation, as well as race specific trends in earnings, we can identify not
just black-white differences in economic status relative to parent cohorts, but
also black-white differences in factors effecting economic status trends.
2 6 For instance, changes in the levels of IAE cannot offer any information on
the influence demographic adjustments on white baby boomers, or the influence
that earnings changes had on black baby boomers; IAE only relates to the overall
net level of economic status.
28
Section III describes the methods of analysis and the data used. Section
IV presents a set o f IAE profiles by race and outlines the use of these profiles
for comparison of baby boomer and pre-boomer parent cohorts by age. Section
V displays earnings profiles by race, and compares baby boomers to pre
boomer parent cohorts by age. Section VI reports on differences in household
composition and labor force participation statistics between baby boomer and
pre-boomer parent cohorts by age. Section VII discusses the empirical and
concludes.
III. DATA AND METHODS
A. Data
The data comes from the March income supplement o f the Current
Population Survey (CPS). This study utilizes quinquennial years from 1965 to
1990. The 1965 data were prepared by Mare and Winship (1973), and the
1970-1990 data come from public use tapes. The income measure used is pre
tax, post-transfer, money income, including public and private pension income,
public assistance, other welfare payments, alimony, and child support payments.
Income is price adjusted by the CPI-X1 index, and reported in 1988 dollars.2 7
2 7 For discussion, see U.S. Congressional Budget Office, 1988.
29
In the CPS, demographic questions refer to the time of the survey but income
questions refer to the previous year. Thus, income data from quinquennial
survey years 1965 to 1990 actually represent years 1964 through 1989 when the
individuals were 9 months younger.2 8
In two of the six income years (now referring to the period 1964-1989
rather than 1965-1990), the national unemployment rate fluctuated significantly;
four of the six years had unemployment rates between 5.2% and 6.2%, but in
1969 and 1984, the rate was 3.5% and 7.5%, respectively. Because of variation
in unemployment rates in 1969 and 1984, analysis of IAE and earnings profiles
rely mostly on the other survey years (1964, 1974, 1979, and 1989).2 9
B. Methodology
The sample includes individuals 15 to 64 years of age. This range was
selected to best capture individuals of working age. The sample was divided
into an exhaustive set of 5-year "synthetic" birth cohorts for the purpose of
inter-cohort comparisons by age.J°
2 8 Since the data is quinquennial and not every year, this difference between
age relating to income data and age relating to demographic data cannot be
reconciled. For simplicity, most of the remaining analysis will refer only to the
time span of 1965-1990 and only one age range for both income and demographic
data.
2 9 Appendix A includes the profiles using all years, but dropping these years
did not affect empirical conclusions.
3 0 A "true" 5-year birth cohort over time would consist of a longitudinal
30
With the data used. 11 mutually exclusive 5-year birth cohorts can be
identified; these include the cohorts of 1910-14 through 1960-64. O f these 11
cohorts, 7 are pre-boomer cohorts (1910 through 1940-44) and 4 are baby
boomer cohorts (1945-49 through 1960-64). Cohort profiles of two economic
measures are created: median IAE profiles and median earnings profiles. Each
profile portrays the trend in the level of an economic measure for a particular
cohort over time (figure 2.1). The level of each economic measure is estimated
by the median value of all individuals in the cohort for that year.3 1
As mentioned above, profiles are created for IAE and earnings. The
reason we examine earnings is to isolate one component of IAE, that which
represents changes in individual earnings capacity but not changes in labor
force or demographic factor decisions. Factors influencing labor market
outcomes for women over the 1964-89 period were complex. Womens’
position in the labor market has changed considerably, mostly in the form of
increased labor market participation. This type of change is a household based
supply shift and related almost exclusively to women of the period.3 2 Our
sampling of a set of individuals born in a certain 5-year time span. But, since
CPS data is not longitudinal, the same individuals cannot be followed across time.
But, given the sampling procedures of the CPS we can track a 5-year population
cohort via cross-sectional samples of that cohort over time.
3 1 The median was chosen over the mean due to its decreased sensitivity to
distributional changes within cohorts.
3 2 That is to say, there was no increasing trend in mens’ labor force
rational for examining earnings alone was to isolate economic changes not
associated with household level decisions. Therefore, earnings profiles are
constricted using only male workers.3 3
An underlying aim of this study is to determine how baby boomers, by
race, have fared compared to their parents when they were the same age."’ 4
To make this comparison, baby boom cohorts are compared to pre-boomer
cohorts born 25 years earlier, but at the same age.3 5 Comparing 5-year
cohorts bom 25 years apart gives the best probability of comparing children to
parents.3 6 So, pre-boomer cohorts that are 25 years older than baby boomer
cohorts are referred as "parent" cohorts.
participation over this period.
J'W e alternatively consider all male workers and then only full-time full-year
male workers.
3 4 Due to sample size limitations, only persons of black and white race are
included.
3 5 To compare the baby boom cohort o f 1960-64 to the pre-boom cohort o f
1935-39 when both cohorts were 25-29 years o f age, we look at 1990 data for the
baby boomers and 1965 data for the pre-boomers.
3 6 Vital Statistics of the U.S. show that in 1965, the average age of mothers at
birth was 26 years and that husbands were 2.2 years older than wives. This
indicates that for people born in 1965, average parents age was about 27.4 years.
So, when using quinquennial data as in this study, assuming either a 25 or 30 year
spread is equally accurate. The data did not allow a 30 year spread so the choice
was clear. Previous research utilizing the 25 year spread include: Easterlin,
Macdonald, an Macunovich, 1990; Macunovich and Easterlin, 1990; Easterlin,
Schaeffer, and Macunovich, 1993; and U.S. Congressional Budget Office (1993).
32
Three variables representing household labor force and demographic
decisions are examined in this analysis:
1) Population proportions of each of three family types: a) Childless. b)
Couples With Own-Children, and c) Single Parentsd1
2) Own-Family Children Under 16 per household.3 8
3) Labor Force Participants per household.
These particular variables were chosen due to their integral relationship to IAE.
The family-type proportions relate to population composition by earnings
capacity and by household demand (e.g. childless persons have lower demand,
and single parents have lower earnings capacity). The number of own-family
children per household is directly related to average family size and thus the
3 7 The childless category includes singles and couples as long as no own-
children are present. The couple with own-children category includes married
individuals or those in marital-type unions, a long as some own-children are
present, and is henceforth referred to simply as couples with children. The Single
Parents category includes individuals who are not married or in a marital-type
union and have own-children present.
3 8 Note that the stratification in item 1) allows for the possibility that
individuals with no own children can still live in a household with children not
their own.
level of household demand. The number of labor force participants per
household is directly related to total household income and earnings capacity.
Comparisons of these three variables between baby boomers and their
parents were performed using only those individuals aged 25-34 years. This
was done because after 35 years of age, decisions about family-type formation,
childbearing, and labor force participation are usually com pleted/9 4 0
Measures of both own-children and labor force participants per household come
from 10-year cohort averages.
Given that much of the theory of the household as applied to baby
boomers points to changes in the household demographic and labor market
actions of women, we also examine shifts in the household economic role of
women. We will explore changes in the percentage of women who worked
between 1965-90 (those who worked at all and those who worked full-time
full-year). We will also establish changes in individual earnings of women
over the period.
In section VII, trends in IAE are decomposed by factors affecting its
growth. The decomposition is performed for the purpose of identifying race
3 9 Since the sample is further stratified by family type, 5-year cohorts are
joined to form 10-year cohorts to maintain sufficient sample size.
4 0 We also looked at those individuals aged 25-44 years, but the relative
comparisons were unchanged.
34
differences in sources of IAE growth. The decomposition will ascertain what
the IAE trend from 1965-90 would have been if the household labor force and
demographic characteristics of baby boomers in 1990 were the same as their
pre-boomer parents in 1965. The trend in decomposed IAE could be thought
of as a "constant labor supply and demographic factor" trend.
The decomposition is performed in three steps: First, median IAE is
obtained for every 5-year age group for quinquennial years from 1965 to 1990.
Second, the proportion in each family-type is determined and the average
number of own-family children and labor force participants per household are
determined for each age-group/family-type combination in 1965. Last, age-
group/family-type specific IAE is adjusted for quinquennial years from 1970-90
to reflect the family-type proportions, household fertility, and household labor
force characteristics of 1965. The resulting IAE trend is called "Constrained
IAE".
IV. IAE OF BABY BOOMERS AND THEIR PARENTS
A. IAE Profiles
How have baby boomers fared economically relative to their parents at
the same age? Empirical evidence toward this question is introduced here.
35
Using a sample of individuals aged 15-64 between 1965 and 1990, two sets of
IAE cohort profiles were created, one for blacks and one for whites (see figure
2.2 and table 2.1).4 1 It is immediately clear that IAE profiles for whites are
higher than for blacks. Every white profile is higher than the corresponding
black profile. The dashed profiles identify baby boomer cohorts and the solid
profiles identify pre-boomer cohorts. The profiles labeled (1) through (11)
correspond to birth cohorts beginning with the youngest and ending with the
oldest; profiles labeled (1) to (11) refer to birth cohorts of 1960-64 to 1910-14,
respectively.4 2 In these IAE profiles, vertical comparison examines cohorts at
different times but at the same age.4 3 With only a few exceptions, vertical
comparison of profiles shows that holding age constant, median IAE of any
cohort exceeded median IAE of previous cohorts. This implies that with few
exceptions, median IAE increased with each successive cohort for both blacks
and whites.
4 1 Recall that income data from 1969 and 1984 was not included when creating
these profiles due to wide fluctuations in unemployment rates. The profile set
including these years is presented in the appendix to chapter 2.
4 2 The individual profiles in figure 2.2 are not fully shaped as the sample
profile in figure 1. This is because the sample data span only 25 years but a
complete age-IAE profile covering the 15-64 working life spans 45 years.
4jThe graphs in Figures 9 and 10 are plotted in ratio scale. Thus, the distances
between profiles portray growth rates of IAE.
The dashed baby boomer profiles, for both blacks and whites, are more
compressed than pre-boomer profiles and are more likely to overlap
immediately proceeding profiles. This is indicative of a slowdown in the
growth in IAE over the 1965 to 1990 period. However, this does not suggest
that baby boomers are not as well off as their parents. Comparisons of
successive profiles only considers a 5-year span, whereas comparison of baby
boomers to their parents should look back 5 profiles to span 25 years.
Reinspection of the profile set will show that for both blacks and whites, all
baby boomer profiles (the dashed curves) lie above all pre-boomer profiles (the
solid curves).
Showing the entire set of profiles is helpful as an exposition of the
general shape of the profiles, but it makes comparison of two profiles out of
the set somewhat arduous. Comparison between individual profiles is best
performed in a separate figure where only the profiles in question are
presented; this is executed below.
B. Comparing Baby Boomers with Their Parents: IAE
Comparison of baby boomers to their parents is accomplished by
vertically comparing IAE profiles of baby boomer cohorts to IAE profiles of
pre-boomer "parent" cohorts (assumed to have been born 25 years earlier).
Since the baby boom spans 20 years, four separate 5-year cohorts qualify as
37
baby boomer cohorts. Correspondingly, comparing baby boomers to their
parents entails making four separate comparisons. More specifically, we
compare the 1960-64 cohort to the 1935-39 cohort at 25-29 years of age
(cohort (1) vs. (6)), the 1955-59 cohort to the 1930-34 cohort at 30-34 years
(cohort (2) vs. (7)), the 1950-54 cohort to the 1925-29 cohort at 35-39 years
(cohort (3) vs. (8)), and the 1945-49 cohort to the 1920-24 cohort at 40-44
years of age (cohort (4) vs. (9)).
In all of these match-ups, black and white baby boomers show
significant gains in IAE over their parents4 4 ; on average, median age specific
IAE of baby boomers exceeded that of their "parents" by about 71% (see
figures 2.3 through 2.6). Additionally, closer examination shows that although
the levels of IAE of black baby boomers are lower than whites, gains by black
baby boomers are greater than for whites (median age specific IAE of black
baby boomers exceeded that o f their "parents" by 92.1% on average, whereas
the median age specific IAE of white baby boomers exceeded that of their
"parents" by about 68.1% on average). IAE gains were larger of older baby
boomer cohorts were larger than younger baby boomers.4 5
4 4 Henceforth, baby boomer "IAE gains" refers to positive differences between
median IAE of baby boomer cohorts and pre-boomer "parent" cohorts.
4 5 For black baby boomers, those aged 25-29, 30-34, 35-39, and 40-44 in 1990
had percentage gains in IAE o f 75.7%, 79.2%, 107.3%, and 106.1%, respectively.
For white baby boomers, those aged 25-29, 30-34, 35-39, and 40-44 in 1990 had
percentage gains in IAE of 63.3%, 62.7%, 73.5%, 72.9%, respectively. Note
C. Summary
Not only have both black and white baby boomers experienced levels of
IAE in excess of their parents, but blacks to a larger degree at every age. The
questions that remain unanswered include: What were the driving forces behind
the observed gains in IAE?, and, Are the these sources similar for blacks and
whites?4 6 To answer these questions, we next examine changes in the levels
of earnings, and then look at statistics on household demographic and labor
force decisions.
V. EARNINGS OF BABY BOOMERS AND THEIR PARENTS
A. Earnings Profiles
The baby boomer comparisons discussed above came from the
construction of IAE profiles. Similarly, the examination of changes in earnings
comes from developing earnings profiles. Since the aim here is to isolate labor
market outcomes not related to household dynamics (such as changes in labor
however, that since the sample only allows one age-range comparison for each
cohort set, we cannot distinguish period effects from age effects.
4 6 It is clear that IAE gains for black baby boomers exceed those of whites.
What is not clear is whether blacks had larger improvements in each component
of IAE (such as earnings or household demographic and labor force
characteristics).
39
force participation), the sample used to obtain measures of median earnings
includes only full-time full-year working males aged 15-64 between 1965 and
1990.4 7
The sets of earning profiles are presented in the same manner as were
IAE profiles (see figure 2.7 and table 2.1). Dashed lines are baby boomer
earnings profiles where solid lines are pre-boomer earnings profiles. Vertical
comparison is analogous to age specific comparison. The profiles labeled (1)
to (11) identify 5-year cohorts bom from 1960-64 to 1910-14, respectively.4 8
Positive growth in earnings for successive cohorts is not as clear as was the
case for IAE.4 9 As before, comparison between baby boomer and parent
cohorts is best presented in separate figures.
4 7 Profiles using all working males were also examined, but the results did not
differ substantially across subsamples and those profiles are not included. Recall
that for earnings analysis, females are excluded from earnings sampling due to
concerns over changes in female labor force participation over the 1965-90 period.
4 S The scale for earnings in Figure 1.11 is not the same as the scale for IAE in
Figure 1.9. Thus, relative distances between cohorts are not comparable.
■"Henceforth, baby boomer "earnings gains" refers to positive differences
between median earnings of full-time full-year baby boomers and full-time full-
year pre-boomer "parents".
40
B. Comparing Baby Boomers with Their Parents: Earnings
As was the case for IAE profiles, earnings comparison between baby
boomers and their parents is performed via four separate pairs of profiles. It is
shown that age specific earnings gains of black and white baby boomers were
positive, but not in the same magnitude as IAE gains (see figures 2.8 through
2.11). Median age specific earnings of baby boomer was about 20% grater
than their parents on average. Similar to IAE gains, earnings gains for black
baby boomer were larger than for whites (median age specific earnings of black
baby boomers exceeded that of their "parents" by 40.4% on average, whereas
median age specific earnings of white baby boomers exceeded that of their
"parents" by about 22.3% on average).
C. Summary'
Just as was the case for IAE, earnings gains by older baby boomer
cohorts were larger than for younger baby boomers.5 0 As is shown, baby
boomer gains in earnings are considerably smaller than gains in IAE, thus
5 0 For black baby boomers, those aged 25-29, 30-34, 35-39, and 40-44 in 1990
had percentage gains in earnings of 18.2%, 31.0%, 59.1%, and 53.2%,
respectively. For white baby boomers, those aged 25-29, 30-34, 35-39, and 40-44
in 1990 had percentage gains in earnings of 10.3%, 18.8%, 26.1%, 33.9%,
respectively. Note however, that since the sample only allows one age-range
comparison for each cohort set, we cannot distinguish period effects from age
effects.
41
implying that additional gains must have come from other sources. We turn to
this issue next.
VI. HOUSEHOLD DEMOGRAPHIC AND LABOR FORCE
CHARACTERISTICS
Analysis in previous sections found that baby boomers have made
substantial gains in IAE over their parents, blacks to a larger degree. It was
also found that the median earnings of full-time full-year working male baby
boomers was higher than for their parents; again, blacks to a larger degree.
However, since the gains in IAE are considerably larger than the gains in
earnings, some source of gains in IAE remains to be explained. In an attempt
to identify additional sources of baby boomer IAE gains, we focus on
household demographic and labor force characteristics. The sample used for
analysis in this section includes individuals aged 25-34 in 1965 (pre-boomers)
and in 1990 (baby boomers).'1 Household demographic components are
5 1 The additional stratification by family type brings about statistical concerns
of sufficient sample size. In response to this concern, 5-year baby boomer cohorts
are combined into 10-year cohorts here. Comparisons between 35-44 year old pre
boomers and baby boomers were also conducted, but the results were not
significantly different, and thus are not discussed here. See appendix A for those
results.
42
explored by way o f family-tvpe composition and the number of own-family
children per household. The household labor force component is examined
through the number of labor force participants per household.
A. Family-Type Composition
Baby boomers have family-type distributions that are quite different than
their parents. The distribution of family-type is influential to overall IAE
because the distribution of IAE is very unequal across family-type. Both black
and white baby boomers are more likely to be childless than their parents at 25-
34 years of age (see table 2.2. columns 1-3). Childlessness among baby
boomers has increased 62% among blacks and 110% among whites (relative to
their parents). Increased childlessness results from several different household
level decisions, including delays in or decisions against marriage, delays in or
decisions against childbearing, and divorce.5 2
Fewer baby boomers aged 25-34 describe themselves as in a couple u-ith
children relative to their parents. The number of baby boomers in a couple
with children decreased 44% and 36% for blacks and whites, respectively,
relative to parent cohorts. As with childlessness, the frequency of couples with
^Increases in divorce effect childlessness in two ways. First, childbearing
falls precipitously after a divorce, and if no children were yet born, then
childlessness results (at least for the time being). Second, even if a divorced
couple already had children, children typically live with only one parent. Thus,
divorce turns a couple with kids into a single childless person and a single parent.
43
children decreases as a result of several household factors, such as delays in or
decisions against marriage, delays in or decisions against childbearing, or
increase in divorce.
Single parenthood increased among baby boomer aged 25-34. This was
the case for blacks and for whites, who increased single parenthood by 67%
and 60%, respectively. A single parent can either be a previously married
individual with a child from that marriage, or a never married individual with a
illegitimate child. Therefore, increases in single parenthood among baby
boomers came from factors increasing either illegitimate births or divorce rates.
Whatever the specific causes that brought about the observed changes in
familv-type composition among baby boomers relative to their parents, it seems
to have affected blacks and whites similarly - or did it? Relative to their
parents, both black and white baby boomers increased childlessness, decreased
couples with kids and increased single parents, but the magnitudes of change
and the actual levels have been quit different. Black and white baby boomers
had similar percentage increases in single parenthood, but blacks have
increased levels that were much larger to begin with. Additionally, white baby
boomers moved to childlessness almost twice as much as blacks, and a higher
percentage of whites remained as couples with kids.
Of critical importance is the relative economic standing of each of these
three family-types. To illuminate the connection between family-type
44
composition and IAE, median IAE of each family-type is indexed to that of
couples with kids (see table 2.2, col. 4). IAE of couples with kids was chosen
as an index because this was the family-type category that baby boomer were
exiting. IAE of childless persons is considerably higher than that of couples
with kids or single parents. IAE of single parents is the lowest of all three
family-types, only 56.0% and 67.0% of couples with kids and just 39.7% and
42.1% of childless IAE for black and white baby boomers, respectfully. The
most salient difference in family-type changes between blacks and whites was
the larger move to childlessness, which had the highest IAE of all family-types.
Thus, although direction of family-type composition changes between
baby boomers and their parents was similar for blacks and whites, the relative
magnitudes of those changes and the subtle shifts toward opposite extremes is
what sets blacks and whites apart. Overall, changes in family-type composition
has positively affected the IAE of white baby boomers to a larger degree than
blacks, whose net change in IAE due to family-type changes is quite possibly
trivial.'"
5 "Increases by black baby boomers in the prevalence of childlessness is a plus,
but the increase in single parenthood has a negative effect on IAE; the resulting
net effect on overall baby boomer IAE is thus vague.
45
B. Own-Children Per Household
The second type of demographic measure examined is the household
number of own-family children under 16. Baby boomers in 1990 have
substantially lower own-family children than their parents in 1965 (table 2.2
columns 5-7). The decrease for black baby boomers in couple with kids and
single parent family-types (35.0% and 35.0%, respectively) was lower than for
whites (21.0% and 21.0%, respectively), but the larger increase in childlessness
by white baby boomers has kept overall decreases similar across race (46.2%
and 45.2% for all blacks and whites, respectively).
Decreases in the number of own-family children for those in couples
resulted from lower desired childbearing as well as delayed marriage given
constant desired childbearing. Similarly, fewer own-family children per
household among single parents resulted from increased divorce and increased
illegitimacy. To the extent that single parenthood increased due to increases in
illegitimacy, fewer own-children among single parents was a consequence of
increases in unrealized expectations of birth ligitimation as well as decreases in
desired childbearing among those never married.
In any case, both black and white baby boomers have lower overall
levels of own-children per household, which in itself positively effects levels of
IAE. There appears no race advantage in the effect of fewer own-children on
levels of IAE; black and white baby boomers benefit to a similar extent.
46
C. Household Labor Force Participation
The last type of household dynamic we contemplate is the number of
household labor force participants. Keep in mind that these figures represent
the total number of labor force participants in the household including all
families. Thus, individuals in single parent families, for example, can have
more than one household labor force participant if other families in the same
household have labor force participants as well. This is the case when single
parent baby boomers have doubled-up with other family members to increase
economic status.
It was found that black baby boomers, who have roughly 1/3 of the
cohort in each family-type, have about the same level of household labor force
participation as their parents (table 2.2 columns 8-10). But, white baby
boomers, on average, had considerably higher levels of household labor force
participation than their parents. For black baby boomers, household labor force
participation decreased among single parents and childless individuals (-9% and
-1%, respectively), but increased among couples with kids (+11%). Therefore,
black baby boomers had no substantial increase over their parents. On the
other hand white baby boomers who were childless and those in couples with
kids, who together constitute 92% of the baby boomer cohort, have increased
labor force participation by 6% and 33% respectively. The only family-type
category among white baby boomers who have decreased their average labor
47
force participation is single parents (-3%), who represent just 8% of the cohort.
Thus, household labor force participation among white baby boomers was
substantially higher than their parents.
The number of labor force participants in the household is determined by
several factors.5 '* One way labor force participation changes is when
household members alter their participation status (those who are labor force
participants exit, or those who are not enter). Another way is by the addition
of new household members who are already in the labor force.
There exists a separate issue that affected the labor force participation of
black baby boomers more than it did for whites. This is the high
unemployment (and low earnings) of unskilled less educated blacks (Freeman
and Holzer, 1987; Levy and Michel, 1991; Card and Kreuger, 1993). This
most likely resulted in race differences in the probability that being in the labor
force translated into being unemployed, and thus the prevalence of discouraged
5 4 Labor force participants were chosen rather than workers in an attempt to
count all individuals who decide to contribute to the work force, whether or not
they were currently employed. A non-labor force individual may decide to enter
the labor force, but this doesn’t always lead to working. This specification has
two implications. First, being in the labor force doesn’t imply a contribution to
household resources (household resources can decline even with constant earnings
and labor force participation). Second, even those individuals who decide to enter
the labor force may become discouraged workers after an episode of unsuccessful
searching. This analysis did not differentiate between discouraged workers and
those who didn’t want to work even when jobs were available.
48
workers. This connection between labor force participation and earnings is not,
unfortunately, reflected in statistics on household labor force participation.
The observed differences between household labor force participation of
baby boomers and their parents has clearly assisted the economic condition of
whites more than blacks. It was not that black baby boomers have eroded
labor force participation relative to their parents, but rather that white baby
boomers made positive changes relative to static rates for black baby boomers.
D. Changes in Women’s Status in the Household
The traditional household role of women was that of a homemaker; one
who gives birth to and raises children, and is involved in household production
but not extensively in the labor market. When women did work, it was often
part-time and in low paying positions. This characterization, while indicative
of one extreme, is more closely representative of pre-boomer women than of
baby boomer women. What remains unclear is the effect changes in womens’
household roles has had on IAE of baby boomers relative to their parents.
Various literature documents recent increases in women's labor force
participation and earnings levels (Levy and Michel, 1991; Blau and Kahn,
1994) and the increasing importance of women’s work to household economic
status (Cancian, Danzigea, and Gottschalk, 1993; Dechter and Smock, 1994).
Cancian et al. found that between 1968 and 1988, average household income
49
for couples has increased by more than $8000 (in 1988 dollars), two thirds of
which is accounted for by increases in womens’ income. Dechter and Smock
subdivided a sample of married households by type of income-provision
scheme; categories included male-provider, female-provider, and co-provider
schemes. Dechter and Smock found that the prevalence of the male-provider
structure is declining while the co-provider structure is increasing.
Additionally, they found that "co-provider marriages are economically
advantaged compared to other income-provision-role arrangements", and that "a
relatively substantial part of the total improvement in younger couples’
economic welfare over time stems from the shift towards co-provider
marriages." These conclusions point to changes in womens’ roles as beneficial
to economic status.
The current interest lies in how changes in the labor force participation
and earnings levels of female baby boomers has effected the IAE among baby
boomers, by race. The IAE gender gap for persons aged 25-44 was slight in
1965 and 1990 and was relatively stable over the period (see table 2.3).5 3
Black pre-boomer males had a small advantage over pre-boomer females in
1965 (5%); a slightly larger advantage existed for baby boomer males in 1990
(8.2%). For whites, pre-boomer females had slightly higher IAE than males in
5 ;,The group of individuals aged 25-44 in 1990 includes all baby boomers (and
only baby boomers); individuals of the same age in 1965 could represent all of
their parents.
1965 (5%), but the gap for baby boomers in 1990 was trivial (1%). The
earnings gap, as opposed to the IAE gap, has narrowed substantially for the
same samples over the same 1965 to 1990 period (also table 2.3). Black
females, who earned only 36% of black males in 1965, increased their ratio to
78% by 1990. White women, who earned just 38% of white males in 1965
jumped to earning 58% of white males in 1990. It is interesting to find that
large increases in womens’ earnings relative to men did not translate to
increases in womens’ IAE relative to men. This is indicative of two important
points. First, the pooling of individual resources in determining IAE has an
equalizing effect on measures of economic status. Second, even though women
are earning relatively more, they are also increasingly responsible for single
parent families.
Employment rates among women aged 25-44 increased dramatically
between 1965 and 1990 (see table 2.3). Employment rates were measured two
ways. First, we determined the proportion of individuals who worked full-time
for the entire year, henceforth referred to as full-time full-year workers.
Second, we determined the proportion of individuals who worked in any
capacity (full or part-time, full or part-year), henceforth referred to as all
workers. Female baby boomers have dramatically increased the proportions of
both full-time full-year workers and all workers relative to their female parents.
51
Between 1965 and 1990, the proportion of black and white women aged
25-44 who were full-time full-year workers increased considerably from 28% to
64.1% and from 20.5% to 57.1%, respectively. Over the same period, the
proportion of full-time full-year workers among black and white men aged 25-
44 increased only slightly or not at all. The proportion increased among blacks
from 63.4% to 71.4% and decreased only slightly among whites from 80/8% to
79.7%. The scenario among all workers is much the same. Between 1965 and
1990, the proportion of female blacks aged 25-44 in the all workers category
increased substantially from 64.8% to 76.7%; the increase among white females
was even larger from 46.3% to 79.0%. The proportion of men aged 25-44 in
the all worker category deceased for blacks and whites over the same period
(from 94.7% to 87.6% among blacks and from 98.0% to 95.8% among whites).
Note that although the full-time full-year worker and all worker
populations increased substantially for black (white) women relative to black
(white) men. men maintained higher proportions in each worker category in
1990. Additionally, race differences in the measurements of changing womens'
roles were rather inconsequential. Black female/male earnings ratios increased
more than whites, but this is most likely a results of the earnings experiences of
young black men rather than a changing role of black women. The proportions
of both black and white female baby boomers who worked (considering either
full-time full-year workers or all workers) increased significantly relative to
52
their parents. The proportion of black all worker women did not increase as
much as for white women, but black women started out at a much higher
proportion, and improvements at the upper end are much harder to attain.
In comparing the figures from 1965 and 1990, it is clear that female
labor force decisions and earnings experiences were quite different for baby
boomers than for their parents. The increased earnings levels and the increased
labor force participation contributed positively to overall IAE. It is clear that
for baby boomers, the changes in the role of women as earner and co-provider
have positively affected IAE levels over the 1965 to 1990 time period.
E. Summary
Subsections A through D have outlined how baby boomers have changed
average levels of three household based factors. It was shown that changes in
family-type proportions were generally beneficial to white baby boomers but
trivial overall for black baby boomers. The decline in own-children per
household was substantial for all baby boomers, and the effect on IAE was
beneficial for black and white baby boomers. Blacks were shown to decrease
the number of family-type specific own-children to a larger degree than whites,
but the stronger tendency for whites to be childless brought their average
decline in own-children in line with that of blacks. Increases in baby boomer
household labor force participation among whites (relative to their parents) had
53
a positive effect on IAE, but among blacks, whose household labor force
participation was relatively unchanged, IAE was unaffected. Thus, white baby
boomers had the relative advantage in improving IAE via changes in household
labor force participants.
The changing household role of women from 1965 to 1990 was shown
to positively affect IAE for both black and white baby boomers. The female
proportion of the work force has increased dramatically among full-time full-
year workers and among all M ’ orkers for blacks and whites, and female earnings
levels have jumped considerably. The changes in womens’ economic roles
across race however, have not been significantly different.
The overall picture portrays white baby boomers as having made larger
gains in IAE from household demographic and labor force characteristics.
What is not clear is the overall effect of these household factors on black and
white baby boomers. In the next section we turn to a simulated decomposition
technique which sheds light on this question.
VII. IAE DECOMPOSITION
Empirical results from previous sections have implied that black and
white baby boomers, both having experienced significant gains in IAE over
54
their parents, have made these gains from different combinations of sources.
Black baby boomers had relatively larger earnings growth while white baby
boomers benefit relatively more from changes in household demographic and
labor force characteristics. But with regard to these household demographic
and labor force characteristics, the net effect on IAE is not clear. For instance,
have the overall changes in household demographic and labor force
characteristics made by black baby boomers (relative to their parents) had a net
positive effect on their IAE?
In this section, we turn to a decomposition technique that simulates the
IAE trend between 1965 and 1990 that would have occurred if the three
household demographic and labor market factors discussed above were held
constant.
The decomposition technique produces an estimate of the IAE trend that
would have occurred from 1965 to 1990 if there had been no changes in the
demographic factors considered above since 1965. This IAE from which the
effects of other factors has been "decomposed" is referred to as "Constrained
IAE". When unconstrained of actual IAE trends are plotted against constrained
IAE, race differences are immediately visible (see figure 2.12). The actual IAE
trend for blacks is consistently lower than the constrained IAE trend (25%
lower in 1990), but the actual IAE trend for whites is consistently higher than
the constrained IAE trend (8.5% higher in 1990). This implies that for blacks,
55
the demographic effects have decreased IAE over the period, but the same
demographic effects have increased IAE over the period for whites.
Additionally, for a simulation of the race differences in the effects of earnings
growth, examine the convergence of the dashed lines, which show the black-
ratio in constrained IAE dropped from 0.54 in 1965 to 0.88 in 1990. This
convergence shows that the effects of earnings on black IAE (net of changes in
household demographic and labor force characteristics) greatly outpaced that of
whites.
This decomposition evidence is consistent with results above, which
indicated that black baby boomers had the relative advantage in earnings
growth whereas white baby boomers had the relative advantage in changing
household demographic and labor force factors to improve IAE. The results of
the decomposition cannot be directly applied to the experiences of baby
boomers alone since the samples used in the decomposition were more
comprehensive and included all individuals, but the general picture remains
quite revealing.
VIII. CONCLUSION
56
The primary aim of this analysis was to estimate race differences in the
economic status of baby boomers relative to their parents. Another important
issue explored was the relative influence of changes in demographic factor
levels between baby boomers and their parents. Using a household based
measure of economic well-being, called Total Real Money Income Per Adult
Equivalent (IAE), this study found that both black and white baby boomers,
despite their relative cohort size, have made significant gains in IAE over their
parents at the same age, blacks to a larger extent. An interesting finding was
that the relative sources of IAE gains differed across race, as black baby
boomers had stronger relative earnings effects and white baby boomers had
stronger household demographic and labor force participation effects.
Analysis of earnings profiles showed that median earnings of black and
white male baby boomers has surpassed that of their male parents. However,
the gains in earnings were significantly lower than the gains in IAE for black
and white baby boomers. Just as with IAE, percentage gains in the earnings of
black male baby boomers, relative to their parents, were substantially higher
than for white male baby boomers.
Examination of changes in household demographic and labor force
characteristics suggests that white baby boomers have positively affected IAE
via changes in household composition and labor supply; the net effect for
blacks is unclear. Black and white baby boomers made similar decreases in the
number of own-family children under 16 and thus decreased average household
demand. But. changes in family-type formation and household labor force
participation have on average mitigated IAE gains among blacks. More
specifically. 30% of black baby boomers are single parents compared to 18%
for their parents, while only 8% of white baby boomers are single parents
compared to 5% for their parents. The number of household labor force
participants for white baby boomers increased overall, but black baby boomers
showed no overall increase.
The changing role of women in the household has positively affected the
IAE of baby boomers relative to their parents. As of 1990, female baby
boomers were earning far more than their parent counterparts in 1965. Also,
the labor force participation of women has increased significantly from that of
their parents. There does not appear to be any substantial race difference in
earnings levels changes or employment rate changes of baby boomers. Both
black and white female baby boomers made similar increases in employment
and experienced similar growth in earnings over the period.
A decomposition analysis concluded that black baby boomers would
have had higher IAE in 1990 if they would have maintained the demographic
and labor force levels o f their parents in 1965. To the contrary, white baby
58
boomers have benefit significantly from making the changes in demographic
and labor force characteristics. Thus, even though black and white baby
boomers have both made gains in IAE over their parents, the relative sources of
these gains were not the same. If alterations in household demographic and
labor force characteristics are thought of as a possible response to perceived
shortfalls in economic status, then the relative lack of response by blacks baby
boomers is consistent with their relative advantage in earnings gains, especially
in light of the fact that overall gains in IAE for baby boomers still exceed those
of whites.
Individual Income
59
FIGURE 2.1
SAMPLE INCOME PROFILE
ratio scale
20000------------------------------------------------------------------------------------------------
15000-
10000-
5000-
15-19 20-24 25-29 30-34 35-39 4 0 - 4 4 45-49 50-54 55-59 60-04
Age Group
Median Income Per Adult Equivalent ($1988)
O
ratio scale
20000-1-------
15000-
10000-
(I)
5 0 0 0 - < J>
FIGURE 2.2
IAE: blacks aged 15-64 by cohort, 1965-1990
(3) / * ,
( 10)
(ID
( 8 )
/
4
(*)
2 ? 2 4 i s i l 9 3 0 0 4 3 5 0 9 4 0 4 4 4 5 J 9 50^54 J ? S 9 6 0 6 4
A ge G ro u p
IAE: whites aged 15-64 by cohort. 1965-1990
ratio scale
20000-
15000-
(U ) > io o c o -
( 10)
500 0 -
15-19 20-24 25-29 30-J4 35-39 4 ^ 4 4 45-49 50-54 55-59 60-64
A ge G ro u p
( 1 ) 19GU-G4 ( 2 ) 1 3 5 5 - 5 9 ( 3 ) 1 9 5 0 -5 4 ( 4 ) 1 9 4 5 -4 9
( 5 ) 1940-4H ( 6 ) 1 9 3 5 - 3 9 ( 7 ) 1 9 3 0 -3 4 ( 8 ) 1 9 2 5 -2 9
( 9 ) 1 9 2 0 -2 4 ( 1 0 ) 1 9 1 5 -1 9 ( 1 1 ) 1 9 1 0 -1 4
6 1
TABLE 2.1 : IAE & EARNINGS OF BABY BOOMERS AND PARENTS
Persons Aged 25-44 by Age Group
25-29
Age Group
30-34 35-39 40-44
M E D IA N I A E
BLACKS
WHITES
Baby boomers in 1990 9924 10144 11705 11897
Parents in 1965 5647 5662 5773 6151
Baby Boomer Gains (%) 75 . 7 79.2 107.3 106.1
Baby boomers in 1990 15538 15702 16372 17219
Parents in 1965 9516 9650 9437 9958
Baby Boomer Gains {%) 63.3 62 . 7 73 .5 72.9
E A R N I N G S : F U L L - T I M E F U L L -Y E A R M A LE W ORKERS
BLACKS Baby boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
17176
14533
18 .2
18130
13841
31.0
22901
14394
59.1
23855
15571
53 .2
WHITES Baby boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
22901
20761
10.3
26718
22491
18.8
30534
24221
26 .1
32443
24221
33 .9
EARNINGS: ALL MALE EARNERS
BLACKS Baby boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
14313
12661
13 .0
15744
12111
30.1
19084
12457
53 .2
19084
13045
46.3
WHITES Baby boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
20038
20501
3.4
23855
23918
11.1
28626
28474
25.3
30534
29613
35 . 8
Median Income Per Adult Equivalent ($iy8S )
(N
V O
FIGURE 2.3
ratio s o le
20000-1
LAE: black 1960-64 baby boomer and 1935-39 pre-boomer cohorts
(1960-64)
/
/
3509 40-44 45^9
A ge G ro u p
ratio scale
20000- t
IAE: white 1960-64 baby boomer and 1935-39 pre-boomer cohorts
e 5000-
(1960-64)
(1935-39)
1M 9 30-34 35-39 40-44 4J-49
A ge G ro u p
50-54 55-59 60-64
Medina Income Per Adult E quivalent ($1988)
m
V J D
FIGURE 2.4
n iio JC ik
20000-t
LAE: black 1955-59 baby boomer ami 1930-34 pre-boomer cohorts
ratio scale
20000-1
IAE: white 1955-59 baby boomer and 1930-34 pre-boomer cohorts
4
(19S5-S9)
(19J0-J4)
a 30CO-
1
s
(1955-59)
(1930-34)
A ge G ro u p 15-19 20-24 25-29
_ j------p
30-34 35-39 40-U 45-49
A ge G ro u p
50-54 55-59 60-64
Median Income Per Adult Equivalent ($1988)
V O
FIGURE 2.5
IAE: black 1950-54 baby boomer and 1925-29 pre-boomer cohorts
ratio scale
20000-j--------------------------------------------------------------------------------------------------------------------------------------------------------------
15000-
I 0 C 0 O *
15-19 20-24 25-29 30-J4 35-39 « W 4 45-49 50-54 55-59 60-64
A ge G ro u p
ratio scale
20000
IAE: white 1950-54 baby boomer and 1925— 29 pre-boomer cohorts
a 5000-
.2
* 2
5
(1925-29)
15-19 20-24 25-29 30-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60-64
Mediitn Income Per Adult E q u iv a le n t
in
V O
FIGURE 2.6
n iio scale
20000-t
IAE: black 1945-49 baby boomer and 1920-24 pre-boomer cohorts
n iio sctJe
20000-r
IAE: white 1945-49 baby boomer and 1920-24 pre-boomer cohorts
4
(1945-49)
(1920-24)
■ 8
S
4
(1945-49)
(1920-24)
15-19 20-24 25-29 JO-34 35-39 40-14 4J-49 50-J4 55-59 6044
A ge G ro u p
15*19 20*24 25-29 30-34 35.J9 40-44
A ge G ro u p
55-59 6 064
Median Individual Earnings ($1988)
V O
V O
FIGURE 2.7
EARNINGS: black full-time full-year working males aged 15-64 by cohort, 1965-1990
(full-tim e full-year black m ale w orkers)
ratio scale
35000-1
EARNINGS: white full-time full-year working males aged 15-64 by cohort, 1965-1990
(full-tim e full-year w hite m ale w orkers)
30000-
2 5 0 0 0 -
20000-
15000-
0000-
15-19 20-24 25-29 J o S 3509 4oC* 45^49
A ge G ro u p
50-54 55-59 60-64
2 5000-
20000- (10)
(11)
1 0 0 0 0 -
a
*S
9
* >
c
500 0 -
1000
15-19 20-24 25-29 30-34 35-39 40-44 4J-49 50-54 55-59 60-64
A ge G ro u p
( I ) 19G0-G4 ( 2 ) 1 9 5 5 -5 9 ( 3 ) 1 9 5 0 -5 4 ( 1 ) 1 9 4 5 -4 9
( 5 ) 1 9 4 0 -4 4 (G ) 1 9 3 5 -3 9 ( 7 ) 1 9 3 0 -3 4 ( 8 ) 1 9 2 5 -2 9
( 9 ) 1 9 2 0 -2 4 ( 1 0 ) 1 9 1 5 -1 9 (II) 1 9 1 0 -1 4
Median Individual Earnings ($1988)
r* -
v o
FIGURE 2.8
EARNINGS: black 1960-6*1 baby boomer and 1935-39 pre-boomer cohorts
(full-iim e full-year black m ale w orkers)
EARNINGS: white 1960-6*1 baby boomer and 1935-39 pre-boomcr cohorts
(full-tim e full-year w hite m ale w orkers)
30000-
25000-
20000-
15000-
ioooo-
(1960-64)
(1935-39)
15-19 20-24 25-29 35-39 40-44
Age Group
45-49 50-54 55-59 60-64
30000-
2 5 0 0 0 -
(1935-39)
(1960-64)
15-19 20-24 25-29 35-39 4G44
Age G ro u p
45*49 50-54 55-59 60-64
Median Individual Earnings ($1988)
oo
'O
FIGURE 2.9
EARNINGS: black 1955-59 baby boomer and 1930-34 pre-boomer cohorts
(full-tim e full-year black male workers)
EARNINGS: white 1955-59 baby boomer and 1930-34 pre-boomer cohorts
(full-tim e full-year w hile m ale w orkers)
30000-
25000-
20000-
15000-
10000-
4
(1955-59)
15-19 20-24 25-29 35-39 40-44 45-49
Age G roup
50-54 55-59 60-64
30000-
2 5 0 0 0 -
-3 500 0 -
(1930-34)
(1955-59)
15-19 20-24 25-29 30^34 35-39 4 0 ^ 4 4 5^9
Age G ro u p
50-54 55-59 60-64
Median Individual Earnings ($1988)
ON
VO
FIGURE 2.10
EARNINGS: black 1950-54 baby boomer and 1925-29 pre-boomer cohorts
(full-lim e full-year black m ale w orkers)
30000-
25000-
( 1 9 5 0 -5 4 )
( 1 9 2 5 -2 9 )
20-24 25-29 30-34 35-39 4tM4
Age G roup
45-49 50-54 55-59
EARNINGS: while 1950*54 baby boomer and 1925-29 pre-boomer cohorts
(full-tim e full-year w hile m ale w orkers)
ratio scale
30000-
25000-
20000-
0 15000-
3
9 \
^ I 0 0 0 0 -
e
j
1 3 0
5
3
* 8
(1925-29)
(1950-54)
15-19 20-24 25-29 30-34 35-39 40-44 45-49
Age G ro u p
50-54 55-59 60-64
Median Individual Earnings ($1988)
o
F IG U R E 2.il
EARNINGS: black 1945-49 baby boomer and 1920-24 pre-boomer cohorts
(full-tim e full-year black m ale w orkers)
30000-
25000-
20000-
15000-
10000-
(1945-49)
20-24 25-29 30-34 3 5 .3 9 40-44 45-49
Age G ro u p
EARNINGS: white 1945-49 baby boomer and 1920-24 pre-boomer cohorts
(full-tim e full-year w hite m ale w orkers)
ratio sc a le
30000-
25000-
£
*
• £ 5 0 0 0 -
.2
• S
(1920-24)
« r
(1945-49)
15-19 20-24 25-29 30-34 35-39 40-44 45-49
Age G ro u p
50-54 55-59 60-64
71
TABLE 2.2 : HOUSEHOLD DEMOGRAPHIC AND LABOR FORCE
STATISTICS
Baby Boomers and Their Parents aged 25-34
IAE Indexed Own-Family Household
Family Type to Couple Children Labor Force
Frequency (%) with Kids Under 16 Participants
Family Cohort Cohort Cohort Cohort
Type ' 1930-39 1955-64 % a 1955-64 1930-39 1955-64 % a 1930-39 1955-64 % *
BLACKS
Childless 23.0 37.0 +62.0 141.0 0.00 0.00 -- 1.74 1.72 - l.C
Couple w/
Kids 59.0 33.0 -44.0 100.0 3.22 2.09 -35.0 1.51 1.68 +11.0
Single
Parents 18.0 30.0 +67.0 56.0 3.33 2.18 -35.0 1.46 1.33 - 9.0
WHITES
Childless 20.0 44.0 +110.0 159.0 0.00 0.00 -- 1.75 1.86 + 6.0
Couple w/
Kids 75.0 48.0 -36.0 100.0 2.46 1.94 -21.0 1.26 1.67 +33.0
Single
Parents 5.0 8.0 +60.0 67.0 2.29 1.81 -21.0 1.66 1.61 - 3.0
72
TABLE 2.3 : IAE, EARNINGS, AND LABOR FORCE PARTICIPATION
Baby Boomers and Their Parents aged 25-44
BLACKS
males
females
f/m ratio
WHITES
males
females
f/m ratio
IAE
(median)
1555 1990
6291 13240
5543 12021
0.95 0.918
9960 17048
10436 16945
1.05 0.99
EARNINGS
(median)
1965 1990
3600 18000
1300 14000
0.36 0.78
6192 25600
2350 14800
0.38 0.58
% FTFY
WORKER
1965 1990
63.4 71.6
28.0 64.1
0.44 0.90
80.8 79.7
20.5 57.1
0.25 0.72
% ANY
WORKER
1965 1990
94 . 7 87.6
64.8 76.7
0.68 0.88
98.0 95.8
46.3 79.0
0.47 0.82
Median Income Per Adult Equivalent ($1988)
73
FIGURE 2.12
ACTUAL .AND CONSTRAINED LAE: persons aged 15-64, 1965-1990
ratio sc ale
20000-
W hite Actual
15000-
White Constrained
Black Constrained
10000-
Black Actual
5000-
1965 1970 1980 1975 1985 1990
Year
APPENDIX TO CHAPTER 2
74
Construction of IAE:
IAE is obtained by dividing the sum of individual income within the
household by the number of adult equivalents in the household:
IAE, = ( T N, income.. ) / AE.
j v ^ 1=1 y ' 1 j
i = 1 to household size and j = household identifier.
Equivalency Scale Specification
The scale used to calculate household adult equivalents is:
1st adult = 1.0; every additional adult = 0.8
1st child = 0.4; every additional child = 0.3
which accounts for varying resource requirements between adults and children,
and for separate economies o f scale among adults and children (Fuchs, 1986).
Note that this specification allocates the same value o f IAE to every member of
the household, thus it assumes equal sharing.
Median Income Per Adult E quivalent ($1988)
ratio scale
20000-. -----
IAE: blacks aged 15-64 by cohort, 1965-1990 IAE: whites aged 15-64 by cohort, 1965-1990
( 10)
0 )
(3)
( 2 )
(tl)
(8 )
4
( 4 )
15-19 2?24 25-29 3004 3509 1044 4 5^9 S0O4 350 9 60^64
Age Gruup
20000-
15 0 0 0
( 11) (3) # > 1 0 0 0 0 -
( 10)
3000-
15-19 20-24 23-29 30-34 33-39 40-44 45-49 50-54 55-59 6 0 ^ 4
Age G roup
(1) I9G0-G4 (2) 1955-59 (3) 1950-54 (4) 1945-49
(5) 1940-44 (G) 1935-39 (7) 1930-34 (8) 1925-29
(9) 1920-24 (10) 1915-19 (II) 1910-14
CHAPTER 3
76
THE IMPACT OF INEQUALITY AND EQUIVALENCY SCALE
SPECIFICATION ON RELATIVE ECONOMIC IMPROVEMENTS:
Baby Boomers and Their Parents
I. ABSTRACT
This study utilizes Current Population Survey Data from 1965-1990 to explore
the effects of increasing economic inequality, by race, on the improvements in
economic status of baby boomers relative to their parents. Results from
Chapter 2 indicate that median Income Per Adult Equivalent (IAE) of both
black and white baby boomers was substantially larger than pre-boomers parent
cohorts. However, the experience of individuals at the ends of the distribution
is unclear. The first aim of this chapter is to analyze the distribution of IAE
for individuals aged 25-44 between 1965 and 1990, and discern whether
changes in this distribution thwarted baby boomer gains at the 25th and 75th
percentile. We find that despite the existence of increasing IAE inequality over
the period, IAE of baby boomers was substantially greater than their parents,
even at the 25th and 75th percentiles. The second aim of this paper is to
77
examine the stability of estimates of economic status across competing
equivalency scale assumptions. The measurement of IAE in Chapter 2 was
based on a particular equivalency scale specification, thus the findings rely on
the validity of the assumed scale. In this study we revisit earlier results and
conclusions. The specific equivalency scale assumptions are relaxed and three
alternative equivalency scale specifications are employed. It is found that when
alternative equivalency scale implemented, results of IAE comparisons are
parallel to those in chapter 2 with regards to trends, but moderate differences
exist in levels.
II. INTRODUCTION
A. Increasing Inequality and Economic Improvement
In chapter 2 we addressed the question of whether baby boomers
experienced levels of economic status in excess of their parents at the same
age, and if the experiences were similar for blacks and whites.5 6 We found
both black and white baby boomers, through slightly different means, did in
5 6 In this chapter we employ the same assumption about baby boomer parents
as in chapter 2. That is, the average age spread between baby boomers and their
parents is assumed to be 25 years, thus 5-year cohorts born on average 25 years
before 5-year baby boomer cohorts are assumed to represent "parents".
78
fact have higher Income Per Adult Equivalent (IAE) than their parents at the
same age. The question was important in light of potential economic pitfalls
faced by baby boomers due to their relative population size, recent stagnation
in earnings growth, and differences in earnings growth between blacks and
whites.
Another manner in which the economic condition faced by baby boomers
differed from that of their parents was in the level of economic inequality.
Economic inequality has attracted increased attention recently, in part due to
the concurrence of increasing inequality and stagnant earnings growth. As long
as earnings growth is substantial while inequality is increasing, the outcome is
simply that some individuals are getting richer faster than others. However,
when economic inequality occurs in the face of stagnant earnings growth, the
result is the rich get richer and the poor get poorer. By most measures,
economic inequality has increased between 1965 and 1990, with most of the
increase occurring from the early 1970’s (Levy and Michel (1991); Karoly
(1992); L ew and Mumane (1992); Burtless (1993)).5 7 The interest in this
chapter lies in the influence that economic inequality may have on economic
5 7 A small proportion of the literature refutes the idea that inequality is on the
rise. Mayer and Jencks (1993) point to other measures of economic status (such
as assets, wealth, use of health and social services) that result in relatively constant
inequality measures. Sleznick (1994) finds that inequality has not generally risen
when expenditures rather than income are used as a measure of economic well
being.
79
improvements of baby boomers over their parents. The findings in chapter 2
led to the conclusion that baby boomers, across race, have made substantial
gains in IAE over their parents at the same age. However, these conclusions
were based on median measures of economic status, and may not be
representative of the whole IAE distribution. It is plausible that a sufficient
widening of the IAE distribution between 1965 and 1990 could lead to
dramatic gains at the upper end while preventing gains at the lower end. It is
also possible that the changes in IAE inequality from 1965 to 1990 have
worsened the relative economic position of black baby boomers within the
distribution, resulting in smaller and larger proportions at the upper and lower
ends, respectively. If these scenarios were the case, then the results of chapter
2 become less applicable to the baby boomer population in general. This study-
wili determine whether increases in IAE inequality have undermined previous
median-based conclusions of substantial gains by baby boomers over their
parents for individuals at lower ends of the IAE distribution. Before moving
on we will discuss common beliefs on forces behind recent increases in
economic inequality, and describe the characteristics of individuals that are
increasingly represented at the lower end of the distributions.
8 0
Explanations for Recent Increases in Economic Inequality
Many different micro-level measures have been used to represent the
economic status of the US population. Such measures include wages, earnings,
and income. The context of these measures has most often been the individual,
but recently more attention has been paid to the household.5 8 While research
attention has largely focused on individual economic inequality without
reference to household, a growing literature has addressed household (or
household based) inequality.5 9 Indeed, measures of household inequality are
heavily influenced - if not dominated - by inequality among the individuals in
the household; but household decisions and characteristics also influence the
overall economic distribution. We first discuss issues related to individual
inequality, and then turn to household economic inequality.
Earnings Inequality
Traditional explanation for changes in inequality center on business cycle
effects. The basic understanding of the forces behind earnings inequality was
questioned after the 1970‘s when trends in inequality were not as closely
5 8 A s in chapter 2, household income can be measured in several ways,
including total household income, per capita household income, and household-
based equivalent income.
5 9 Household inequality refers to an unequal distribution of household
aggregated income measures, whereas household based inequality refers to an
unequal distribution of individual income adjusted for household characteristics.
81
related to observed business cycles. Earnings inequality increased since the
early 1970’s despite expansionary cycles in the 1970’s and 1980’s. This new
development brought into question the general understanding of the factors
influencing earnings inequality, and brought about increased academic attention.
An excellent review by Levy and Murnane (1992) comprehensively
summarized the literature on earnings inequality. As stated by Levy and
Murnane,
Explanation of inequality trends is usefully divided into two parts. The
first part examines the inequality between groups of workers defined by
age/experience and education, inequality which, for men was stable or
declining in the 1970s and grew sharply in the 1980s. The second part
examines those factors that may be responsible for growing within-gxowp
inequality, the steadily growing earnings inequality among men with the
same education and experience, (pp. 1353-54)
The factors effecting these two components are quite different, and not equally
understood. Using an economic framework, factors affecting these two
components could be thought of as supply based (affecting the size,
composition, or quality of the labor force) or demand based (affecting the
structure of firm demand for labor given the characteristics of the labor
force).6 0
Between-Group Earnings Inequality Between-group earnings
inequality was stable or declining in the 1970’s, but increased through the
6 0 This division does not preclude a feed-back mechanism between lagged
demand and current supply.
8 2
1980’s. What combination of supply and demand factors effected this trend?
Katz and Murphy (1992) find that fluctuations in both the age and education
premia (the differential earnings associated with higher age and educational
attainment) are associated with growth rates of different labor force groups
(supply side), as well as stable growth rates in the relative demand for college
educated workers (demand side).
Supply side changes in labor force composition include the relative
proportion of high school and college graduates, changes in the age
distribution, and representation of women in the labor force. The entrance of
relatively large and highly educated baby boomer cohorts into the labor force,
for example, contributed to changes in the age/education distribution of the
labor force, and reduced the earnings of young workers compared to that of
older experienced workers.
Several demand side factors influenced age and education premia in the
1970‘s and 1980’s. Changes in the industrial structure in the US decreased the
number of manufacturing jobs (a large source of demand for low
skill/education workers) and increased the number of service sector jobs (with a
typically larger proportion o f low paid workers), resulting in lower relative
wages for low educated young workers. Non-neutral technological change
increased the productivity of highly skilled and educated workers and increased
the education premium. Other explanations include increased outsourcing of
83
lower paying manufacturing jobs and changes in wage setting institutions (i.e.
decreased unionization) which both decrease the relative earnings of lower
educated lower paid workers.
Between-group inequality trends have reflected the relative importance of
these factors. The education premium, which decreased in the 1970's. more
than offset the increased age/experience premium and resulted in decreases in
between-group inequality in the 1970’s. During the 1980’s, the education
premium increased dramatically along with the age premium, thus resulting in
increased between-group inequality during the 1980’s. However, trends in
between-group earnings inequality are only part of the inequality story.
W ithin-G roup Earnings Inequality' While the recent trends in within-
group inequality are well established, much less is understood about the causal
factors. Since the early 1970’s, inequality within age/experience, education,
and gender groups has steadily increased (Katz and Murphy (1992). Levy and
Murnane divide the hypotheses explaining the increases in within-group
inequality into five categories:
1. Changing Characteristics o f the Labor Force (other than
age/experience, education, and gender), which fall on the supply
side.
2. Increasing Returns to Skill (same covariates as #1), which fall on
the demand side.
3. Increasing Industry Specific Wage Differentials, which fall on the
demand side.
84
4. Plant or Firm Specific Wage Differentials Within Industry, which
fall on the demand side.
5. Changing Wage Setting Institutions, which fall on the demand
side.
Items (1) and (2) refer to worker characteristics that are latent in data sets such
as the CPS. Items (3) and (4) refer to situations where the earnings of certain
labor groups are not equal across all employers. Item (5) refers to changes in
wage setting conventions that may be unrelated to changes in productivity or
final demand. Support for these hypotheses is at best preliminary and
agreement on their relevance is incomplete, but a short discussion of each
proposition is in order.
The idea behind Changing Characteristics o f the Labor Force is that in
addition to age/experience, education, and gender, other personal characteristics
further stratify the labor market, and changing the population proportions across
these classifications changes the relative wages in each. So in fact, this is still
a type of berween-groxxp inequality factor that seems to affect within-group
inequality due to latency of determinant characteristics. Increasing Returns to
Skill also relate to these additional latent personal characteristics, inasmuch as
they are rewarded in the labor market. Howell and Wolff (1991) found that
shifts in labor force composition have motivated increased demand for
interactive skills, defined as skill in mentoring, negotiating, instructing,
supervising, and persuading. This type of Increasing Returns to Skills could
85
also be considered a between-group inequality effect. But this effect, due to
latency of certain variables, seem to affect within-group inequality.
Two other proposed sources of within-group inequality are Increasing
Industry Specific Wage Differentials and Plant or Firm Specific Wage
Differentials Within Industry. Each relate to a widening dispersion of earnings
(by industry or plant/firm within industry, respectively) for individuals with
similar age/experience, education, and gender characteristics. Blackburn (1990)
found that 15% of increases in within-group inequality for men stemmed from
worker movements from industries with low residual variation in earned
income (e.g. primary goods) to industries with high residual variation in earned
income (e.g. services). Even within industry, increasing wage dispersion
among firms contributes to overall within-group earnings inequality (Groshen
(1989); Davis and Haltwinger (1991)). Neither Increasing Industry Specific
Wage Differentials or Plant or Firm Specific Wage Differentials Within
Industry hypotheses provide causal reasons for increases in within-group
inequality, but they do identify areas for research.
Despite a clear understanding of the causes for increased within-group
earnings inequality, it has been established that such inequality has increased
steadily since the early 1970’s. The steady increase in within-group earnings
inequality coupled with the 1970’s decrease and the 1980’s increase in
between-group earnings inequality resulted in slow increases in overall earnings
8 6
inequality in the 1970’s followed by large increases in inequality through the
1980’s. The next step is to identify the connection between individual earnings
inequality and household inequality.
Household Economic Inequality
Discussion in chapter 2 outlined the advantages of household-based
equivalent income measures in representing the economic status of individuals.
Following that reasoning, efforts to determine the level of inequality in the
economic status of individuals should start by making use of the most
appropriate measures of economic status. Such an interest in household-based
inequality is supported by recent economic literature (Hanratty and Blank
(1992); Karoly (1992); Blackburn, Mckinley, and Bloom (1993); Burtless
(1993); Haveman and Buron (1993); Meyer and Jencks (1993); Browning,
Bourguignon, Chiappori, and Lechene (1994)).
To be clear, we do not discount the relevance of earnings inequality to
household income inequality, as they are in fact forever linked. Variations in
earnings levels are an important component, if not the most important
component, of variations in household income. Recent studies addressing both
earnings and household income inequality have concluded that the observed
increase in household inequality is really a story about men’s earnings
inequality (Gronau (1982); Levy and Michel (1991); Cancian, Danziger, and
Gottschalk (1993); Blackburn and Bloom (1994); Topel (1994)). Additionally,
recent research has concluded that the distribution of women’s earnings has
actually had an egalitarian effect on household inequality. If women’s earnings
and labor force participation had been constant since the early 1970’s,
economic inequality would have been even higher than it has been (Gronau
(1982); Blau and Kahn (1994), Cancian, Danziger, and Gottschalk (1993); ).
However, considering earnings inequality alone ignores additional sources of
inequality related to household size and composition as well as labor force
participation. Burtless (1994) points to the importance of considering the
household as the context when determining levels of inequality. Burtless states:
In order to see the effects of changing earnings and employment
patterns on the overall income distribution, it is necessary to examine
earnings patterns among earners ranked by their fam ily incomes rather
than by their own individual earnings, (p. 133)
Inequality in total household income will capture the effects of changes in
household labor force participation on inequality, this includes the effects of
observed changes in women’s labor force participation and women’s earnings.
To go a step further, one would examine inequality in equivalent
income, as measured by IAE in chapter 2. Measures of equivalent income
capture not only the effects of earnings and labor force participation in
inequality, but also the effects of changes in household composition (including
family type and composition, Karoly (1992)). Inequality in IAE will differ
8 8
from inequality in earnings or total household income inasmuch as the
components unique IAE have independently affected inequality.6 1 In chapter
2, IAE was used as the most appropriate measure of economic status because it
captures the joint effects of total household resources and total household
demands on individual economic status. For the same reasons, IAE is the
economic measure employed when examining inequality.
IAE Inequality Among Baby Boomers
The reason we turn to economic inequality in this chapter is to scrutinize
the results of chapter 2. The results in chapter 2 found that across race, baby
boomers experienced median levels of economic status substantially greater
than their parents. Those estimates may not represent the economic experience
of individuals at the upper and lower tails of the distribution. Chapter 2 also
identified earnings growth as a relatively stronger source of baby boomer gains
for blacks, and identified household demographic and labor force changes as
the stronger element in the source of baby boomer gains for whites. If
changing inequality in IAE has a stronger effect on earnings than other
components of IAE, then race specific trends in IAE inequality may have
6 1 Differences in the inequality of IAE compared to other measures of
economic status are discussed further in following sections.
89
differed significantly. If this was the case, the experiences of baby boomers at
the lower ends of the distribution would also differ by race.
We now ask a specific question: Has the race specific level of inequality
in IAE between 1965 and 1990 increased in such a manner as to render
median-based results unapplicable to baby boomers at lower ends of the
distribution? The empirical results below will show that the answer is no, for
both black and white baby boomers.
B. Alternative Equivalency Scale Specification
Chapter 2 examined the economic status of baby boomers relative to
their parents. Economic status was measured by IAE, one form of equivalent
income. The concept of equivalent income relates to the adjustment of total
household income to account for the effects of variations in age/size
composition on the demand structure of the household. Under this concept, a
higher level of income in a larger household could be considered "equivalent"
to a lower level of income in a smaller household. This is because in either
case, total household resources applied to total household demand result in
equivalent economic well-being per person.
Any equivalent income transformation requires the application of a
specific equivalency scale. The equivalency scale represents the particular
structural assumptions made about the connection between age/size composition
90
and household demand. In Chapter 2 the equivalency scale utilized was first
proposed by Fuchs (1986).6 2 Over the years many different equivalency scale
specifications have appeared in economic literature. Previous efforts to develop
equivalency scales include Orshansky (1965), Lazear and Michael (1988),
Buhmann, Rainwater, Schmaus, and Smeeding (1988), Ruggles (1990), Mertz,
Gardner, Smeeding, Faik, and Johnson (1993), and Phipps and Gardner (1994).
Despite the different considerations and issues these studies have addressed, no
single equivalency scale has emerged as superior. The reason for this lack of
consensus is stated quite clearly by Ruggles: "It is easier to identify the
problems with current [equivalency scale] adjustments than it is to design a
new set." (p. 73). Incidentally, one argument regarding proper use of
equivalency scales disputes the existence of any "best" scale, but instead
supports the use of different scales for different purposes.
F our Alternative Equivalency Scales
Since there are infinite possible equivalency scale specifications, the
choice of one scale is a constraint, but is it a binding constraint? In this
chapter we relax this constraint and compare outcomes when four different
equivalency scale specifications are implemented. The four equivalency scales
we adopt are: 1) that proposed by Fuchs (1986) and used in chapter 2, 2) a per
6:For specifics on Fuchs’ scale, see appendix to chapter 2.
capita scheme, 3) the scheme implicit in the official US poverty thresholds, and
4) A parametric scheme similar to that of Buhmann et al (1988). The
discovery of a "best" equivalency scale is certainly beyond the scope of this
study. We can however estimate the "cost" of misspecifying the equivalency
scale. That is to say, while we do not identify a single "best" scale, we apply
several different scales and examine the stability of empirical estimates across
alternative specifications. The aim of this exercise is to demonstrate that the
essence of the results in chapter 2 are robust to alternative equivalency scale
specifications, and that measures of inequality in this chapter are relatively
stable across specification.
Fuchs’ Equivalency Scale The first equivalency scale we discuss was
outlined by Fuchs (1986) and used in the analysis of chapter 2. Fuchs’ scale
was developed to systematically incorporate two specific concepts in the
measurement of household demand. The first is the concept that individual
demand varies considerably by age. Fuchs asserts that a base level of adult
demand is larger than for a child. Thus, in a structure where individual
weights are representative of their demand, and if the demand of a single adult
is set equal to 1.0, then the weight of a child is less than 1.0 but greater than
zero. The second specific concept incorporated into Fuchs’ weighing structure
is the existence of household economies of scale. A household exhibits
economies of scale if adding individuals increases total demand less than
92
household size. Fuchs presumes that economies of scale do exist both in the
number of adults and in the number of children. That is, the additional demand
of each adult beyond the first is only a fraction of the first adults' demand.
Similarly, the demand of each additional child (beyond the first) is only a
fraction of the first childs’ demand.
The specific structure of Fuchs’ equivalency scale is not complicated.
The first adult in a household is weighted at 1, and every additional adult is
weighted at .8. This implies that when household size exceeds one, each new
adult can maintain the economic status of the original adult at 80% of what it
cost the first adult. A similar discounting scheme was employed for calculating
child weights. Additional demand associated with adding the first child is
assumed to be 40% of the first adult, thus the first child is weighted at .4. As
with additional adults, it is assumed that if additional children are added to the
household, the economic status of the new children can equal that of the first
child with 25% less income. Hence, every child after the first is weighted at
63 64
■ J.
6jIn three examples of Fuchs’ scale, the demand of two adults and two
children in one household is equivalent to the demand of 2.5 adults living alone,
and, the demand of six adults living in one household is equivalent to the demand
of five adults living alone, and, the demand of one adult and three children in one
household is equivalent to the demand of two adults living alone.
wAs in the previous chapter, a child is defined as any unmarried person less
than 16 years of age, not living alone.
The Per Capita Equivalency Scale The first alternative equivalency
scale we turn to is not always recognized as an equivalency scale. We are
referring to the equivalency scale implicit in a per capita representation. The
use of the per capita representation is a wide spread and long standing practice
across disciplines and institutions. The term per capita comes directly from the
latin phrase "by head", which is closely related to its equivalency scale
implications.6 ' A per capita transformation adjusts any given total based on a
number of related persons or "heads", without regard to age or scale. Thus, the
equivalency scale applied by the use of a per capita transformation assumes no
economies of scale and equal demand across all ages. In the per capita scheme,
all individuals are given the weight of l.O.6 6
The Official US Equivalency Scale The second alternative equivalency
scale utilized could be referred to as the Official US equivalency scale. In
reality, there is no official US equivalency scale, but there is a set of Official
US Poverty thresholds, and the determination of these poverty thresholds
requires the use of an equivalency scale. So in a sense, this assumed
equivalency scale is an official US equivalency scale.6 7
“ Oxford Encyclopedic English Dictionary, 1991.
“ Sometimes a per capita representation only considers adults. In this case,
there is still no allowance for economies of scale, and children are assumed to be
costless.
6 7 To be clear, the equivalency scale implicit in the set of Official US Poverty
94
The Official US poverty thresholds were first developed by Orshansky
(1965) for the purpose of determining poverty lines for families in the US,
accounting for age/size variations across families as well as region and
farm/non-farm residence. Orshansky’s thresholds were based on the cost of the
minimum adequate diet for a household. In 1969, the US Bureau of the
Budget implemented Orshansky’s poverty thresholds as Official US Poverty
Thresholds. The Official US Poverty Thresholds have since been updated,
omitting differentials by region and farm/non-farm residence, and including
differentials for the elderly.6 8 These thresholds have Orshansky’s equivalency
scales embedded them. If differences in poverty thresholds are defined by the
age/size composition of the family, it is assumed that household demand differs
by age/size composition, and an equivalency scale is implicit.6 9
The set o f Official US Poverty Thresholds for 1992 includes 50
thresholds accounting for family sizes from "one" to "nine or more", and for
Thresholds is not officially used for inferences about the full distribution of
income, but only the very low end, at the point of poverty. Use of the weighing
scheme across the entire income spectrum is in fact not official.
6 S In the development of the Official US Poverty Thresholds, the US Bureau
of the Census account for elderly persons only if they are the head of the
household, and only if total family size is less than three.
6 9 The Official US Poverty Thresholds, and thus the implicit equivalency scales,
refer specifically to variations in the compositions of families. I assume that if
multiple families reside with one household, cost-cutting and sharing behavior
exists within that household. Thus, I treat all individuals in a household as one
family.
95
number of children from "one" to "eight or more" (see table 3.1). Indexing the
50 thresholds to the "under 65" one person threshold ($7,299) transforms the
thresholds to a set of 50 weights, which together constitute a specific
equivalency scale (see table 3.2). As the tables illustrate, there is no systematic
function for relating additional demands to additional persons, whether children
or adults. The set of weights represent total "living-alone adult" equivalents for
the household. The full set of weights in table 3.2 is used as the official US
Poverty Line equivalency scale.
The Parametric Equivalency Scale The last alternative equivalency
scale to be employed is one application of a class of "parametric" equivalency
scales. Scales in this class have a common functional form, and are adjustable
by specification of one or more parameters which reflect the scale rate of
elasticity by household size. Variation in parameter(s) values represent
different assumptions as to scale rates across households. One characteristic of
this class of scales is the smooth relation between equivalent income and
household size. One advantage to this class of scales is that they can be
specified to approximate other non-parametric equivalency scales by
adjustments in parameter values.
One type of the parametric class of equivalency scales was developed by
Buhmann et al. (1988) and incorporates a single parameter (e) as the elasticity
of the scale rate to household size. The scale is defined by the function:
(2.1) ES = HHI / HHSe
96
where ES is the economic status of the household7 0 , HHI is total household
income, HHS is household size, and (e) is the single elasticity parameter. The
value of e falls somewhere between 0 and 1, where e = 1 implies no economies
of scale (this implementation would be identical to a per capita representation)
and e = 0 implies perfect economies of scale (all persons in the household
beyond the first are costless). Note that this parametric scale ignores the ages
of individuals in the household.
Buhmann et al. have estimated the overall scale elasticity in other
equivalency scales, and have reported that a single parameter parametric scale
(such as their own) could be specified to approximate a smooth version of the
other scales. As pointed out by Ruggles (1990), the estimates of overall scale
elasticity in other "expert" non-parametric equivalency scales fall at or about .5.
For use in this study, we apply the Buhmann parametric specification with e =
.5. This results in a "smooth" functional form with a scale elasticity close to
other "expert" non-parametric equivalency scales.
7 0 Through this chapter, economic status is determined at the household level,
and every person in that household is assumed to have the same level of individual
economic status. This treatment implies the assumption of equal sharing in the
household.
97
In chapter 2 we found that both black and white baby boomers have
experienced median levels of IAE substantially higher than pre-boomer "parent"
cohorts of the same age. The validity of these results may lie in the sensitivity
to the equivalency scale assumption. If some significant but unobserved
component of household economic status is overlooked or misconstrued by
certain equivalency scale assumptions, then the specific choice of equivalency
scale becomes essential to accurate results. Thus, we proceed by re-estimating
previous results under these three alternative equivalency scale specifications,
addressing levels of baby boomer gains in IAE and IAE inequality.
III. METHODOLOGY7 1
A. IAE Inequality
To investigate the impact of changing IAE inequality on the economic
status of baby boomers three separate procedures are employed. First, we
calculate Gini coefficients by year and race as an overall measure of IAE
inequality. Second, we compare race specific levels of IAE for baby boomers
and their parents at the 75th and 25th percentiles.7 2 These comparisons
7 1 The data used in this chapter is that used in chapter 2; see chapter 2 section
III.A for a discussion.
7 2 The main interest here is really at the lower ends of the distribution, but we
98
establish the relative IAE gains of baby boomers at upper and lower tails of the
IAE distribution. Third, we identify the proportions o f blacks and whites with
IAE in the top and bottom 10% of the overall distribution. Obtaining these
measures from 1965 to 1990 will depict race representation at extreme ends of
the distribution for pre-boomer parents up to the time when the baby boomers
were the same age.
Gini coefficients are obtained for all persons aged 25-44, as well as for
blacks and whites separately, for quinquennial years between 1965 to 1990.7 j
The trend in the all-person or "overall" Gini shows what has happened to the
distribution of IAE over the period. Notice that persons aged 25-44 in 1965
are the baby boomers’ parents and persons aged 25-44 in 1990 are the baby
boomers. The Gini for blacks (or whites) measures inequality in the
distribution of IAE among blacks (or whites) . Comparing a race specific Gini
to the overall Gini emphasizes the relative distribution of that race. For
example, if Gini coefficients for whites are lower than overall Ginis, then white
IAE is not as unequally distributed as blacks.
The second aspect of inequality examined is the relative gains in baby
boomer IAE at the 75th and 25th percentiles. The examination procedure is
also examine the upper end for comparison.
7 3 Gini coefficients for IAE are determined by the construction of Lorenz
curves based on decile tabulation.
99
nearly identical to comparisons of IAE in chapter 2, the only difference lies in
the IAE statistic examined. In chapter 2 the metric examined was median IAE
of 5-year birth cohorts from 1910-14 through 1960-64, and here we look at
75th and 25th percentiles of the same cohorts. Results of comparisons in
chapter 2 were reported as percentage gains in median IAE of baby boomers
over their parents at the same age. We now determine whether baby boomers
made gains over their parents at the 75th and 25th percentiles, and if so, how
the percentage gains compare to those found at the median.
The last glance at IAE inequality focuses on the extreme ends of the
distribution, more specifically the highest and lowest 10%. A specific interest
is how changing inequality between 1965 and 1990 has effected the race
representation at the upper and lower tails. Using the whole 25-44 population
in quinquennial years from 1965 to 1990, we identify IAE at the 10th and 90th
percentile. For each year, we determine the proportion of blacks and whites
that fall below the 10th percentile and the proportion that are above the 90th
percentile. This procedure will shed light on race differences in the prevalence
and trends in impoverishment and prosperity.
B. Application of Alternative Equivalency Scales
The second objective of this chapter is to relax the equivalency scale
assumption in the determination of economic status, and examine the stability
100
of results under three alternative equivalency scale assumptions. First, we
revisit certain results of chapter 2 where the IAE of baby boomers was
compared to the IAE of their parents.7 4 We conduct the same analysis as in
chapter 2 changing only the equivalency scale specification. The questions to
be answered by these comparisons was whether baby boomers experienced
levels of equivalent income in excess of their parents, and if so, by how much;
this is the approach taken here as well. We obtain three more measures of
median equivalent income for 5-year birth cohorts by age and race, one using
the per capita equivalency scale, the second using the US poverty line
equivalency scale, and a third using the parametric equivalency scale. With
these three measures of equivalent income in hand, we obtain three alternative
estimates of baby boomers economic gains. These three new sets of estimates
are presented along with the analogous results of chapter 2 where Fuchs’
equivalency scale was employed.
Second, we turn to the estimates of inequality in IAE discussed earlier in
this chapter. If alternative equivalency scale specifications result in different
levels and trends of equivalent income, then they may also result in different
7 4 IAE, or income per adult equivalent, is an adjusted income measure that
accounts for age size variation in household size by applying a certain equivalency
scale. Changing the equivalency scale assumption still results in a measure of
equivalent income and can still be referred to as IAE, but the levels of the
measure would be different. In this chapter we refer to measures of equivalent
income by the specific equivalency scale used, such as "Fuchs’ scale" or "Per
capita scale".
101
levels and trends in inequality. To examine this possibility, we calculate three
more sets of Gini coefficients, on for each alternative equivalency scale. These
Gini coefficients will report estimates of inequality in equivalent income,
overall and by race.
IV. RESULTS
A. IAE Inequality
IAE Gini Coefficients
Gini coefficients indicate that inequality in the distribution of equivalent
income for all individuals aged 25-44 has increased over the 1965 to 1990
period (see figure 3.1 and table 3.3). The overall Gini coefficient actually
decreased from .328 to .311 between 1965 and 1970, but increased thereafter to
.358 in 1990. The increase in inequality between 1970 and 1990 was only
slight with the exception of a dramatic increase between 1980 and 1985.
Inequality among whites is slightly lower than for all persons, and the trend is
almost identical.7 5 The white Gini coefficient was .317 in 1965, dropped to
7 3 The similarity in trend should not be surprising given that whites make up
approximately 86% of the sample.
.302 in 1970, then increased steadily to .347 in 1990. Inequality among blacks
is somewhat higher than for all persons in every sample year from 1965 to
1990, and the trend is slightly different. The black Gini coefficient started at
.384 in 1965, fell to .359 in 1970, and increased thereafter to .397 in 1990.
Increases in the black Gini coefficient between 1970 and 1990 were smaller
and more stable than among whites, with 1970 to 1975 being the only period
showing an exaggerated increase. The ratio of black to white Gini coefficients
showed no consistent trend between 1965 and 1980, but from 1980 to 1990 the
ratio dropped slightly, indicating that white inequality grew somewhat faster
over this period (also table 3.3).
Recall that the age range of 25-44 in 1990 captures all the baby boomers
and only baby boomers. The same age range in 1965 represents their pre
boomer parents. It appears that inequality in the distribution of IAE has
increased substantially between the generations. This is also the case among
blacks and whites separately. What remains to be seen is how this increase in
inequality affected baby boomer gains in IAE at upper and lower ends of the
distribution.
IAE at the 75th and 25th Percentile
The next procedure evaluates the effect of rising inequality on baby
boomer IAE gains. Using the 75th percentile of IAE for 5-year birth cohorts,
103
we first constructed a set of 75th percentile IAE profiles for blacks and for
whites (see figure 3.2). From these profile sets, we extracted eight profile pairs
(four for blacks and four for whites) to compare baby boomer profiles to pre
boomer parent profiles (see figure 3.3 through 3.6).7 6 It is clear from these
figures that both black and white baby boomers have made substantial gains in
75th percentile IAE over the pre-boomer parent cohorts. As expected, IAE
gains at the 75th percentile which ranged from 85.5% to 117.4% for blacks and
from 77.3% to 103.6% for whites are larger than gains at the median, which
ranged from 75.7% to 107.3% for blacks and from 62.7% to 73.5% for whites
(see table 3.4). Just as for comparisons at the median, black baby boomers
have made larger gains than whites.
We next examined IAE at the 25th percentile for baby boomers and their
parents. We constructed a set of IAE profiles, this time at 25th percentile, by
race (see figure 3.7). Clearly, both black and white 25th percent profile sets
are more compressed than the 75th percentile sets in figure 3.2. We extracted
four profile pairs for blacks and four for whites to make comparisons between
baby boomer and parent cohorts (see figures 3.8 through 3.11). We find that
baby boomers, even at the 25th percentile, have made considerable gains in
IAE over their parents. Baby boomer gains at the 25th percentile are on
7 6 The baby boom generation spans 20 years, so if 5-year cohorts have been
defined then four different cohorts are baby boomer cohorts. Comparison of baby
boomers to their parents, therefore, requires four sets of comparisons.
104
average slightly lower than gains at the median (see table 3.4). Gains at the
25th percentile for black baby boomers ranged from 59.8% to 102.9%, for
whites the range was 49.0% to 65.3%.7 7 Thus, by comparing the specific IAE
improvements by baby boomers relative to their parents at the 75th and 25th
percentiles, we see that increases in IAE inequality resulted in larger IAE gains
at the 75th percentile and lower gains at the 25th percentile. The effects of
increased IAE inequality were not strong enough to keep either black of white
baby boomers from experiencing substantial gains in IAE over their parents,
even at the 25th percentile.
Race Proportions at the top and bottom decile of IAE
The last means by which we explore the effects of increasing IAE
inequality is by determining the proportion of blacks and whites aged 25-44 in
the top and bottom 10% of the IAE distribution for all persons aged 25-44.
This approach shows the differences in how IAE inequality has effected the
levels and trends o f race representation at the extreme ends of the IAE
distribution (see table 3.5). The findings show that blacks are consistently
underrepresented in the top 10% and overrepresented in the bottom 10%; the
opposite is true for whites, who are consistently overrepresented in the upper
7 7 Note that with one exception, IAE gains by black baby boomers at the 25th
percentile were larger than white baby boomer gains at the median.
105
10% and underrepresented in the lower 10%. With regard to the upper 10%.
little has changed from 1965 to 1990 for blacks or for whites. Exactly 3% of
blacks aged 25-44 in 1990 found themselves in the top 10% of the IAE
distribution for all persons aged 25-44. This figure increased to about 4.7% in
between 1970 and 1980. but fell again in 1985 to 3.6% where it stayed through
1990. In 1965, 10.8% of whites aged 25-44 were in the top 10%, and this
proportion was stable through 1990 when an identical 10.8% of whites were in
the top 10% of the IAE distribution.
Something a bit more interesting occurred at the lower 10% of the IAE
distribution. Between 1965 and 1990 and for individuals aged 25-44, there was
a substantial decrease in the proportion of blacks in the lowest IAE decile
(from 30.6% to 23.1%). This occurred over a period when the proportion of
whites in the lowest 10% increased slightly (7.5% to 8%). So, during a time
period when overall inequality in the distribution of IAE was increasing for
those aged 25-44, blacks aged 25-44 were improving their position relative to
whites, albeit moderately. Similar changes have occurred in race specific
poverty rates over the 1965 to 1990 period, where the proportion of blacks and
whites below the poverty line decreased from 41.8% to 31.9% and from 9.9%
to 10.7%, respectively.7 8 Keep in mind that changes in these proportions are
only relative and do not preclude a concurrent increase in inequality. A
7 S U.S. Bureau of the Census, Current Population Reports, P60-185.
106
appropriate view of these figures would be that increases in inequality were
relatively larger for whites than blacks, and thus blacks were less frequently
found in the bottom decile of IAE in 1990 than in 1965.
With the exception of fewer blacks in the bottom decile of IAE, the
relative proportion of black and white baby boomers in the top and bottom
10% of the IAE distribution was mostly unchanged since 1965. This finding
lies in contrast to earlier proposed concerns that black baby boomers may have
experienced relative increases in IAE inequality over the 1965-90 period.
Nonetheless, blacks remain underrepresented in the top decile of IAE and
heavily over-represented in the bottom 10% as of 1990.
B. Stability of Results Under Alternative Equivalency Scale
Specifications
The second aim of this paper is to examine the stability of estimates of
economic status across competing equivalency scale assumptions. The
measurement of IAE in Chapter 2 and of IAE inequality in section IV.A above
was based on a particular equivalency scale specification, thus the findings rely
on the validity of the assumed scale. By relaxing the specific equivalency scale
assumption and applying three alternative equivalency scales, we effectively
obtain four alternative measures of equivalent income and inequality in
107
equivalent income. This section reports on differences and/or similarities in
these alternative measures.
Trends in Alternative Measures of Equivalent Income
The trends in the four measures of equivalent income have been
surprisingly similar across race (see fig 3.12 and table 3.6). By any of the four
measures, blacks and whites aged 25-44 experienced modest growth in
equivalent income from 1965 to 1980, stagnancy or slight declines between
1980 and 1985, and modest growth again from 1985 to 1990. With respect to
each measure of equivalent income, blacks have consistently lower levels than
whites over the entire period. Equivalent income based on US Poverty Line
and Parametric equivalence scales are almost indistinguishable for blacks and
for whites. IAE by use of Fuchs’ scale (that used in chapter 2 and referred to
as IAE) falls consistently in the middle o f the group for both blacks and whites.
Equivalent income by use of the per capita scale (hereafter referred to as per
capita income) is consistently the lowest measure.
The relative levels of these equivalent income measures relate directly to
the assumptions of each specification. Both the US poverty line scale and the
parametric scale assume very large economies of scale, Fuchs’ scale assumes
modest economies of scale, and per capita income assumes no economies of
scale. The larger the economies of scale, the higher the purchasing power of a
108
given amount of income for household size greater than one. For example,
assume that economies o f scale are such that two adults in a household are
equally well off as one adult living alone when the two individuals have 80%
more total household income than the adult living alone.7 9 Under this
assumption, S18,000 in a two adult household goes just as far as $10,000 in a
one person household, and all three individuals have equivalent incomes of
$10,000.8 0 But, if the economies of scale assumption dictates that two adults
need only 40% more than a living-alone adult to be equally well off, then
$18,000 for two is more than $10,000 for one. In this case, $18,000 and
$10,000 of total household income relate to $12,857, and $10,000 of equivalent
income, respectively. Thus, equivalency scales with higher assumed economies
of scale naturally result in higher average equivalent income.
The levels of equivalent income in figure 3.12 are associated with
individuals in the entire 25-44 age range. To compare baby boomer cohorts
with their parent cohorts, we compare similar figures by age group in 1965 and
1990.
"This is the same as Fuchs’ scale.
S 0 If a second adult in a household increased demand by 80%, the weight for
the second adult would be .8. Thus the number of adult equivalent in a two adult
household is 1.8, and $18,000 divided by 1.8 = $10,000. The total number of
adult equivalents for a one person household is of course 1, and total household
income = equivalent income = $10,000.
109
Estimates of Gains in Equivalent Income Across Equivalency Scales
In chapter 2, equivalent income of baby boomers, referred to as IAE,
was compared to that of their parents was by comparing median IAE of a 5-
year baby boomer cohort to that of a 5-year pre-boomer cohort bom on average
25 years earlier. Results in chapter 2 showed that median IAE of baby
boomers exceeded that of pre-boomer parents; growth in IAE ranged from
62.7% to 107.3%, depending on race and cohort. These results were based on
the use of Fuchs’ equivalency scale specification. Our current interest lies in
the sensitivity of these results to that equivalency scale restriction. Therefore,
we performed the same comparison three more times using three alternative
equivalency scales (see table 3.7).
The results coming from the four separate equivalency scale
specifications are similar in many ways. Under all four scales, black and white
baby boomers have experienced significant gains in equivalent income over
their parents, blacks to a larger extent. Despite the equivalency scale
employed, the gains in equivalent income are larger for the two oldest baby
boomer cohorts, across race. In three of the four equivalency scale
specifications (Fuchs’ scale, the parametric scale, and the US poverty line
scale), the percentage gains are also similar, falling between 72.6% and 111.8%
for blacks and between 60.7% and 73.9% for whites. Only the percentage
gains in equivalent income based on the per capita equivalency scale were
110
clearly higher than the rest, falling between 101% and 132% for blacks and
between 83.7% for whites. One reason that gains in equivalent income were
larger by the per capita scale than by other scales is that the per capita scale
assumes no difference in the demand of adults and children, and assumes no
economies of scale. As outlined above, these characteristics of the per capita
scale resulted in lower relative levels of equivalent income. Since the per
capita specification counts children as adults and allows no demand "discount"
for additional scale, reductions in children and/or in household size have an
even larger positive effect on equivalent income, and percentage gains appear
larger. Withstanding the larger gains in equivalent income rising from the use
of the per capita scale, the conclusion drawn from the comparison of alternative
equivalency scales is clear. By 1990, black and white baby boomers made
significant gains in equivalent income relative to their parents. The percentage
gains in equivalent income by black baby boomers were consistently larger than
for whites. After relaxing the equivalency scale assumption and employing
three competing scales, we found that these conclusions are in fact robust to
equivalency scale specification.3 1
8‘The only significant difference arising from alternative scales came from the
per capita scale, and the result was even larger gains.
Ill
Estimates of Equivalent Income Inequality Across Equivalency Scales
Just as levels of equivalent income are influenced by equivalency scale
specification, so too are measures of inequality. Section IV.B above discussed
measures of IAE inequality, but these measures, like the first estimates of IAE
gains in chapter 2, were based on Fuchs’ equivalency scale specification. In
this section we report on three more evaluations of inequality, utilizing the
same three alternative equivalency scales.
We compare measures of inequality across equivalency scales by
calculating three more sets of Gini coefficients. As in section IV.B, these Gini
coefficients are determined for all persons (and for blacks and whites
separately) aged 25-44 (see table 3.8).8 2 For all equivalency scale
specifications, the trends in Gini coefficients show common traits. The Gini
coefficients fall between 1965 and 1970, increase significantly between 1970
and 1985, then increase only slightly (if at all) from 1985 to 1990. Regardless
of the equivalency scale used, inequality among blacks is consistently higher
than among whites. As was the case for gains in equivalent income, inequality
in equivalent income is similar among three of the four equivalency scales.
Only the per capita scale resulted in significantly different measures of
inequality. Overall Gini coefficients obtained using the Fuchs, parametric, and
8 2 Recall that this age range is considered so that the sample in 1990 is the
baby boomers and the sample in 1965 is their parents.
112
US poverty line equivalency scales fall between .324 and .340 in 1965, fall to
between .305 and .319 in 1970, and increase through 1990 to levels between
.351 and .359. When the per capita scale is used, the overall Gini coefficient is
.381 in 1965, drops to .363 in 1975, and increases through 1990 to .399. A
main reason for the difference in inequality using the per capita scale comes
from ignoring economies of scale and age differences in demand. When
economies of scale and the age composition of the household are ignored - as
is the case with the per capita scale - larger households and households with
children are overrepresented among the poor, and a larger poor population
results. But, small families and families without children are relatively
unaffected by economies of scale or age difference in demand issues, thus
equivalent income under the per capita scale is similar to other scales for these
households. The result is a larger distribution of equivalent income, clearly
shown in table 3.8.
Even though one of the three alternative equivalency scales resulted in
higher measures of equivalent income inequality, the trends in inequality across
scales have followed the same general pattern: a decrease from 1965 to 1970, a
slight increase from 1970 to 1980, a substantial increase from 1980 to 1985,
and a slight increase (if at all) from 1985 to 1990. Additionally, race
differences in Gini coefficients are resilient to equivalency scale specification.
The black/white ratio in Gini coefficients, by reported by equivalency scale, are
113
evidence of this (also table 3.8). The black to white ratio of the Gini
coefficient in 1965 was about 1.213 (between 1.211 and 1.215), and by 1990 it
had dropped to roughly 1.135 (between 1.121 and 1.150). It is clear that race
differences in IAE inequality and the trend in inequality across race are robust
to equivalency scale specification among the scales examined.
Results have shown that analysis of baby boomer equivalent income,
whether comparing gains over their parents, measuring inequality, or examining
race differences in either, is insensitive to equivalency scale specification.
Utilization of three alternative equivalency scales failed to lead to significant
differences in results or to changes in conclusions based on Fuchs’ equivalency
scale, the scale used at the onset.
V. CONCLUSION
This analysis has established that the inequality in the distribution of IAE
for individuals aged 25-44 has increased significantly between 1965 and 1990.
Inequality was consistently higher among blacks than whites, but the
black/white ratio of inequality decreased over the period.
Results in chapter 2 showed that median IAE of both black and white
baby boomers exceeded that of their parents at the same age. One concern
114
about the rise IAE inequality is the effect it may have on baby boomers at the
lower end of the distribution. Empirical findings in this chapter found that
despite the increases in inequality over the period, both black and white baby
boomers experienced substantial gains in IAE even at the 25th and 75th
percentiles. Additionally, it was shown that the race proportions in the lowest
and highest 10% of the IAE distribution were mostly unchanged: in 1965 and
1990 blacks are heavily overrepresented in the lowest 10% and
underrepresented in the top 10%, and vice versa for whites. The only
significant change in race representations in these deciles was a decrease in the
proportion of blacks in the lowest 10% from 30.6% in 1965 to 23.1% in 1990.
Whatever the effects of rising IAE inequality have been on the baby
boomers, they haven’t effected gains IAE at the 25th and 75th percentiles
dramatically different than at the median. Estimated race proportions at the
bottom and top deciles of IAE in 1965 and 1990 show minimal race differences
in the effect of IAE at the distant ends of the IAE distribution, with the
exception of decreasing proportions of blacks in the lowest decile.
The measurement of equivalent income requires the specification of an
equivalency scale. This scale represents the assumptions made regarding
household economies o f scale and age differences in demand structure. The
measurement of equivalent income in chapter 2 and the initial analysis of
inequality in this chapter relied on the use of one equivalency scale, that
115
proposed by Fuchs (1986). Since the validity of results in these sections may
rely on the equivalency scale specification choice, the second part of this
chapter focused on the effects of utilizing three other competing equivalency
scales; a per capita scale, the scale implicit in the official US poverty
thresholds, and a parametric scale.
The levels of equivalent income from 1965 to 1990 were sensitive to
equivalency scale specification, but the trends were not. The differences in the
levels of equivalent income across equivalency scale are largely attributed to
assumptions about economies of scale and age differences in demand. The
trend in equivalent income, despite equivalency scale, was steadily increasing
over the 1965 to 1990 period. It was shown that the percentage gains in
equivalent income of black and white baby boomers over their parents was
quite similar across equivalency scales. Only one scale, the per capita scale,
gave significantly different results, and these results pointed to even larger
gains than with the other three equivalency scales.
Also examined was the effect of equivalency scale specification on the
measurement of equivalent income inequality. As with baby boomer gains,
three of the four scales resulted in similar measures of inequality; only the per
capita equivalency scale implied that inequality was higher than when using
other scales. The per capita scale results in higher estimates of inequality
primarily due to assumptions of no economies of scale and no age difference in
116
demand. Such assumptions result in the overrepresentation of large families
and families with children among the poor, and a wider distribution of
equivalent income.
The use of alternative equivalency scales changes the results relatively
little. More importantly, the conclusions tied to the specific interest in chapter
2 - the economic status of baby boomers relative to their parents - are
supported by all four of the equivalency scales employed. The analysis of
equivalent income seems mostly insensitive to equivalency scale specification,
and the results pertaining to baby boomer gains, across race, are robust.
117
TABLE 3.1 : OFFICIAL U.S. POVERTY THRESHOLDS, 1992
Family Unit Size Related Children Under IB Years
(Persons)
Eight
None One Two Three Four Five Six Seven or more
One 7143
under 65 7299
65 or older 6729
Two 9137
head under 65 9395 9670
head 65 or older 8480 9634
Three 10974 11293 11304
Four 14471 14708 14228
Five 17451 17705 17163
Six 20072 20152 19737
Seven 23096 23240 23743
Eight 25831 26059 25590
Nine or more 31073 31223 30808
14277
16743 16487
19339 18747 18396
22396 21751 20998 20171
25179 24596 23855 23085 22889
30459 29887 29099 28387 28221
Source: Current Population Reports, series P-SO
(1992$)
118
TABLE 3.2 : ADULT EQUIVALENCY SCALE IMPLICIT IN OFFICIAL
U.S. POVERTY THRESHOLDS
Family Unit Size Related Children Under 18 Years
(persons)
Eight
None One Two Three Four Five Six Seven cr more
One 0.98
under 65 1.00
65 or older 0.92
Two 1.25
head under 65 1.29
head 65 or older 1.16
Three 1.50
Four 1.98
Five 2.39
Six 2.75
Seven 3.16
Eight 3 .54
Nine cr more 4.26
1.32
1.55 1. 55
2 . 02 1.95 1.96
2.43 2.35 2 .29 2.26
2.76 2.70 2.65 2.57
3 .18 3 .12 3 . 07 2.98
3.57 3.51 3 .45 3.37
4.28 4 .22 4 .17 4 . 09
2 . 52
2.88 2.76
3 .27 3 .16 3 .14
3 . 99 3 .89 3 .87 3
Gini Coefficients
FIGURE 3.1
119
GINI COEFFICIENTS FOR IAE: persons aged 25-44
Blacks 25-44
All persons 25-44
Whiles 25-44
1965 1975 v 1980
Y ear
1970 1985 1990
120
TABLE 3.3 : GINI COEFFICIENTS FOR IAE
Persons Aged 25-44
1965 1970 1975 1980 1985 1990
All Individuals 0.328 0.311 0.319 0.321 0.354 0.358
Blacks 0 .384 0.359 0.379 0.378 0.395 0.397
Whites 0.317 0.302 0.308 0.310 0 .344 0.347
Black/White Ratio 1.211 1.189 1.231 1.220 1. 148 1.144
25th percentile Income Per Adult E quivalent ($1988)
FIGURE 3.2
25th PERCENTILE IAE: blacks aged I5-6‘I by cohort, 1965-1990
ratio scale
15000-
10000-
5000-
( 10)
(ID
15-19 20-24 25-29 JO-34 J5-39 40-44 45-49 50-54 55-59 60-64
A ge G ro u p
raiio scale
I500CH
25th PERCENTILE IAE: whiles aged 15-64 by cohort, 1965-1990
0 ) i
(2 ) <
u
- 50004
(5) 16) (7) (8) < 9>
4
(4)
15-19 20-24 25-29 30-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60-64
( 1 ) 1 9 0 0 -0 1 ( 2 ) 1955-511 ( 3 ) 1 9 5 0 -5 1 ( 1 ) 1 9 1 5 -1 9
( 5 ) 1 9 1 0 -1 1 ( 6 ) 1 9 3 5 - 3 9 ( 7 ) 1 9 3 0 -3 1 ( 8 ) 1 9 2 5 -2 9
( 9 ) 1 9 2 0 -2 1 ( 1 0 ) 1 9 1 5 -1 9 ( 1 1 ) 1 9 1 0 -1 1
25th percentile Income Per Adult E quivalent ($1988)
(N
(N
FIGURE 3.3
25th PERCENTILE IAE: black 1960-64 baby boomer and 1935-39 pre-boomer cohorts 25th PERCENTILE IAE: white 1960-64 baby boomer and 1935-39 pre-boomer cohorts
n u o tcaJe , J *
ratio sciie
15000-
1 0 0 0 0 -
£ 1 0 0 0 0 -
{1960-64)
(1935-39)
(1960-64)
20-24 25-29 30-J4 35-39 40-44 45-49 50-54 55-59 60-64
Age Group
15-19 20-24 25-29 30-54 35-39 40-44 434 9 50-54 55-59 60-64
A ge G ro u p
co
2 FIGURE 3.4
25th PERCENTILE IAE: black 1955-59 baby boomer and 1930-34 pre-boomer cohorts 25th PERCENTILE IAE: white 1955-59 baby boomer and 1930-34 pre-boomer cohorts
n iio tc iJ e ratio icale
13000- 15000-
ao 10000- 30 10000-
(1930-34)
5000-
(1930-34)
15-19 20-24 25-29 30-34 35-39 40-44 45-19 50-54 55-59 6 0 * 4
A ge G r o u p A ge G ro u p
25th percentile Income Per Adult Equivalent ($1988)
3 FIGURE 3.5
25th PERCENTILE IAE: black 1950-54 baby boomer and 1925-29 pre-boomer cohorts
ratio sciie
25th PERCENTILE IAE: white 1950-54 baby boomer and 1925-29 pre-boomer cohorts
ratio scale
(1950-54)
- 500 0 -
3
u
3
15-19 20-24 25-29 30-54 33-39 4 0 -«
Age G ro u p
50-54 55-59 60-64
(1925-29)
15-19 20-24 25-29 30^54 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60- 6*
3 FIGURE 3.6
25th PERCENTILE LAE: black 1945-49 baby boomer and 1920-24 pre-boomer cohorts
m io tctlc
25th PERCENTILE IAE: white 1945-49 baby boomer and 1920-24 pre-boomer cohorts
ratio »aUe
4
(1945-49)
15-19 20-24 25-29 30-34 35-39 40-44 45-49
Age G ro u p
W
- 5 0 0 0 -
4
( 1 9 4 5 - 4 9 )
15-19 20-24 25-29 50-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60-64
126
TABLE 3.4 : IAE LEVELS OF BABY BOOMERS AND THEIR
PARENTS
Persons Aged 25-44 by Age Group
Age Group
25-29 30-34 35-39 40-44
MEDIAN IA E
BLACKS
WHITES
Baby boomers in 1990 9924 10144 11705 11897
Parents in 1965 5647 5662 5773 6151
Baby Boomer Gains (%) 75 .7 79.2 107.3 106.1
Baby boomers in 1990 15538 15702 16372 17219
Parents in 1965 9516 9845 9437 9958
Baby Boomer Gains <%) 63 .3 62.7 73 .5 72.9
25TH PERCENTILE IAE
BLACKS
WHITES
Baby boomers in 1990 5364 4937 5896 6228
Parents in 1965 3274 3089 3278 3069
Baby Boomer Gains (%) 63.8 59.8 79.9 102 . 9
Baby boomers in 1990 9854 10022 10496 11091
Parents in 1965 6614 6548 6550 6710
Baby Boomer Gains (%) 49.0 53 .1 60.2 65 .3
75TH PERCENTILE IAE
BLACKS
WHITES
Baby boomers in 1990 16048 16882 18706 19554
Parents in 1965 8651 8499 8606 9689
Baby Boomer Gains {%) 85.5 98.6 117.4 101. 8
Baby boomers in 1990 23378 23855 24577 25838
Parents in 1965 13129 13453 13594 14163
Baby Boomer Gains (%) 78 .1 77.3 80 . 8 103 .6
75th percentile Income Per Adult Equivalent ($1988)
rsi
FIGURE 3.7
ratio scale
35000-1
75lli PERCENTILE IAE: blacks aged 15-64 by cohort, 1965-1990
( 10)
< ! > * ,
(2) 4
(ID
(8 )
/
4
(4)
ratio scale
35000-j
30000-
00
0 0
25 25 0 0 0 -
4*
S
■= 20000-
75th PERCENTILE IAE: whites aged 15-64 by cohort, 1965-1990
3
c
V
U
u
&
(U)
( 10)
( 8 )
(7)
(G)
A ge G ro u p A ge G ro u p
( 1 ) I9G 0-G 1 ( 2 ) 1 3 5 5 -5 'J ( 2 ) 1 0 5 0 -5 4 ( 1 ) 1 9 1 5 -1 9
( 5 ) 1 9 1 0 -1 1 (G ) 1 9 3 5 -3 9 ( 7 ) 1 9 3 0 -3 1 ( 8 ) 1 9 2 5 -2 9
( 9 ) 1 9 2 0 -2 1 ( 1 0 ) 1 9 1 5 -1 9 ( 1 1 ) 1 9 1 0 -1 1
7 5 l h percentile Income Per Adult Equivalent ($19t& )
FIGURE 3.8
75th PERCENTILE IAE: black 1960-64 baby boomer and 1935-39 pre-boomer cohorts 75th PERCENTILE IAE: white 1960-64 baby boomer and 1935-39 pre-boomer cohorts
rmiio i d l e rauo scale
35000-.
5000-
4
(1960-64)
13-19 20-24 25-29 30-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60^64
3
2 25000-
■3 20000-
3 15000-
<
(1960-64)
15-19 20-24 25-29 30-34 35-39 4(M 4 45-49
A ge G ro u p
50-54 35-59 60-64
75th percentile income Per Adult E quivalent ($1988)
§ FIGURE 3.9
75lh PERCENTILE IAE: black 1955-59 baby boomer and 1930-34 pre-boomer cohorts
ratio scale
75th PERCENTILE IAE: white 1955-59 baby boomer and 1930-34 pre-boomer cohorts
ratio scale
4
(1955-59)
¥
(1930-34)
15-19 20-24 25-29 30-34 35-39 40-44 4 5 ^ 9
A ge G ro u p
50-54 55-59 60-64
2 25000-
4
(1955-59)
15-19 20-24 25-29 3CF34 35-39 40^44 45^49
A ge G ro u p
50-54 55-59 60-64
75th percentile Income Per Adult E quivalent ($1988)
O
~ FIGURE 3.10
75th PERCENTILE IAE: bhick 1950-54 baby boomer and 1925-29 pre-boomer cohorts
nlin in U ' * ratio scale
35U JO -
15-19 20-24 25-29 30-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 60-64
75th PERCENTILE IAE: white 1950-54 baby boomer and 1925-29 pre-boomer cohorts
ratio scale
35000-
30000-
s
25 0 0 0 -
4*
8
s 20 0 0 0 -
>
3
$
3
*o
15000-
E
0
a iOOOQ-
1
s
a
&
■ a
( l 950-54J
15-19 20124 25-29 30-34 35-39 4 044 454 9
A ge G ro u p
50-54 55-59 60-04
75th percentile Income Per Adult E quivalent ($1988)
FIGURE 3.11
75th PERCENTILE IAE: black 1945-19 baby boomer and 1920-24 pre-boomer cohorts
ratio scale
35000-t--- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- --- --- -- --- --- -- --- ----
30000-
75th PERCENTILE IAE: white 1945-49 baby boomer and 1920-24 pre-boomer cohorts
ratio scale
3 5 0 0 0 - 1 ---- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- --- -
3 0000-
4
(1945-49)
*5 25000-
a
• 5 20c
>
• 3
a '
v
3
a
w
e
&
4
(1945-19)
4
(1920-24)
Age G ro u p
15-19 20-24 30-34 35-39 40-44 45-49
A ge G ro u p
50-54 55-59 « W 4
132
TABLE 3.5 : RACE PRO PORTIONS AT HIGHEST AND LOW EST
DECILES
Persons Aged 25-44
1965 1970 1975 1980 1985 1990
BLACKS
below 10th percentile 30.6 28.7 28.0 26.0 23.8 23.1
above 90th percentile 3.0 4.7 4.7 4.8 3.6 3.6
WHITES
below 10th percentile 7.5 7.8 7.9 7.9 8.0 8.0
above 90th percentile 10.8 10.6 10.6 10.7 10.9 10.8
Median Equivalent Income ($1988)
FIGURE 3.12
EQUIVALENT INCOME BY FOUR EQUIVALENCY SCALES: Blacks aged 25-44
ratio sole
30000-r— — ■ -------- ---------------------------------------------------------------------------------------------------------------------------
25000-
20000-
T
1965 1975 1980
Year
US Poverty Thrwbold Scale
Parametric Scale
Fuchs' S o le
Per CapiutSok
~T~
1985
EQUIVALENT INCOME BY FOUR EQUIVALENCY SCALES: Whites aged 25-44
redo scale
30000-1-----------------------------------------------------------------------------------------------------------------------------------------—
25000-
20000-
s
9\
£ 1 0 0 0 0 -
s
>
'3
3
e
3
* 2
1975 1980
Y e a r
US Poverty Threshold S o le
Parametric S o le
Fuchs* S o le
TABLE 3.6 : EQUIVALENT INCOME BY FOUR ALTERNATIVE
EQUIVALENCY SCALES
Persons Aged 25-44
1965 1970 1975 1980 1985 1990
BY FUCHS' SCALE
Blacks 5685 7641 8668 9976 9463 10783
Whites 9633 12012 13045 14723 14797 16119
BY INCOME PER CAPITA
Blacks 3744 5130 5923 7182 7048 8063
Whites 6471 7941 8998 10551 10820 11916
BY PARAMETRIC ADULT EQUIVALENCY SCALE
Blacks 7737 10572 12063 13571 12896 14444
Whites 13027 16240 17648 19762 19629 21373
BY ADULT EQUIVALENCY SCALE IMPLICIT IN THE US POVERTY THRESHOLD
Blacks
Whites
7542 10397 11916 13752 13180 14841
13276 16233 17818 20313 20329 22170
135
TABLE 3.7 : EQUIVALENT INCOME BY FOUR ALTERNATIVE
EQUIVALENCY SCALES
Persons Aged 25-44 by Age Group
Age Group
25-29 30-34 35-39 40-44
BY FUCHS' SCALE
BLACKS Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
9924
5647
75 .7
10144
5662
79 .2
11705
5647
107.3
11897
5773
106 . 1
WHITES Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains {%)
15538
9516
63 .3
15702
9650
62 . 7
16372
9437
73 .5
17219
9958
72 . 9
BY INCOME PER CAPITA
BLACKS Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
7335
3651
101.0
7379
3460
113.3
8723
3757
132.2
9394
4200
123.7
WHITES Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains {%)
11784
6211
89.7
11032
6005
83.7
11768
6278
87.4
13550
7353
84.3
BY PARAMETRIC ADULT EQUIVALENCY SCALE
BLACKS Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
12836
7439
72.6
13607
7340
85 .4
15921
8150
95.3
17048
8400
101.0
WHITES Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
20119
12234
64.5
20116
12518
60 . 7
21815
12996
67.9
24324
14384
69 .1
BY ADULT EQUIVALENCY SCALE IMPLICIT IN THE US POVERTY THRESHOLD
BLACKS Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains {%)
13314
7412
79 .6
14111
7317
92.9
15971
7542
111. 8
17190
8212
109.3
WHITES Baby Boomers in 1990
Parents in 1965
Baby Boomer Gains (%)
21183
12581
68 .4
20870
12501
66.9
22502
12939
73 .9
25204
14715
71.3
136
TABLE 3.8 : GINI COEFFICIENTS FOR EQUIVALENT INCOME,
FOUR ALTERNATIVE EQUIVALENCY SCALES
Persons Aged 25-44
1965 1970 1975 1980 1985 1990
BY FUCHS' SCALE
All Individuals
Blacks
Whites
Black/White Ratio
0.326
0 .384
0.317
1.211
0.311
0 .359
0.302
1.189
0.319
0 .379
0.308
1.231
0 .321
0.378
0.310
1.220
0.354
0.395
0.344
1.148
0.353
0 .397
0.347
1. 144
BY INCOME PER CAPITA
All Individuals
Blacks
Whites
Black/White Ratio
0.381
0 .445
0.367
1.213
0.363
0.420
0.354
1.186
0.368
0 .430
0.359
1.198
0.367
0 .425
0.357
1.190
0.396
0.432
0.387
1.116
0.399
0.436
0.389
1.121
BY PARAMETRIC ADULT EQUIVALENCY SCALE
All Individuals
Blacks
Whites
Black/White Ratio
0.324
0.379
0.312
1.215
0.305
0.352
0.295
1.193
0 .311
0.364
0.301
1.209
0.313
0.366
0.302
1.212
0.347
0.391
0.337
1.160
0.351
0.391
0.340
1.150
BY ADULT EQUIVALENCY SCALE IMPLICIT IN THE US POVERTY THRESHOLD
All Individuals
Blacks
Whites
Black/White Ratio
0 .340
0 .397
0 .327
1.214
0 .319
0.369
0 .309
1.194
0.323
0.380
0.313
1.214
0.323
0.377
0 .312
1.208
0.355
0 .398
0.345
1.154
0.359
0.397
0 .348
1.141
137
CHAPTER 4
BACKGROUND CHARACTERISTICS AND COLLEGE EXPERIENCE
AS DETERMINANTS OF INCOME:
A Sample of Elite College Graduates
I. ABSTRACT
This study explores the effects of micro-level characteristics on short-term
income among a sample of graduates from highly selective, prestigious private
colleges and universities. The choice of occupation is found to be an
endogenous event, determined by a wide range of personal and family
characteristics. Two-stage instrumental variables (2SIV) estimation shows that
income is significantly influenced by variables representing race, formation of
marital-type unions (and if in a union: gender and education of spouse),
parents’ income, father’s education, gpa ranking, professional doctorate degrees,
occupation type, and experience. Income in this sample was not significantly
influenced by gender (outside of marital-type unions), school type (university
vs. college), masters degrees, or doctorates in the arts & sciences. Failure to
138
model occupation as endogenous furnished significantly different results, both
in the magnitude and the significance of various effects.
II. INTRODUCTION
The aim in this study is to explore the connection between personal and
family background variables and individual income among a sample of highly
educated young Americans. The nature of the data set enables us, in the
context of income determination, to examine two topics commonly overlooked
in the literature. First, the affect of extensive sample selection on the
correlation between personal and family background characteristics and income,
and second, the role played by personal and family background characteristics
in intermediate processes that structurally influence income.
The structure of the U.S. educational process is complex, comprised of
multiple phases or stages. Educational success comes in many forms:
attendance, achievement, or graduation could all be considered outcomes.
Every individual’s past experiences, typically measured by some set of personal
and family background variables, affects the likelihood that they will be
successful in meeting each of many educational objectives. When a person’s
individual past experiences structurally influence their success in any stage of
139
the educational process, selection has occurred. Stated alternatively, the fact
that an individual has, for example, completed high school with a high grade
point average is not a random event. Rather, it is an endogenous outcome
influenced by that individual’s past experiences.
Even though childhood education in the U.S. is compulsory, certain
personal and family characteristics have been found to jointly determine
scholastic achievement in that environment (Angrist and Krueger, 1991).
During high school, personal and family characteristics have been shown to
influence achievement, school completion, and students’ perception of the
returns to education (Behrman, Kletzer, McPherson and Schapiro, 1994; Blake,
1989; Manski. 1993). Progression to post-secondary education is another
process affected by selection, both self selection and institutional selection
(Willis and Rosen, 1979; Heckman, 1980). Additionally, previous literature on
the educational selection process has proposed that survival of the selection
process is a specific signal, and such a signal may itself be an effective
outcome of the educational process (Spence, 1973; Riley, 1979).
If a person’s income is affected by the extent to which he/she grew up in
a high income home, a significant portion of the "parents’ income effect" could
be attributed to increasing the likelihood of successful passage through one or
more selection processes. All of the effects of personal and family background
experiences could be partitioned into two components, one component
140
influencing the selection process, and another component influencing decisions
given the selection outcome. The second category of effects, those net of
sample selection, is where our research interest lies. So in a sense, the
existence of sample selection is used as a tool for isolating specific secondary
effects.
The data used come from two surveys, an initial and a follow-up, of
graduates from elite, private universities in the U.S.; all individuals in the
sample are graduates of the class of 1984. The individuals in the sample could
be thought of as a group who have all "survived" a highly rigorous and
competitive selection process. The objective in using this sample was to
identify the extent to which a fairly standard set of personal characteristics
remain important in determining income, even after such thorough selection to
the sample. Whatever the influence past experiences may have had on success
in the selection process, individuals who remain part of the sample are now, in
some respects, on fairly equal ground.
The second main interest of this study is related to model specification,
specifically the treatment of occupational choice. It is assumed that all
individuals are rational utility maximizers, that utility is directly related to
income, and that individuals make decisions and take actions to maximize
income. Utility of individual i can be expressed as in equation (3.1):
141
(3.1)
where: utility
I,
income
background characteristics
random error term
where Bn and Bi2 may or may not be mutually exclusive.
The mechanisms by which personal and family background
characteristics influence individual economic outcomes are surely numerous,
and certainly include being influential in the learning/development process as
well as preference formation process.8 3 Examples of these preferences
include, but are not limited to: preferences for leisure, preferences for
schooling, and preferences for high income (which in turn effect preferences
for additional schooling and/or training). That is to say, a person’s background
can affect their income earning ability as well as their income earning desire.
While individual preferences cannot be observed directly, many personal and
family characteristics (which are systematically related these preferences), are
observed. It is also assumed that college experiences (such as gpa, college
type, and obtaining graduate degrees) influence income, either directly or
through preference formation. College related preferences that affect income
8jIn the context of education, for example, the preference for schooling, or the
preference for high income (and therefore a preference for training or education
that leads to high income).
142
include occupational preference, lifestyle preference, preference for a high
income career, and a preference for working for social change. Therefore,
income of individual i can be expressed by equations (3.2) and (3.3):
(j.2) Ij = X U B, + e2 i
(3-3) [XH ] = [P j|F jJCjIOj]
where [X,j] is a vector of determinants that can be partitioned into the variable
groups used for empirical work in subsequent sections:
Pj = personal background characteristics, such as race, gender, and marital
status.8 4
Fi = family background characteristics, such as parents’ education and
parents’ income.
Cj = college experience characteristics, such as gpa, college type, and highest
earned degree.
0; = occupational experience characteristics, such as occupation type and
experience.
“ individuals are identified as being in a "marital type" union or not. A
marital type union includes married individuals and individuals who identified
themselves as living with a partner.
143
e,i= random error term.
Equation (3.3) can be rewritten as equation (3.4):
(3.4) If = I(Pi,Fi,Ci,Oi,e2 i)
However, this is the point where the treatment of occupational choice is
critical. Even after an individual is admitted to - and subsequently graduates
from - a highly selective and prestigious post secondary institution, several
sequential decisions remain (for example: occupation, progression to graduate
school, and marriage). Occupation may be more accurately modeled as an
endogenous component of income rather than an exogenous explanatory
variable.8 5 Failure to account for the endogeneity of explanatory variables can
result in misspecification of the model and inefficient estimates. Given data
limitations and the strong influence of occupation on realized income, attention
8 5 0ther variables representing gpa, highest completed degree, and socio
economic values may also be endogenous to income. Hausman type specification
tests rejected the endogeneity of gpa in this sample, thus gpa is assumed
exogenous here. Schapiro, O’Malley, and Litten (1991) have shown the intentions
to progress to graduate school by COFHE graduates to be endogenous. However,
graduate degrees are not treated as endogenous in this study, as many individuals
are still in graduate school and have no reported income (22.6% of the sample
reported that working was not their current activity in 1991). The possibility of
endogeneity in explanatory variables representing socio-economic values is not
considered problematic here; this assumption is discussed in a later section.
144
is centered on the possible endogeneity of occupation, and the effect such
endogeneity has on estimation of short-term income.
As the model stands in equation 3.4, occupational choice and all other
explanatory variables are explicitly treated as independent to income. Stated
alternatively, occupational choice is assumed to be pre-determined random
variable to income.8 6 As discussed above, many of the explanatory variables
that explain income may do so via their influence on occupation choice. If this
is the case, occupational choice is not exogenous but rather endogenous to
income, being determined by a separate but structurally related process. This
relationship between occupation of individual i and his/her personal and family
background characteristics can be expressed as in equations (3.5), (3.6), and
(3.7):
(3.5) 0| - X2 iB, + e3 i
(3.6) [X2 i] = [PJFJCJ
(3.7)
“ Assuming occupation is exogenous to income does not require the
assumption that occupation is totally random, just random with respect to income.
145
Hence, based on personal and family background and college
characteristics, each individual maximizes utility after college by choosing an
optimal occupation that is expected to result in maximum utility. Each
individual with characteristics P, F, and C chooses occupation Oj if:
(3.8) E[V(0-J] > E[U(Ok )]
for all Ok not equal to Oj, where E[.] represents the mathematical expectation.
The utility maximization can be expressed generally by equation (3.9):
(3.9) max U -, = U*(P*n! F’n, C*n, 0*(P*i:, F*i2 , C i2, e3 i), e:i, eu)
where 0(.) is a qualitative relation nested in a continuous linear equation.8 7
This study addresses two research questions regarding graduates from a
set of elite private colleges and universities. First, which personal and family
background characteristics significantly explain differences in short term
income, and what are the magnitudes of their effects?8 8 Second, what is the
effect of modeling occupation type as endogenous (relative to assuming
8 7 The qualitative nature of occupation choice will be discussed in a following
section.
8sShort term here meaning approximately six years after graduation in 1984.
146
exogeneity) in the estimation of income? More specifically, how does the set
of 2SIV coefficient estimates differ from ordinary least squares (OLS)
coefficient estimates?
Section III. contains a description of the data, an outline of the variables
used in estimation, and descriptive statistics. Section IV. provides a discussion
of estimation methods. Section V. reports on and discusses econometric
results. Sections VI. concludes.
m. DATA
A. Source
The data were supplied by The Consortium on Financing Higher
Education (COFHE). COFHE represents a group of 32 prestigious, selective
private U.S. colleges and universities, located predominantly in the northeast
(see table 4.1 for a list). In the spring of 1984, 23 COFHE schools conducted
a survey of their graduating seniors. In 1991, 15 of these original COFHE
schools conducted a follow-up mail survey of the class of 1984; the response
rate was 48% of the deliverable mailings.8 9 The number of graduates that
completed both the "Senior Survey" and the "Follow-up Survey" is 1526,
8 9 The identities of the 15 schools were not provided.
147
representing 12 schools. For the purposes of this study, only those individuals
who indicated that their principle activity in February 1991 was employment
are included.9 0
As mentioned in the previous section, this sample is a result of
considerable selection, a typically undesirable attribute. The selection has
occurred at many steps, including completion of secondary schooling,
acceptance to a COFHE school, graduation from the COFHE school, and
employment at the time of the follow-up survey.9 1 Given the objectives of
this study, the extent that selection shapes the sample into the "best of the best"
is beneficial rather than problematic. The interest here is how personal and
family background variables affect individuals who have received maximum
benefit from the educational system. Any selection that isolates the high
achievers, the highly motivated, and the most employable, is consistent with the
aims of the study.
However, other types of inherent selection are ambiguous in effect and
not so easily dismissed. Many individuals who were not employed at the time
9 0 This implies that 22.6% of the original sample were either students or were
in some "other" activity. For actual sample sizes in the estimations, see Tables 3
and 4. Also, individual records that had missing values for any variables used
were dropped, so actual sample sizes for estimation were smaller. See each table
for associated sample.
9lThis is not to mention selection via survey attrition. Survey attrition is
discussed subsequently.
148
of the follow-up survey (and were therefore excluded from the sample) were in
school, presumably graduate school. The elimination of these individuals from
the sample cannot be welcomed as improving sample "quality" in any
systematic way. Additionally, the attrition of individuals between the senior
and follow-up surveys, arising from loss of entire schools as well as individuals
within schools, could possibly introduce estimation bias. However, the effects
from these second types of selection do not significantly affect the descriptive
sample statistics o f the utilized explanatory variables, and are therefore assumed
to affect overall results minimally.9 2
B. Descriptive Statistics
Table 4.2 presents a list of the variables with descriptions and sample
statistics; the statistics are reported for all individuals, as well as for men and
women separately. No recovery schemes for missing data were employed, and
individuals with missing data for any variable used in the system were dropped
from the sample. The resulting sample sizes for the occupation and income
equations were 951 and 887, respectively.
9 2 Due to the overwhelming weight that to follow-up variables play in the
income determination system, corrections for the selection arising from attrition
are not performed.
149
The sample is predominantly white (87.6% overall); asians, blacks and
hispanics each constitute between 2.5% to 5% of the sample.9 3 Men make up
just under half of the sample (48.5%). A little more than half of the sample
was married or with a partner by 1991, men to a slightly larger degree. Out of
those who married (or formed a marital type union), about half had spouses (or
partners) that were also college graduates.9 4 About 9% of the sample report
their parents’ annual income (in 1983, the year prior to graduation) was less
than $20,000 (the low category), about 61% report parents’ annual income
between $20,000 and $80,000 (the medium category), and about 30% of
individuals report parents’ annual income above $80,000 (the high category);
notice that a larger proportion of men reported having parents’ income in the
high category. The fathers of the sample COFHE graduates are highly
educated, in that about 53% of fathers had graduate degrees; a slightly larger
percentage of women than men had fathers with graduate degrees. Many
COFHE graduates themselves have pursued graduate degrees, and 46% of
COFHE graduates reported already having obtained a graduate degree within
9 "Keep in mind that this sample reflects only 12 COFHE schools, and
represents only those who completed both surveys. For entering freshmen of
COFHE schools in general, the breakdown for 1983 was: ASIAN 7.1%, BLACK
6.0%, HISPANIC 3.4%, and WHITE 83.5% (source:COFHE).
9 4 Survey respondents self reported if their spouse/partner was a college
graduate without specification as to what type of graduate.
150
seven years of undergraduate matriculation; similar proportions of men and
women have obtained masters degrees, but a larger proportion of men have
completed doctorate degrees.
The follow-up survey asked individuals to identify their principle activity
in 1991 as either "student", "employment", or "other activity". If
"employment" was chosen, then the survey asked the individual to identify their
field of work; there were 33 separate occupation categories from which to
choose. For the purposes of this study, the list of occupations was condensed
into the following seven categories:
1. Business (including accountant, actuary, advertising, public relations,
banking, finance, business consultant/analyst, marketing, and sales).
2. Technical (including architect, urban planner/designer,computer
programmer/analyst, engineer, and statistician).
3. Public administration/social service/government (including
college/university administrator, foreign service, government, public
administration, library, museum, religious, social/welfare/recreation
worker, teacher, and educational administrator).
4. Health, excluding physicians (including nursing, allied health, and
clinical psychologists).
5. Academic or non-academic researcher (including college/university
teacher or researcher, non-academic scientific research, non-academic
social scientist).
6. Arts/Entertainment/Communications (including visual or performing
artist, communications - film, radio, TV, journalism, and other
professional writing).
151
7. Professional Doctor (including physician, dentist, veterinarian, lawyer,
and judge).
The most popular career category was business (34.6%), and the second
most popular was professional doctors (26.7%). Notice that considerably more
men than women chose OCCUP2 (technical), and significantly more women
than men choose either OCCUP3 (public admin./social service/govt) or
OCCUP4 (health). Men and women had almost identical within-school gpa
rank (50.7% for men vs. 49.7% for women). The average number of years of
work experience was 3.71 years, with men and women and women reporting
similar levels o f experience (3.76 and 3.67 years, respectively). Average
income six years after graduating from a COFHE school was $41,530; women
on average earned 74% of men.9 5 Such a low level of experience associated
with this average amount of income clearly shows the economically elite nature
of the sample.
9 :The large gap between men and women’s earnings is due to differences in
gender specific means of explanatory variables. Results in subsequent sections
show how the effects of gender alone are insignificant.
IV. METHODOLOGY
152
As mentioned above, the treatment of occupational choice as an
explanatory variable is a central research issue. To account for the endogeneity
of occupation on income, income is estimated by the technique of two stage
instrumental variables (2SIV), using predicted occupation type as a second
stage instrument. The first stage of the 2SIV procedure involves estimating
occupational choice and obtaining measures of predicted occupation. The
second stage of the 2SIV procedure involves estimating the income equation by
OLS using predicted occupation rather than actual occupation. After estimating
income by 2SIV, we test for endogeneity of occupational choice by comparing
2SIV coefficient estimates to OLS estimates via a Hausman-type specification
test (Hausman, 1978).
A. First Stage Estimation of Occupational Choice
Occupational choice is an unordered qualitative outcome and therefore is
non-linear in nature. Occupational choice is estimated using the multi-nomial
logit model (Theil, 1969; Nerlove and Press, 1973; and Schmidt and Strauss,
1985b). Use of this model in the context of occupational choice assumes
independence among occupational alternatives.9 6 Predicted occupation type
9 6 While it is preferred not to make this restriction, current modeling
from the logit model could come in one of two ways: (1) estimated occupation
could be represented by the one occupation category a person is most likely to
choose, or (2) estimated occupation could be represented by the whole set of
estimated probabilities of choosing each occupation type. The second option,
the set of estimated probabilities was chosen as instruments for this analysis.9 7
GPA from the survey is categorical, in that respondents identified their average
grade category from "A" down to "C or below". Due to possible differences in
the relative value of grades across schools, a new measure of achievement was
created. Using the reported category of gpa and the distribution of gpa in an
individual’s school, each respondents gpa ranking is determined, by school.9 8
This new measure of GPA is referred to as "GPARANK"; the values of
GPARANK fall between 0 and 1.
alternatives that allow for dependence between choices restrict the number of
choices modelled. Independence is assumed and a large set of occupational
choices is maintained.
9 7 The estimates of occupation as probabilities are more precisely a combination
of the independent regressors alone, thus more effectively eliminating the
correlation between occupation and income equation errors.
9 S This makes quite a difference in the overall sample; an "A-\B+" for example,
puts an individual as high as the 92 percentile in one school and as low as the 70
percentile in another.
154
B. Second Stage Income Estimation
The income measure is categorical self-reported own income in 1990."
Categorical income was transformed into the mean dollar value in the range,
and treated as continuous.1 0 0 The dependent variable in the estimation is the
natural log of income. Income was estimated by OLS using instrumented
occupation as an explanatory variable; the results of this 2SIV procedure, along
with OLS estimates, provide the basis of specification testing discussed below.
Standard errors in the second stage are corrected to reflect the fact that the
instruments themselves are stochastic (Murphy and Topel, 1985; Green, 1990).
V. RESULTS
A. Occupation Type Estimation
Table 4.3 presents empirical results for the multi-nomial logit estimation
of occupation type, given individual characteristics.1 0 1 The estimated
"This means that "own income" relates to income earned between 5/4 and 61 /;
years after graduation.
lo o Since there are 11 income categories, the statistical costs of this
transformation should be minimal.
,olThe occupation equation coefficient estimates in table 4.3 are based on the
entire sample. The estimation was also run for men and women separately, but
this reduced samples sizes below 500 and some occupation categories frequencies
155
coefficients represent the independent effects of each variable on the log of the
ratio of the probability of choosing each occupation over another (hereafter
referred to as the log-ratio-probability). Since there are seven occupational
categories, there are 21 different sets of coefficients.1 0 2 Although each
variable is modeled to independently affect each log-ratio-probability, the
results show that not all variables significantly affect all log-ratio-probabilities.
Personal Characteristics:
Personal characteristics that significantly affect multiple log-ratio-
probabilities are: gender, marital or union status, and considering "the
opportunity to work for social change" an essential career characteristic. Being
male shows a tendency to favor technical academic/nonacademic research
careers, and to avoid pub.admin./soc.svc./govt and health careers. Having a
marital-type partner is significantly associated with not choosing either a
pub.admin/soc.svc./govt career or a professional doctor career. Individuals that
below 30, casting suspicion on the reliability of the estimates. Thus, the
prediction of occupation comes from the combined gender sample.
1 0 “The effect that a certain explanatory variable has on the log ratio of the
probability of choosing occupation i to the probability of choosing occupation j,
is equal to the negative of the effect that the explanatory variable has on the log
ratio the probability of choosing occupation j to the probability of choosing
occupation i.
log«/(Pi/Pj) = -log^Pj/Pi)
156
identified themselves as desiring a career that provides "the opportunity to work
for social change" have a significant tendency to choose
pub.admin./soc.svc./govt careers, health careers, or professional doctor careers;
all these professions are associated with working for or helping the public.
Many personal characteristic variables failed to significantly affect occupational
choice, including race and education level of spouse/partner (if in a union);
these variables were therefore excluded from the equation.
Parent characteristics:
Having parents with income in the $80k + category also significantly
affected several log-ratio-probabilities. Generally, individuals with high
parents' income had a significant tendency to choose a professional doctor
career over all others except business, and a tendency to choose business
careers over technical or pub.admin./soc.svc./govt careers. Neither mothers’ or
fathers’ education level significantly affected occupational choices, so these
variables were dropped from the equation.
Educational Experience Characteristics:
Having higher GPARANK positively affects choosing either a
pub.admin./soc.svc./govt., a professor/researcher, or a professional doctor
career, and decreases the chances of choosing business or technical careers. All
157
COFHE graduates attended either a university or a liberal arts college.1 0 3
Those who attended a university favor health careers, and avoid
pub.admin./soc.svc/govt careers. Four variables represent the completion of
graduate degrees: MASTERA&S, MASTEROTH, DOCTORA&S, and
DOCTOROTH. Masters degrees in the arts and sciences (MASTRA&S) were
found to be insignificant to occupational choice, and excluded from the
equation. The other three variables representing completion of graduate
degrees were significant to occupational choice. Masters degrees in areas
outside the arts and sciences (MASTROTH) give a significant tendency to
choose a business career, or a technical career (if not business). Individuals
with doctorates in the arts and sciences (DOCTRA&S) have a significant
tendency to avoid business occupations and favor academic/nonacademic
research careers. Individuals with doctorates outside the arts and sciences
(DOCTROTH) also have a significant tendency to avoid business careers but
tend to favor professional doctor careers rather than research careers.
The estimation of the multi-nomial logit model revealed predicted
probabilities of choosing each occupation choice. Using these predicted
probabilities, the occupation equation correctly predicted 57.2% of the actual
occupational types; this indicates considerable correlation between chosen
occupation and the independent variables. Such correlation and predictive
'“ "College" includes coed and single-sex colleges.
158
power supports the concern of the endogeneity of occupational choice to
income. Hausman (1978) has devised a specification test that can be used in
the context of possible endogeneity; discussion of these tests follows in section
V.
B. Income Equation Estimation by Instrumental Variables
The second stage of the estimation is the regression of the log of own
income on a set of 20 explanatory variables (table 4.4); this includes
instrumented occupation obtained in the first stage occupation estimation.
Since the dependent variable is the log of income, the linear coefficient
estimates can be roughly interpreted as percentage effects on actual income
when all other variables are at the sample means.1 0 4 Included with the 2SIV
coefficient estimates are the OLS coefficient estimates; the OLS estimates result
from treating occupation as exogenous and omitting the first stage occupation
equation. The following discussion focuses on the 2SIV coefficient estimates,
but highlights significant differences between 2SIV and OLS results.
1 0 4 That is to say, a coefficient of .10 for gender would imply that males earn
.10 more log of own income, or alternatively, they earn approximately 10% more
income. More precisely, coefficients of .10 and -.10 relate to a 10.5% increase
and a 9.5% decrease in income, respectively, coefficients of .20 and -.20 relate to
a 22.1% increase and a IS.2% decrease in income, respectively, and coefficients
of .30 and -.30 relate to a 35% increase and a 26% decrease in income,
respectively, etc.
159
Personal Characteristics:
Being black has a positive and significant effect on income (21.8%);
OLS estimates, however, indicate that being black has an insignificant effect.
Men have no significant advantage over women in this sample, unless they
have formed a marital-type union. If an individual has formed a marital-type
union, then the regressors PARTNER, GENDER* PARTNER, and PTNRGRAD
must be considered jointly.1 0 5 If a union has been formed, both the gender of
the individual and the education level of the spouse enter into the picture, and
the sum of the effects produces the overall effect. Other things equal, being
married or with a partner has a large significant and negative effect on income
(-30.3%), indicating an income penalty associated with forming a union.
However, the positive and significant coefficient (+28%) for PTNRGRAD
almost completely offsets the negative coefficient for PARTNER, resulting in
little effect. In truth, the income effect isn’t from forming a marital-type union
alone, but rather from forming a union with a relatively uneducated partner.
Additionally, the significant and positive coefficient (+15.1%) for
GENDER*PARTNER shows that men in unions are affected quite differently
than women in unions.1 0 6 This suggests that for this sample, men in marital-
1 < b The inclusion of GENDER*PARTNER and PTNRGRAD as explanatory
variables shed light on how forming a marital-type union and how having a
college educated partner affects men and women quite differently.
1 0 6 Earlier regressions included a PTNRGRAD*GENDER interaction term, but
160
type unions may act differently, with regards to obtaining income, than women
in such unions.
When the effects of GENDER, GENDER*PARTNER, and PTNRGRAD
are considered jointly, we see the following. While women in unions with
non-college grads have 30.3% less income, men in unions with non-grads have
only 15.2% less income. While women in unions with college grads have
2.2% less income, men in unions with college grads actually have 12.9 more
income. OLS estimates relate quite a different story. Although 2SIV and OLS
report similar coefficients for PARTNER (in a marital-type union) and
PTNRGRAD (partner is a college graduate), the effects across gender are much
different. OLS estimates indicate that men earn more than women regardless
of marital-tvpe union formation, and forming such a union affects men and
women similarly. The structure of OLS estimation, which models occupation
as exogenous, assumes that women, whether or not in a union, make occupation
choices identically to men. Hence, differences in OLS and 2SIV results here
suggest this is a limiting and potentially misleading assumption.
As mentioned earlier in this chapter, the income estimation was
performed separately for men and women. However, the resulting sample sizes
for each gender group were less than 500, and more importantly, the
the coefficient was insignificant. Evidently, although forming a marital-type union
affects men and women differently, having a college educated spouse affects them
the same.
161
frequencies in each occupation category dropped below 30 in several cases.
Given the importance of occupation choice to this model, such data problems
are critical, and the reliability of resulting estimates is questionable. Therefore,
our discussion of results is based on the pooled sample. However, coefficient
estimates from 2SIV and OLS for sample by gender are presented in the
appendix to chapter 4.
Parent Characteristics:
Individuals whose parents’ had low or medium income have, on average,
lower levels of own income relative to individuals with higher parents’ income.
More specifically, at the sample means, low and medium parents’ income
relates to 18.9% and 4.6% lower own income, respectively, both relative to the
reference group of high parents’ income. 2SIV estimates indicate the effect of
having low (versus high) parents’ income is significant but the effect of having
medium (versus high) parents’ income is not, while OLS estimates imply that
the effects of both low and medium parents’ income are significant. Once
again, the comparison between OLS and 2SIV coefficients suggests that
modeling occupation as endogenous weakens the direct link between
explanatory variables and income. Fathers’ education measured here by
completion of graduate school, appears to significantly affect income. In this
1 6 2
sample, individuals whose fathers have a graduate degree earn 8% less income
than those whose fathers did not.1 0 7 1 0 8
Educational Experience Characteristics:
Overall undergraduate academic achievement, as measured by the
variable GPARANK, has a positive and significant effect on income. Recall
that the structural range of GPARANK values is 0 and 1. The minimum and
maximum values actually observed are .02 and .94. Thus, the coefficient
estimate of .313, which implies a income effect of approximately 31% from
GPARANK of 0 to GPARANK of 1, slightly overstates the maximum possible
income effect. The maximum income effect is approximately 28%, and that
comes from comparing the highest observed GPARANK to the lowest. OLS
estimates attributed only 19.6% more income to this gpa difference.
Individuals who consider "high income potential" an essential career
characteristic have on average 27% more income than those who don't; OLS
estimates report a smaller 17.8% income effect from desiring high income
1 0 'This variable should be interpreted with caution. Since this sample excludes
individuals who are still in school, and since individuals whose fathers have
graduate degrees may have a tendency to pursue graduate degrees themselves, the
sample may be censored, and the coefficients biased.
1 0 s Variables representing mothers’ education were previously included in the
regressions. These variables were found to be irrelevant to income, however, and
were excluded.
163
potential. Attending a university had no significant affect on income by 2SIV,
but OLS estimates report a significant 9% university income premium. This
difference between the 2SIV and OLS significance of UNIV as an explanatory
variable no doubt relates to differences in typical curriculum/major options (and
hence occupation choices) between COFHE universities and colleges.
Surprisingly, the effect of graduate degrees on income is largely
insignificant, and when it is significant, the effect is negative.1 0 9 The
coefficient for the variable DOCTROTH (doctorate degrees in areas other than
the arts and sciences) is negative and significant (-68.3%), and the coefficient
for MASTROTH (master degrees in areas other than the arts and sciences) was
insignificant. OLS estimates attribute positive and significant income effects to
MASTROTH but the coefficient on DOCTROTH is insignificant.1 1 0
Occupational Characteristics:
Business (OCCUP1) was the reference category for estimating the effect
of occupation type on income. Thus, relative to choosing OCCUP1, all other
occupations except professional doctors (OCCUP7) have significant negative
1 0 9 This ambiguous or negative effect from graduate degrees may rise out of the
inherent education/experience tradeoff in cross sectional analysis.
1 1 0 Variables representing masters and doctorates in the arts and sciences were
insignificant under both 2SIV and OLS structures, and were excluded from the
model.
164
effects on income. This is also the case with OLS estimates, except the
coefficients are much smaller.
Experience is also a significant factor in explaining income. The
experience variable, EXPER, has a significant positive coefficient indicating
that every year of experience is related to 6% more income. Previous versions
of the model included a term for the square of experience, typical of classic
age-eamings or experience-eamings profile models (Becker, 1965; Ben Porath,
1965; Mincer, 1974).1 1 1 The squared experience term, however, was
consistently insignificant, and thus dropped from the model. Within this
sample, the interpretation of coefficients on experience is difficult. Although
there is a wide distribution of experience values, all individuals come from one
graduation cohort and are roughly the same age.1 1 2 The education quality of
this sample minimizes relative unemployment, thus implying that low levels of
experience are most likely associated with additional investment in
education.1 1 ' In this context, the returns to experience are difficult to isolate.
' 1 'These data represent a cross-sectional sample, so variations in experience are
not associated with variations in age, but are associated with variations in
completed education (i.e. individuals who went on to full-time graduate school
have not been able to acquire as much work experience as those who did not).
1 1 2 This fact alone differentiates this analysis from classic types of age-earnings
models, which have commonly assumed experience to be equal to age - schooling
- 6 .
ll3Some careers do not require schooling beyond the baccalaureate (e.g.
engineers, artists, social service), while other careers have a graduate education
165
C. Tests for Endogeneity of Occupational Choice
Overall, the first stage occupational choice equation correctly predicted
nearly 60% of the actual occupational choices made, showing that the
correlation between occupational choice and the nine explanatory variables is
significant and wide spread.1 1 4 It is just this type of correlation and
predictive power which supports the concern about the endogeneity of
occupational choice. A specification test devised by Hausman (1978) can be
used to explore the endogeneity of occupational choice in this context.
The endogeneity test compares two sets of coefficients, one assuming
exogeneity and another assuming endogeneity. The test is based on the idea
that under the null hypothesis of no endogeneity, OLS estimation and 2SIV
estimation are both consistent, but 2SIV is inefficient. Under the alternative
hypothesis of endogeneity, 2SIV estimation is consistent but OLS is not. The
coefficients from the 2SIV and OLS estimations are compared in the form of a
Wald criterion, which results in the chi-squared statistic of 71.6896, for which
the critical value with 7 df is 14.07 (at 95% significance). Thus, the hypothesis
of exogeneity is rejected, and the hypothesis of endogeneity is supported.
requirement (e.g. physicians, dentists, academic research). Thus, individuals who
undertook additional schooling as a requirement for a career will have a
combination of lower experience and higher levels of education than individuals
who moved directly from undergraduate education to the workforce.
I,4A random draw of an occupation type has 14% probability of being correct.
166
These test results suggest that modeling occupation as exogenous (as in
the OLS estimation) leads to inconsistent estimation, and misunderstanding of
how personal and family background characteristics truly affect income.
Modeling occupational choice as endogenous to income (as in 2SIV estimation)
leads to consistent estimation and more reliable empirical results.
VI. CONCLUSION
This study utilizes an ’elite’ sample of college graduates and explores the
explanatory power of micro-level personal and family background variables on
observed differences in short-term income. A central concern of the essay was
the appropriate treatment of occupation choice as an explanatory variable to
income. The logit estimation of occupational choice shows that a wide range
of personal characteristics significantly affect the choice of occupation.
Subsequent hypothesis tests support the argument for the endogeneity of
occupation to income.
Assuming the endogeneity of occupational choice, income is estimated
by a two stage instrumental variables procedure. Results show that
withstanding extreme similarity in the high quantity and quality of education
across individuals, many personal and family background characteristics are
167
significantly influential to income. Characteristics found to be statistically
significant to income were: (1) race (being black), (2) gender (if in a marital-
type union),1 1 5 and (3) marital-type partner being a college graduate (if in
such a union), (4) parents' income, (5) father having a graduate degree, (6)
undergraduate GPA ranking, (7) considering "High income potential" an
essential career characteristic, (8) obtaining a doctorate degree in a field other
than arts and sciences, (9) predicted occupation type, and (10) years of
experience. Characteristics insignificant to income were: (1) gender (outside
marital-type unions), (2) mother’s education, (3) considering "The opportunity
to work for social change" an essential career characteristic, (4) school type, (5)
obtaining masters degrees, and (6) obtaining doctorate degrees in the arts &
sciences.1 1 6 Evidently, the strong effect from obtaining a college degree, and
the even stronger effect of obtaining a COFHE school degree, cannot
undermine the income effects coming from variables representing personal and
family background characteristics and college experiences.
It is also shown that treating occupation as exogenous has salient and
important repercussions. Ignoring the endogeneity of occupation to income
leads to coefficient estimates that are often at odds with estimates resulting
from the incorporation of endogenous occupation in the model. Assuming
1 1 5 Outside of marital-tvpe unions, gender is insignificant.
1 1 6 The square of experience is also insignificant.
168
exogeneity of occupation leads to the incorrect conclusion that certain
insignificant characteristics are in fact significant (e.g. gender (outside of
marital-type unions), attending a university (rather than a liberal arts college),
and obtaining masters degrees), and that some significant characteristics are in
fact insignificant (e.g. being black, gender (within marital-type unions), and
possessing doctorate degrees in fields other than the arts and sciences). Even
when the two estimation methods agree on the significance of a characteristic,
the empirical cost of assuming exogeneity of occupation is captured through the
differences in magnitudes of certain coefficients (e.g GPARANK, HINCPOT,
and four occupation variables).
The question of endogeneity of occupational choice is clearly non-trivial.
Incorrect specification in this income model would lead a researcher to believe
that some personal variable effects are significant when they are not, and vice-
versa. Even when the interpretation o f factor significance is robust to
endogeneity assumption, estimates of certain coefficients are quite sensitive.
Two important implications come from this study. First, despite the
strong - and potentially equalizing - effect that a top quality educations has on
short term income, measures of certain personal and family background
variables are found to be significantly influential to that income. Second,
assumptions about specific pathways by which these variables affect income are
critical to empirical results and corresponding scientific conclusions. More
169
specifically, this study shows that when factors which are influential to
occupational choice are assumed to affect income directly (rather than indirectly
through endogenous occupation) coefficient estimates are inconsistent, biased,
and therefore misleading.
TABLE 4.1 : Consortium on Financing Higher Education
Member Institutions:
Amherst College
Barnard College
Brown University
Bryn Mawr College
Carleton College
Columbia University
Cornell University
Dartmouth University
Duke University
Georgetown University
Harvard University
The John Hopkins University
Massachusets Institute of Technology
Mount Holyoke College
Northwestern University
Oberlin College
Pomona College
Princeton Universtity
Radcliff University
Rice Universtiy
Smith College
Stanford University
Swarthmore College
Trinity College
The University of Chicago
University of Pennsylvania
The University of Rochester
Washington University
Wellesley College
Wesleyan University
Williams College
Yale University
TABLE 4.2 : DESCRIPTIVE STATISTICS
171
Indep.
Variables Description
all
%=1
men
%=1
womer
%=1
ASIAN Race is Asian 4.9 3.9 5 . 9
BLACK Race is Black 2.9 2 . 8 3.1
WHITE Race is White 87 . 6 87.6 87 . 6
HISPA Race is Hispanic 2.5 2 . 8 2.2
GENDER Gender is Male 48.3 - -
PARTNER Married or have "Partner1 1 57.4 57.9 56.9
PTNRGRAD Spouse/partner is College Graduate 51.1 52.1 50 .1
PINCLOW Parents' income < $20k 8.7 7.8 9.6
PINCMED $20k < Parents' income < $80k 60 . 7 58 .4 62 .9
PINCHIGH Parents' income > $80k 30.6 33 .8 27.5
FTHRGRDS Father has a graduate degree 53 .0 50.2 55.6
HINCPOT Consider "high income potential" an essential
career characteristic
17.1 22 . 0 12.6
SOCIALCH Consider "opportunity to work for social
change" an essential career characteristic
20.0 15.2 24.5
UNIV COFHE school attended was a university 68.8 82.0 56.4
MASTRA&S Highest degree is masters, arts & sciences 11.0 10 . 0 12 . 0
MASTROTH Highest degree is masters, not arts & sciences 11.2 12 .4 10 . 0
DOCTRAScS Highest degree is doctorate, arts & sciences 7 .1 8.7 5.7
DOCTROTH Highest degree is doctorate, not arts & sciences 16.6 20 . 1 13 .3
OCCUP1 Business 34.7 35.6 34 .0
OCCUP2 Technical 8.7 12.2 5.5
OCCUP3 Public administration, social service, or govt 13 .3 8.6 17 . 7
OCCUP4 Health field, not physicians 3.4 0 . 7 5 . 9
OCCUP5 Academic or nonacademic research 7.1 8.6 5.7
OCCUPS Arts, entertainment, or communications 5.9 5.1 6.5
OCCUF7 Professional doctor {physician, lawyer, dentist,
or veterinarian)
26.9 29.2 24 .8
mean mean mean
GPARANK Own-school GPA percentile 49.2 50.4 49 .7
EXPER Years experience in current field 3 .71 3 .76 3 .67
LOWNINC Log of own income ($l,000's) 41.5 48.5 35.9
Sample Size 887 428 459
172
TABLE 4.3 : COEFFICIENT ESTIMATES, OCCUPATION EQUATION
Dependent Variable: Occupation Category
Indep.
Variable
log.
(p2/pl)
log.
(p3/pl)
log.
(p4/pl)
log.
(p5/pl)
log.
(p6/pl)
log.
(p7/pl)
CONSTANT -1.1814 ** -0.3869 -3.4432 ** -2.9340 ** -2.0007 ** -1.7386 **
(0.3597) (0.3042) (0.7231) (0.5153) (0.4376) (0.3530)
GENDER 0.7805 ** -0.3926 -2.5591 ** 0.3877 -0.3746 -0.1700
(0.2773) (0.2489) (0.6275) (0.3588) (0.3105) (0.2558)
PARTNER 0.2256 -0.5715 ** 0.2881 -0.2765 -0.1452 -0.6374 **
(0.2607) (0.2273) (0 .4241) (0.3327) (0.2938) (0.2413)
PINCHIGK -0.6417 ** -0.9894 ** -0.6197 -0.9690 ** -0.0600 0.2260
(0.2946) (0.2885) (0.4624) (0 .4078) (0.3096) (0.2521)
GPARANK -0 .3857 0.8322 ** -0.3628 2.2461 * ■ * ■ 1.2644 ** 1.4730 **
(0.4784) (0.4189) (0.7916) (0.9129) (0.5449) (0.4530)
SOCIALCH 0.4563 1.7684 ** 1.4673 ** 0.7589 * 0.4967 1.2579 **
(0.3825) (0.2816) (0.4523) (0.4423) (0.4353) (0.3163)
UNIV -0.5585 ** -0.6818 ** 1.9500 ** -0.5829 0.0803 -0.1009
(0.2857) (0.2466) (0.6348) (0.3765) (0.3423) (0.2783)
MASTROTH -0.6848 * -1.5989 ** 0.0389 ■11.1451 -2.1511 ** -1.8379 **
(0 .3571) (0.4266) (0.5098) (85.972) (0.7342) (0.5386)
DOCTRA&S 1.6898 * -8.7257 2.6227 ** 4.7864 ** 0.6948 3.4778 **
(0.9294) (72.904) (1.0673) (0.7641) (1.2368) (0.7540)
DOCTROTH -8.3021 1.9573 * -6.8340 3.4846 ** -8.1226 6.2077 **
(115.87) (1.1693) (103.80) (1.1447) (124 .37) (1.0179)
Category 1 = "business", 2 = "technical", 3 = "public administration, social service,
or government", 4 = "health, except physicians", 5 = "academic or non-academic
research", 6 = "arts, entertainment, communications", 7 = "professional doctors".
173
TABLE 4.3 : (Continued)
Indep.
Variable (p3/p2)
log.
(p4/p2)
log.
(p5/p2)
log,
(p6/p2)
log.
(p7/p2)
leg.
(p4/p3)
CONSTANT 0.7944 ** -2.2518 ** -1.7526 ** -0.8193 -0.5572 -3.0563
(0.4053) (0.7753) (0.5757) (0.5159) (0.4471) (0.7437)
GENDER -1.1732 ** -3.3397 ** -0.3928 -1.1552 ** -0.9505 ** -2.1665
(0.3257) (0.6550) (0.4134) (0.3803) (0.3351) (0.5519)
PARTNER -0.7971 ** 0.0624 -0.5022 -0.3709 -0.8630 ** 0.8595
(0.3014) (0.4738) (0.3833) (0.3574) (0.3132) (0.4473)
PINCHIGH -0.3477 0.0220 -0.3271 0.5817 0.8678 »* 0.3697
(0.3763) (0.5265) (0.4693) (0.3935) (0.3475) (0.5137)
GPARANK 1.2179 ** 0.0228 2.6319 ** 1.6501 ** 1.8588 '* -1.1950
(0.5535) (0.8458) (0.7307) (0.6588) (0.5819) (0.7969
SOCIALCH 1.3121 ** 1.0109 0.3025 0.0403 0.8015 ”* -0.3011
(0.3825) (0.5298) (0.5068) (0.5088) (0.4097) (0.4542)
UNIV -0.1233 2.5085 ** -0.0243 0.6388 0.4576 2.6319
(0.3250) (0.6720) (0.4258) (0.4049) (0.3501) (0.6477)
MASTROTH -0.9141 * 0. 723 7 -10.4603 -1.4663 * -1.1531 * 1.6378
(0.5202) (0.5993) (85.9724) (0.7955) (0.6188) (0.6305)
DOCTRA&S -10.4155 0.9328 3.0966 ** -0.9949 1.7880 ** 11.3484
(72.904) (1.0015) (0.6562) (1.1778) (0.6534) (72.905)
DOCTROTH 10.2594 1.4680 11.7867 0.1794 14.5098 -3.7913
(115.87) (155.56) (115.87) (169.98) (115.87) (103.80)
Category 1 = "business", 2 = "technical", 3 = "public administration, social service,
or government", 4 = "health, except physicians", 5 = "academic or non-academic
research", 6 = "arts, entertainment, communications", 7 = "professional doctors".
174
TABLE 4.3 : (Continued)
Indep.
Variable
log,
(p5/p3)
log.
(pS/p3)
log.
(p7/p3)
log,
(p5/p4)
log,
(p6/p4)
log,
(p7/p4)
CONSTANT -2.5471 ** -1.6138 ** -1.3517 ** 0.5092 1.4424
*
1.7045 **
(0.5420) (0.4725) (0.3931) (0.8491) (0.8106) 1 (0.7657)
GENDER 0.7803 ** 0.0180 0.2226 2.9468 ** 2.1845
■ * *
2.3891 **
(0.3911) (0 .3545) (0.2988) (0.6980) (0.6798) 1
(0.6538)
PARTNER 0 .2949 0.4262 -0.0659 -0.5647 -0.4334 -0.9256 **
(0.3577) (0.3265) (0.2738) (0.5102) (0.4870)
1
(0.4558)
PINCHIGH 0.0205 0.9294 ** 1.2155 ** -0.3491 0.5597 0.8458 *
(0.4S26) (0.3813) (0.3285) (0.5872) (0.5279)
1
(0.4932)
GPARANK 1.4139 ** 0.4322 0.6408 ** 2.6090 ** 1.6273 1.8359 ’ >*
(0.5350) (0.6033) (0.5077) (0.9357) (0.8743) (0.8178)
SOCIALCH -1.0095 ** -1.2717 ** -0.5105 * -0.7083 -0.9705
★
-0.2093
(0.4312) (0.4282) (0.3026) (0.5615) (0.5597) (0 .4730)
UNIV 0.0989 0.7622 ** 0.5809 * -2.5329 ** -1.8697
* *
-2.0510 *-
(0.3964) (0.3687) (0.3019) (0.7061) (0.6942) (0.6616)
MASTROTH -9.5462 -0.5521 -0.2389 •11.1840 -2.1900
★ ★
-1.8768 **
(85.972) (0.8252) (0.6519) (85.973) (0.8710) (0 . 7147)
DOCTRA&S 13.5122 9.4206 12.2036 2.1637 ** -1.9278 0.8551
(72.901) (72.908) (72.902) (0.8417) (1.2830) (0.8298)
DOCTROTH 1.5272 * -•10.0800 4.2503 ** 10.3186 -1.2886 13.0417
(0.3036) (124.37) (0.6098) (103.80) (161.99) (103.80)
Category 1 = "business", 2 = "technical", 3 = "public administration, social service,
or government", 4 = "health, except physicians", 5 = "academic or non-academic
research", 6 = "arts, entertainment, communications", 7 = "professional doctors".
175
TABLE 4.3 : (Continued)
Indep.
Variable
log.
(pS/p5)
log.
(p7/pS)
log.
(p7/P6)
CONSTANT 0.9332 1.1953 '* 0.2621
(0.5266) (0.5233) (0.5068)
GENDER -0.7623 * -0.5577 0.2046
(0.4356) (0.3596) (0.3577)
PARTNER 0.1313 -0.3608 -0.4921
(0.4052) (0 .3314) (0.3358)
PINCHIGH 0.9089 * 1.1949 ** 0.2860
(0.4759) (0.3975) (0.3536)
GPARANK -0.9817 -0.7730 0.2086
(0 .7741) (0.6501) (0.6278)
SOCIALCH -0.2621 0.4990 0.7611
(0.5439) (0.4062) (0.4501)
UNIV 0.6632 0.4819 -0.1812
(0.4639) (0 .3756) (0.3903)
MASTROTH 8.9940 9.3072 0.3132
(85.974) (85.973) (0.8881)
DOCTRA&S -4.0916 ** -1.3086 ** 2.7830
(1.0483) (0.3709) (1.0399)
DOCTROTH -11.6073 2.7231 ** 14.3304
(124.37) (0.5686) (124.37)
Category 1 = "business", 2 = "technical", 3 = "public administration, social service,
or government", 4 = "health, except physicians", 5 = "academic or non-academic
research", 6 = "arts, entertainment, communications", 7 = "professional doctors".
Auxiliary Statistics at convergence initial
Log Likelihood
Sample Size
% Correctly Predicted
-1151.6
951
57.203
-1850.5
176
TABLE 4.4 : COEFFICIENT ESTIMATES, INCOME EQUATION
Dependent Variable: Log of Income
OLS 2SIV
Independent Coefficient Coefficient
Variable: Estimates Estimates
CONSTANT 10.4671 ** 10.8354 *
(0.0816) (0.2606)
BLACK 0.1503 0.2178 *
(0.1040) (0.1103)
GENDER 0.1372 ** 0.0997
(0.0539) (0.0753
PARTNER -0.2919 ** -0.3031 *
(0.0806) (0.0894)
GENDER*PARTNER 0.0834 0.1510 *
(0.0697) (0.0759)
PTNRGRAD 0.2568 ** 0.2801 *
(0.0733) (0.0778)
PINCLOW -0.2407 ** -0.1884 *
(0.0687) (0.0892)
PINCMED -0.1453 ** -0.0457
(0.0399) (0.0693)
FTHRGRDS -0.0637 * -0.0811 *
(0.0362) (0.0385)
GPARANK 0.1957 ** 0.3130 *
(0.0660) (0.1187)
HINCPOT 0.1782 ** 0.2695 *
(0.0481) (0.0503)
UNIV 0.0895 ** 0.0575
(0.0395) (0.0632)
MASTROTH 0.1201 ** -0.1448
(0.0585) (0.1764)
DOCTROTH -0.0769 -0.6837 *'
(0.0630) (0.3078)
0CCUP2 -0.2485 ** -1.4144 *
(0.0658) (0.8604)
TABLE 4.4 : (Continued)
Independent
Variable:
OLS
Coefficient
Estimates
2SIV
Coefficient
Estimates
0CCUP3
0CCUP4
OCCUP5
0CCUP5
0CCUP7
EXPER
-0.5430 **
(0.0583)
-0.4595 **
(0 .1 0 0 2)
-0.3441 **
(0.0780)
-0.5683 **
(0.0775)
-0.0251
(0 .0609)
0.0617 **
(0.0098)
-0.9562 **
(0.3140)
-0.9978 *
(0.5757)
-1.1946 **
(0.4111)
-3.1675 **
(1.5142)
0.1342
(0.3645)
0.0602 **
(0 .0 1 0 0)
Sample Size 887 887
R-Squared 0.325 0.226
Adj R-Squared 0.310 0.202
SSE 222.29 255.10
SE of Regr. 0.507 0.543
DW 2.019 2.029
SE Adj. Fact. 0.933
Hausman Stat. 71.69
Crit. Value 14.67
(standard errors in parentheses)
** indicates 95% significance.
* indicates 90% significance.
APPENDIX TO CHAPTER 4
178
TABLE B.l
Independent
Variable
CONSTANT
BLACK
PARTNER
PTNRGRAD
PINCBOT2
FTHRGRDS
FTHRSCOLL
HINCPOT
UNIV
MASTROTH
DOCTROTH
OCCUP2
OCCUP3
OCCUP4
OCCUP5
OCCUP6
OCCUP7
EXPER
COEFFICIENT ESTIMATES, INCOME EQUATION
by GENDER
Dependent Variable: Log of Income
Men Women
OLS 2SIV OLS 2SIV
Coef. Coef. Coef. Coef.
Estimates Estimates Estimates Estimates
10.4235 **
(0.1147)
0.1323
(0.1423)
-0.1610
(0.1046)
0.1854 *
(0.1030)
0.1828 **
(0.0520)
0.1753 **
(0.0672)
0.1579 **
(0.0589)
0.1147 *
(0.0614)
-0.0015
(0.0785)
-0.1207
(0.0S26)
-0.4160 **
(0.0790)
-0.6268 *'
(0.0938)
-0.6S45 **
(0.279S)
-0.5260 **
(0.1017)
-0.6551 **
(0.1117)
-0.1578 *
(0.0839)
0.0743 **
(0.0140)
10.3282 **
(0.4256)
0.3057 **
(0.1437)
0.0044
( 0 . 1 1 2 2 )
0.1571
(0.1030)
-0.0392
(0.0916)
0.1691 **
(0.0670)
0.2823 **
(0.0582)
-0.0159
(0.1008)
0.3420 **
(0.2065)
-1.1671 **
(0.3984)
-1.6018
(1.1551)
-1.2079 **
(0 .4712)
-7.9588
(5.7163)
-0.8518 **
(0.4888)
2.9152
(1.8053)
1.1292 **
(0.4266)
0.0845 **
(0.0135)
10.5410 **
(0.0902)
-0.3120 **
(0.1034)
0.2905 **
(0.1025)
-0.1290 **
(0.0565)
-0.1393 **
(0.0533)
0.1790 **
(0.0775)
0.1751 **
(0.0866)
-0.0710
(0.1142)
-0.4707 **
(0.0745)
-0.3014 **
(0.1123)
-0.1148
(0.1171)
-0.4705 **
(0.1055)
0.1482 **
(0.0707)
0.0442 **
(0.0135)
10.7305 **
(0.3439)
-0.3343 **
(0.1073)
0.3183 **
(0.1019)
-0.1241 **
(0.0577)
-0.1433 **
(0.0533)
0.2277 **
(0.0765)
0.0590
(0.1844)
-1.2334
(1.4939)
-0.4705 *
(0.2824)
-0.1492
(0.5871)
-0.3305
(0.3439)
-0.8048
(1.3484)
-0.0125
(0.3609)
0.0342 **
(0.0131)
179
TABLE B.l : (Continued)
Sample Size 428 428 459 459
R-Squared 0 .385 0 .313 0.231 0.113
Adj R-Squared 0.361 0.287 0 . 201 0.087
SSE 91.97 102 .75 122 .99 141.85
SE of Regr. 0.473 0 . 500 0 . 526 0 .565
DW 1. 934 1. 841 2.045 2 . 029
SE Adj. Fact. 0.946 0.931
Hausman Stat. 51.48 24.06
Crit. Value 14.67 14 . 67
(standard errors in parentheses)
** indicates 95% significance.
* indicates 90% significance.
CHAPTER 5
180
SUMMARY
This dissertation contained three separate economic analyses. Each
analysis underlined the importance of the connection between demographic
factors and the level and distribution of economic status.
Chapter 2 examined estimated race differences in the economic status, or
IAE, of baby boomers relative to their parents. An integral element of the IAE
analysis was the relative influence of changes in household demographic and
labor force attributes for baby boomers and their parents. Using IAE, this
study found that both black and white baby boomers have made significant
gains in IAE over their parents at the same age, blacks to a slightly larger
extent. The relative sources of IAE gains differed somewhat across race.
Black baby boomers had stronger relative earnings effects and white baby
boomers had stronger household demographic and labor force participation
effects.
Earnings profiles demonstrated how median earnings of black and white
male baby boomers surpassed that of their male parents. However, gains in
earnings were significantly lower for blacks and whites than gains in IAE. Just
181
as with IAE, percentage gains in earnings of black male baby boomers were
higher than for white male baby boomers.
Changes in household demographic and labor force characteristics have
positively affected the IAE of white baby boomers via changes in household
composition and labor supply; the net effect for blacks is ambiguous. Black
and white baby boomers made similar decreases in the number of own-family
children under 16, thus decreasing average household demand. However,
changes in family-type frequency and household labor force participation,
which improved IAE of white baby boomers, on average blunted IAE gains
among blacks. More specifically, 30% of 25-34 year old black baby boomers
in 1990 were single parents compared to 18% for their parents in 1965, while
only 8% of white baby boomers and 5% of pre-boomers were single parents in
1990 and 1965, respectively.
The changing household role of women facilitated IAE growth o f baby
boomers relative to their parents. As of 1990, female baby boomers were
working more and earning more than their parent counterparts in 1965. There
does not appear to be any substantial race difference in earnings growth or
employment rate increases of female baby boomers; both black and white
female baby boomers made similar increases in employment and experienced
similar growth in earnings over the period.
182
Decomposition of IAE trends disclosed that black baby boomers had
lower IAE in 1990 than would have been the case if they maintained the
demographic and labor force levels of their parents in 1965. To the contrary,
white baby boomers have higher IAE because of their changes in demographic
and labor force characteristics. Even though black and white baby boomers
have higher IAE than their parents, the relative sources of these gains differed
across race. If adjustments in household demographic and labor force factors
are thought of as responses to perceived shortfalls in economic status, then the
relative lack of response by black baby boomers is consistent with their relative
advantage in earnings growth, especially since overall gains in IAE for black
still exceed those of whites.
The second essay in this dissertation established that inequality in the
distribution of IAE increased significantly between 1965 and 1990. The
inequality trend was consistently higher among blacks than whites, but the
black/white ratio of inequality decreased over the period.
Results in chapter 2 displayed that median IAE of both black and white
baby boomers exceeded that of their parents at the same age. One concern
about the rise IAE inequality is the effect it may have on baby boomers at the
lower end of the distribution. Empirical results in this chapter established that
despite the increases in inequality over the period, both black and white baby
boomers experienced substantial gains in IAE at the 25th and 75th percentiles.
183
It was also found that race proportions in the lowest and highest 10% of the
IAE distribution were mostly static. In 1965 and 1990, blacks were
overrepresented in the lowest 10% of IAE and underrepresented in the top 10%
of IAE; the opposite is true for whites. The only significant change in race
representations in these deciles over the period was a decrease in the proportion
of blacks in the lowest 10% of IAE, from 30.6% to 23.1%. Whatever the
effects of rising IAE inequality have been on baby boomers, they have not kept
baby boomers at the 25th and 75th percentiles of IAE from gaining relative to
their parents.
The measurement of equivalent income, measured by IAE in chapter 2,
requires an equivalency scale. This scale represents the assumptions made
regarding household economies of scale and age differences in demand. The
measurement of IAE in chapter 2 (as well as the analysis of inequality in this
chapter 3) relied on the use of an equivalency scale proposed by Fuchs (1986).
Since the relevance of these results relies on the validity of the equivalency
scale specification, the second part of this chapter focused on the effects of
using three alternative equivalency scales. These other equivalency scales are:
a per capita scale, the scale implicit in the official US poverty thresholds, and a
parametric scale.
The levels of IAE from 1965 to 1990 were sensitive to equivalency scale
specification, but the trends were not. The differences in the levels of IAE
184
across equivalency scale are largely attributed to different assumptions about
economies of scale and age differences in demand. The trend in equivalent
income, by any equivalency scale, was steadily increasing over the 1965 to
1990 period. It was shown that the percentage gains in IAE of black and white
baby boomers over their parents was quite similar across equivalency scales.
Only one scale, the per capita scale, gave significantly different results, and
these results pointed to even larger gains than with the other three equivalency
scales.
The effect of equivalency scale specification on the measurement of IAE
inequality was also examined. As with baby boomer IAE gains, three of the
four scales resulted in similar measures of inequality; the single outlying
equivalency scale, the per capita scale, indicated that inequality was higher than
by other scales. This disparity is primarily due to assumptions of no economies
of scale and no age difference in demand. Such an assumption results in
overrepresentation of large families and all families with children among the
poor. This overrepresentation results in a wider distribution of IAE.
Previous results change little when alternative equivalency scales are
employed. More importantly, the conclusions tied to the specific interest in
chapter 2 - the economic status of baby boomers relative to their parents - are
supported by all four of the equivalency scales employed. The analysis of IAE
185
seems mostly insensitive to equivalency scale specification, and the results
pertaining to baby boomer gains, across race, are robust.
The analysis of chapter 4 utilizes an ’elite* sample of college graduates
and examines the explanatory power of micro-level personal and family
background variables on observed differences in short-term income. A central
concern of the essay is the appropriate treatment of occupation choice as an
explanatory variable to income. A logit estimation of occupational choice
indicated that a wide range of personal characteristics significantly affect the
choice of occupation. Subsequent hypothesis tests support the endogeneity of
occupation choice.
Assuming the endogeneity of occupational choice, income is estimated
by a two stage instrumental variables procedure. Results show that
withstanding extreme similarity in the high quantity and quality of education
across individuals, many personal and family background characteristics are
significantly influential to income. Characteristics found to significantly
explain income differences are: (1) being black, (2) gender (if in a marital-type
union).1 1 7 (3) marital-type partner being a college graduate (if in a marital-
type union), (4) parents’ income, (5) father having a graduate degree, (6) GPA
ranking (undergraduate), (7) considering "High income potential" an essential
career characteristic, (8) obtaining a doctorate degree in a field other than arts
ll7Outside of marital-type unions, gender is insignificant.
186
and sciences. (9) predicted occupation type, and (10) years of experience.
Characteristics insignificant to income are: (1) gender (outside marital-type
unions), (2) mother’s education, (3) considering "The opportunity to work for
social change" an essential career characteristic, (4) school type, (5) obtaining
masters degrees, and (6) obtaining doctorate degrees in the arts & sciences.1 1 8
Evidently, the strong effect from obtaining a college degree, and the even
stronger effect of obtaining a COFHE degree, cannot undermine the income
effects coming from variables representing personal and family background
characteristics.
It was shown that treating occupation as exogenous has salient and
important repercussions. Assuming exogeneity of occupation leads to the
incorrect conclusion that certain insignificant characteristics are in fact
significant (e.g. gender (outside of marital-type unions), attending a university
(rather than a liberal arts college), and obtaining masters degrees), and that
some significant characteristics are in fact insignificant (e.g. being black,
gender (within marital-type unions), and possessing doctorate degrees in fields
other than the arts and sciences). Even when the two estimation methods agree
on the statistical significance of a characteristic, the magnitudes of certain
coefficients differ substantially (e.g GPARANK, HINCPOT, and four
occupation variables).
1,sThe square o f experience is also insignificant.
187
The question of endogeneity of occupational choice is clearly non-trivial.
Incorrect specification in this income model leads to the belief that some
personal variable effects are significant when they are not, and vice-versa.
Even when the interpretation of factor significance is robust to endogeneity
assumption, estimates of certain coefficients are quite sensitive.
Two important implications come from this study. First, despite the
strong - and potentially equalizing - effect that a top quality college education
has on short term income, certain personal and family background
characteristics still influence income. Second, assumptions about specific
pathways by which these variables affect income are critical to empirical
conclusions. More specifically, this study shows that when factors which are
influential to occupational choice are assumed to affect income directly (rather
than indirectly through endogenous occupation) coefficient estimates are
inconsistent and biased, and therefore misleading.
CHAPTER 6
188
BIBLIOGRAPHY
Angrist, Joshua, and Alan Krueger, 1991, "Does compulsory school attendance
affect schooling from and earnings?," Quarterly Journal of Economics. 106,
979-1030.
Arrow, Kenneth, 1973, "The Theory of Discrimination," in Discrimination in
Labor Markets, Ashenfelter and Rees eds., Princeton: Princeton University
Press.
Barten, A. P., 1964, "Family Composition, Prices, and Expenditure Patterns," in
P. E. Hart et al. eds., Econometric Analysis for National Economic Planning.
London.
Becker, Gary S., 1960, "The Economic Analysis of Fertility," in Demographic
and Economic Change in Developed Countries. Universities-National Bureau
Conference Series No. 11, Princeton: Princeton University.
__________ , 1965, "Human Capital," NBER. New York.
__________ , 1981, The Treatise on the Family. Cambridge: Harvard University
Press.
Becker, Gary S., and Barry Chiswick, 1966, "Education and the Distribution of
Earnings," American Economic Review. Papers and Proceedings. May, 358-69.
Becker, Gary S., and H. Gregg Lewis, 1973, "On the interaction between the
quantity and quality of children," Journal of Political Economy. 81:2, part 2
(March/April), s279-88.
Becker, William B., 1990, "The Demand for Higher Education," in The
Economics of American Universities. Hoenack and Collins, eds., 158-88,
Albany: The State Univ. of New York Press.
189
Behrman, Jere R., 1988, "Intrahousehold Allocation of Nutrients in Rural India:
Are Boys Favored?, Do Parents Exhibit Inequality Aversion?," Oxford
Economic Papers 40, 32-54.
Behrman, Jere R., Robert A. Pollack, and Paul Taubman, 1982, "Parental
Preferences and the Provision for Progeny," Journal of Political Economy. 90:1,
52-73.
Behrman, Jere R., and Paul Taubman, 1986, "The Effects of Number and
Position of Siblings on Child and Adult Outcomes," Social Bioloev. , 33:1-2, 33-
34.
__________ . 1989, "Is schooling ’mostly in the genes’? Nature-nurture
decomposition with data on relatives," Journal of Political Economy. 97:6,
(December), 1425-46.
Behrman, Jere R., Lori G. Kletzer, Michael S. McPherson, and Morton O.
Schapiro, 1994, "How family background sequentially affects college
educational investments: high school achievement, college enrollment and
college quality choices," mimeo, Dept, o f Economics, University' of
Pennsylvania.
Ben-Porath, Yoram, 1965, "The Production of Human Capital and the Lifecycle
of Earnings," Journal of Political Economy. 75, 352-65.
Berger, Mark C., 1985, "The Effect of Cohort Size on Earnings Growth."
Journal of Political Economy. 93, 561-73.
__________ . 1989, "Demographic Cycles, Cohort Size, and Earnings,"
Demography. 26:2, 311-21.
Berman, Eli. John Bound, and Zvi Griliches, 1994, "Changes in the Demand
for Skilled Labor Within U.S. Manufacturing: Evidence From the Annual
Survey of Manufacturers," Quarterly Journal of Economics. 109:2, 367.97.
Blackburn, McKinley L., 1990, "What Can Explain the Increase in Earnings
Inequality Among Males?," Industrial Relations. 29:3, 441-56.
Blackburn, McKinley L., and David E. Bloom, 1991, "The Distribution of
Family Income: Measuring and Explaining Changes in the 1980’s for Canada
and The United States," NBER Working Paper # 3659. NBER: Cambridge.
190
__________ , 1994, "Changes in the Structure of Family Income Inequality in
the U.S. and Other Industrial Nations During the 1980’s," NBER Working
Paper # 4754. NBER: Cambridge.
Blackburn, McKinley L., David E. Bloom, and Richard E. Freeman, 1990,
"The Declining Economic Position of Less Skilled American Men," in G.
Burtless ed., A Future of Lousv Jobs?. Ch. 2, Washington, D.C.: Brookings
Institute.
__________ , 1993, "Changes in Earnings Differentials in the 1980’s:
Concordance, Convergence, Causes, and Consequences," in D. Papadimitriou
and E. Wolf Eds., Poverty and Prosperity in the USA in the Late 20th Centurv.
Ch. 9, New York: St Martins Press.
Blackorby, Charles, and David Donaldson, 1989, "Adult Equivalence Scales,
Interpersonal Comparisons of Well-Being, and Applied Welfare Economics,"
Discussion Paper No.89-24, University of British Columbia.
Blake, Judith, 1989, Family Size and Achievement, Berkeley: University of
California Press.
Blau. Francine D., and Lawrence M. Kahn, 1994, "Rising Wage Inequality and
the U.S. Gender Gap," American Economic Review. 84:2, 23-8.
Borjas, George J., and Valerie A. Ramey, 1994, "Rising Wage Inequality in the
U.S.: Causes and Consequences," American Economic Review, 84:2, 10-16.
Boskin, Michael J., 1974, "A conditional logit model of occupational choice,"
Journal of Political Economy. 82:2, part 1 (March/April), 389-98.
Bound, John, David A. Jaeger, and Regina Baker, 1993, "The cure can be
worse than the disease: a cautionary tale regarding instrumental variables,"
NBER Technical Working Paper NO. 137. Cambridge, MA: NBER.
Bound, John, and Richard B. Freeman, 1989, "Black Economic Progress:
Erosion of the Post-1965 Gains in the 80’s?," in S. Shulman and W. Darity Jr.
eds., The Question of Discrimination. Racial Inequality in the U.S. Labor
Market. Ch. 3., Middletown, Conn.: Wesleyan Univ. Press.
191
__________ , 1991, "What Went Wrong? The Erosion o f Relative Earnings and
Employment Among Young Black Men in the 1980’s," NBER Working Paper
# 3778. NBER: Cambridge.
Bound, John, and George Johnson, 1992, "Changes in the Structure of Wages
in the 1980’s: An Evaluation of Alternative Explanations," American Economic
Review. 82:3, 371-92.
Bronars, Stephen G., and Jeff Grogger, 1994, "The Economic Consequences of
Unwed Motherhood: Using Twin Births as a Natural Experiment," American
Economic Review. 84:5, 1141-57.
Browning, Martin, Francois Bourguignon, Pierre-Andre Chiappori, and Valerie
Lechene, 1994, "Incomes and Outcomes: A Structural Model of Intrahousehold
Allocation," Journal o f Political Economy. 102:6, 1067-96.
Buhmann, Brigitte, Lee Rainwater, Guenther Schmaus, and Timothy M.
Smeeding, 1988, "Equivalence Scales, Well-Being, Inequality, and Poverty:
Sensitivity Estimates Across Ten Countries Using the Luxembourg Income
Study (LIS) Database," Review of Income and Wealth. 34, 115-142.
Burtless, Gary, 1990, "Earnings Inequality Over the Business and Demographic
Cycles," in G. Burtless ed., A Future of Lousv Jobs?. Ch. 3, Washington, D.C.:
Brookings Institute.
__________ , 1993, "The Contribution of Employment and Hours Changes to
Family Income Inequality," American Economic Review. 83:2, 131-5.
Butler. Richard, and Richard McDonald, 1989, "Interdistributional Income
Inequality," Journal of Business and Economic Statistics. 5, 13-18.
Cain, Glen G., 1986, "The Economic Analysis of Labor Market Discrimination:
A Survey," in The Handbook of Labor Economics, v .l, chapter 13,
Amsterdam: North Holland.
Cancian, Maria, Sheldon Danziger, and Peter Gottschalk, 1993, "The Changing
Contributions of Men and Women to Levels and Distribution of Family
Income, 1968-88," in D. Papadimitriou and E. W olf Eds., Poverty and
Prosperity in the USA in the Late 20th Century. Ch. 9, New York: St Martins
Press.
192
Card, David, and Alan B. Krueger, 1992a, "School quality and black-white
relative earnings: a direct assessment," Quarterly Journal of Econometrics.
107:1, (February), 151-200.
___________. 1992b, "Does School Quality Matter? Returns to Education and
the Characteristics of Public Schools in the U.S.," Journal of Political
Economy. 100:1, 1-40.
__________ , 1993, "The Economic Status of Black Americans: What Can We
Do About it?," American Economic Review. 83:2, 85-91.
Card. David, and Thomas Lemieux, 1994, "Changing Wage Structure and
Black-White Wage Differentials," American Economic Review. 84:2, 29-33.
Davis, Steve J., and John Haltiwanger, 1991, "Wage Dispersion Between and
Within U.S. Manufacturing Plants, 1963-86," Brookings Pap. Econ. Act..
Special Issue, 115-80.
Dooley, M. D., and Peter Gottschalk, 1984, "Earnings Inequality Among Males
in the U.S.: Trends and the Effect of Labor Force Growth." Journal of Political
Economy, 92. 59-89.
Chiswick, Barry, 1988, "Differences in Education and Earnings Across Racial
and Ethnic Groups: Tastes, Discrimination, and Investments in Child Quality,"
Quarterly Journal of Economics. 103:3, 571-97.
Clotfelter, Charles T., and Michael Rothschild, 1993, Studies of Supply and
Demand in Higher Education, Cambridge, MA: NBER.
Connelly, Rachel, 1986, "A Framework for Analyzing the Impact of Cohort
Size on Education and Labor Earnings," Journal of Human Resources. 21:4,
543-62.
Darity, William, 1974, "Illusions of Black Progress," Review of Black Political
Economy, 10, 153-68.
Deaton, Angus, 1988, "The Allocation of Goods within the Household," The
World Bank LSMS Working Papers. No. 39.
Deaton, Angus, and John Muellbauer, 1980, Economics and Consumer
Behavior. Chapter 8, London: Cambridge University Press.
193
Dechter, Aimee, and Pamela Smock, 1994, "The Fading Breadwinner Role and
the Economic Implications for Young Couples," Institute for Research on
Poverty Discussion Paper # 1051-94. Univ. of Wisconsin-Madison.
del Pinal, Jorge H., 1992, "Exploring Alternative Race Ethnic Comparison
Groups in Current Population Surveys," U.S. Department of Commerce.
Current Population Reports. Series P23-182.
Donohue, John J. Ill, and James Heckman, 1991, "Continuous Versus Episodic
Change: The Impact of Civil Rights Policy on the Economic Status of Blacks."
Journal of Economic Literature. 29, 1603-43.
Easterlin, Richard A., 1966, "On the Relation of Economic Factors to Recent
and Projected Fertility Changes," Demography. 3:1, 131-51.
__________ , 1980, Birth and Fortune: The Impact of Numbers on Personal
Welfare. Chicago: University of Chicago Press.
__________ , 1994, "Preferences and Prices in Choice of Career: The Switch to
Business, 1972-87," mimeo, Department of Economics, University of Southern
California.
Easterlin, Richard A., and Eileen M. Crimmins, 1985, The Fertility Revolution:
A Supplv-Demand Analysis. Chicago: University of Chicago Press.
Easterlin, Richard A., Christine Macdonald, and Diane J. Macunovich, 1990,
"Retirement Prospects of the Baby Boom Generation: A Different Perspective,"
The Gerontologist. 30:6, 776-83.
Easterlin, Richard A., Christine M. Schaffer, and Diane J. Macunovich, 1993.
"Will the Baby Boomers Be Less Well Off Than Their Parents? Income,
Wealth, and Family Circumstances Over the Life Cycle in the United States,"
Population and Development Review. 19:3, 497-522.
Ferber, Marianne A., and Carole A. Green, 1991, "Occupational Segregation
and the Earnings Gap," in Essays on the Economics of Discrimination.
Kalamazoo, MI: W. E. Upjohn Institute for Employment Research.
Freeman, Richard B., 1971, The Market for College-trained Manpower.
Cambridge, MA: Harvard Univ. Press.
194
___________, 1976, The Black Elite. New York: Mcgraw Hill.
___________, 1979, "The Effects of Demographic Factors on Age Earnings
Profiles," Journal of Human Resources. 16, 289-318.
___________, 1990, "Employment and Earnings of Disadvantaged Young Men
in a Labor Shortage Economy," NBER Working Paper # 3444. NBER:
Cambridge.
___________, 1994, "Program Report: Labor Studies," NBER Reporter. NBER:
Cambridge.
Freeman, Richard B., and Harry J. Holzer, 1987, "The Black Youth Crisis:
Summary and Findings," in R. Freeman and H. Holzer eds., The Black Youth
Employment Crisis. Ch. 1, Chicago: University of Chicago Press.
Fuchs, Victor R., 1986, "Sex Differences in Economic Well-Being," Science.
22, 459-464.
Gastwirth, Joseph, Tapan Nayak, and Jane-Ling Wang, 1988, "Statistical
Properties of Between Group Income Differentials," Journal of Econometrics.
42:2, 5-19.
Gottschalk, Peter, 1993, "Changes in Inequality of Family Income in Seven
Industrial Countries," American Economic Review. 83:2, 136-42.
Green, William H., 1990, Econometric Analysis. 300-02, New York: Macmillan
Publishers.
Gronau, Reuben, 1982, "Inequality of Family Income: Do Wives’ Earnings
Matter?," Population and Development Review. 8supplement, 119-136.
___________, 1986, "Home Production: A Survey," in The Handbook o f Labor
Economics. 1, chapter 1, Amsterdam: North Holland.
Groshen, Erica L, 1989, "Do Wage Differences Among Employees Last?,"
Working Paper No. 8906. Federal Reserve Bank of Cleveland.
Hanratty, Maria, and Rebecca Blank, 1992, "Down and Out in North America:
Recent Trends in Poverty Rates in the U.S. and Canada," Quarterly Journal of
Economics. 107, 233-54.
195
Hanushek, Eric A., 1981, "Alternative Models of Earnings Determination and
Labor Market Structures," Journal of Human Resources. 238-59.
Hausman, Jerry, 1978, "Specification tests in econometrics," Econometrica. 46,
1251-71.
Haveman, Robert, and Larry Buron, 1993, "Who are the Truly Poor? Patterns
of Official and Net Earnings Capacity Poverty 1973-88," in D. Papadimitriou
and E. Wolf Eds., Poverty and Prosperity in the USA in the Late 20th Centurv.
Ch. 3, New York: St Martins Press.
Heckman, James J., 1980, "Sample selection bias as a specification error," in
Female Labor Supply. James P. Smith ed., Princeton: Princeton University
Press.
__________ , 1989, "The Impact of Government on the Status of Black
Americans," in S. Shulman and W. Darity Jr. eds., The Question of
Discrimination. Racial Inequality in the U.S. Labor Market. Ch. 7.,
Middletown, Conn.: Wesleyan Univ. Press.
Hill. C. R., and F. P. Stafford, 1980, "Parental Care of Children: Time Diary
Estimates of Quantity Predictability and Variety," Journal of Human Resources.
15.
Howell, david R., and Edward N. Wolff, 1991, "Trends in the Growth and
Distribution of Skills in the U.S. Workplace," Industrial and Labor Relations
Review. 43:3, 486-502.
James, Estelle, 1978, "Product Mix and Cost Disaggregation: A Reinterpretation
of the Economics of Higher Education," Journal of Human Resources, 12: 157-
86.
__________ . 1990, "Decision Processes and Priorities in Higher Education," in
The Economics of American Universities. Hoenack and Collins, eds., 77-106,
Albany: The State Univ. of New York Press.
___________, 1986, "Cross-subsidization in Higher Education: Does in Prevent
Private Choice and Public Policy?," in Private Education: Studies in Choice and
Public Policy. D. Levy, ed., 237-57, New York: Oxford Univ. Press.
196
James, Estelle, and Egon Neuberger, 1981, "The University Department as a
Non-profit Labor Cooperative," Public Choice. 36: 585-612.
Juhn, Chinhui, Kevin M. Murphy, and Brooks Pierce, 1993, "Wage Inequality
and the Rise in the Return to Skill," Journal of Political Economy. 101:3, 410-
42.
Kane, Thomas J., 1994, "College Entry by Blacks Since 1970: The Role of
College Costs, Family Background, and Returns to Education," Journal of
Political Economy. 102:5, 878-911.
Karoly, Lynn A., 1992, "The Trend in Inequality Among Families, Individuals,
and Workers in the U.S. A Twenty Five Year Perspective," R-4206-RC. Santa
Monica, CA: Rand Corp.
Katz, Lawrence F., Gary W. Loveman, and David B. Blanchflower, 1993, "A
Comparison of Changes in the Structure of Wages in Four OECD Countries,"
NBER Working Paper # 4297. NBER: Cambridge.
Katz, Lawrence F., and Kevin M. Murphy, 1992, "Changes in Relative Wages,
1965-87," Quarterly Journal of Economics. 107:1, 35-78.
Lazear, Edward, 1979, "The Narrowing of Black-White Wage Differentials is
Illusory," American Economic Review. 69:4, 553-64.
Lazear, Edward, and R. T. Michael, 1988, "Family Size and the Distribution of
Real Per Capita Income," American Economic Review. 70:1, 91-107.
Lehrer, Evelyn, and Marc Nerlove, 1984, "A Lifecycle Analysis of Family
Income Distribution," Economic Inquiry. 22:3, pp.-74.
Leibowitz, Aaron, 1974, "Home Investments in Children," Journal of Political
Economy, pt 2.
Leslie, Larry L., and Paul T. Brinkman, 1987, "Student Price Response in
Higher Education," Journal of Higher Education. 58: 181-203.
Levy, Frank, and Richard J. Murnane, 1993, "U.S. Earnings Levels and
Earnings Inequality: A Review of Recent Trends and Proposed Explanations,"
Journal of Economic Literature. 30:2, 1333-81.
197
Levy, Frank, and Richard C. Michel, 1991, The Economic Future of American
Families. Income and Wealth Trends, Washington, D.C.: Urban Institute.
Lilliard, Lee, 1977, "Inequality: Earnings vs. Human Wealth," American
Economic Review. 67, 42-53.
Low, Stuart A., and Daniel J. Villegas, 1987, "An Alternative Approach to the
Analysis of Wage Differentials," Southern Economic Journal. 54:2, 449-61.
Macunovich. Diane J., 1993, "A Review of Recent Developments in the
Economics of Fertility," Labor and Population Program Working Paper Series
93-06. Santa Monica, CA: Rand Corp.
Macunovich, Diane J., and Richard A. Easterlin, 1990, "How Parents Have
Coped: The Effect of Life Cycle Demographic Decisions on the Economic
Status of Pre-School Age Children, 1964-87," Population and Development
Review. 16:2, 301-25.
Manski, Charles F., 1993, "Adolescent Econometricians: How Do Youth Infer
the Returns to Schooling?," in Studies of Supply and Demand in Higher
Education, eds. Charles T. Clotfelter and Michael Rothschild, 43-57,
Cambridge, MA: NBER.
Manski, C., and David Wise, 1983, College Choice in America. Chapter 6,
Cambridge, MA: Harvard Univ. Press.
Maxwell, Nan, 1987, "Occupational Differences in the Determination of U.S.
Workers Wages: Both the Human Capital and the Structured Labor Market
Hypothesis are Useful in Analysis," American Journal of Economics and
Sociology. 46:4, 431-43.
Mertz, J., T. Garner, T. M. Smeeding, J. Faik, and D. Johnson, 1993, "Two
Scales, One Methodology-Expenditure Based Equivalence Scales for the United
States and Germany," Cross National Studies in Aging Program Project Paper
N. 8. All University Gerontology Center, Syracuse, NY: Syracuse University.
Menchik, P., 1980, "Primogeniture, Equal Sharing and the U.S. Distribution of
Wealth," Quarterly Journal of Economics. March.
198
Mincer, Jacob, 1974, "Schooling. Experience, and Earnings." New York:
Cambridge University Press.
Meyer, Susan E., and Christopher Jencks, 1993, "Recent Trends in Economic
Inequality in the U.S.: Income Versus Expenditures Versus Material Well-
Being," in D. Papadimitriou and E. W olf Eds., Poverty and Prosperity in the
USA in the Late 20th Centurv. Ch. 5, New York: St Martins Press.
Moffltt, Robert A., 1990, "The Distribution of Earnings and the Welfare State."
in G. Burtless ed., A Future of Lousv Jobs?. Ch. 6, Washington, D.C.:
Brookings Institute.
Muellbauer, J., 1977b, "Testing the Barten Model of Household Composition
Effect and the Cost of Children," Economic Journal. 87, 460-87.
Murphy, Kevin, and Robert Topel, 1985, "Estimation and inference in two-step
econometric models," Journal of Business and Economic Statistics. 3, 370-79.
Murphy, Kevin, and Finis Welch, 1989, "Wage Premiums for Recent College
Graduates: Recent Growth and Possible Explanations," Educational Researcher.
18(4): 17-26.
___________, 1993, "Occupational Change and the Demand for Skill, 1940-90,"
American Economic Review. 83:2, 122-6.
Nelson, Richard, and Richard Startz, 1990a, "The distribution of the
instrumental variables estimator and the t-ratios when the instrument is a poor
one," Journal of Business. 63:1.2, sl25-40.
Nerlove, M. and S. Press, 1973, "Univariate and Multivariate Log-Linear and
Logistic Models," R1306-EDA/NIH. Santa Monica, CA: Rand Corp.
Oaxaca, Ronald, 1973, "Male-Female Wage Differentials in Urban Labor
Markets." International Economic Review. October, 693-709.
Orshansky, Mollie, 1963, "Children of the Poor," Social Security Bulletin. 26,
3-13.
___________, 1965, "Counting the Poor: Another Look at the Poverty Profile,"
Social Security Bulletin. 28, 3-29.
199
___________, 1988, "Commentary: The Poverty Measure," Social Security
Bulletin. 51, 1-4.
Person, Torsten, and Guido Tabellini, 1994, "Is Inequality Harmful to
Growth?," American Economic Review. 84:3, 600-21.
Pitt, Mark M., M. R. Rosenzweig, and M. D. Hassan, 1990, "Productivity,
Health, and Inequality in the Intrahousehold Distribution of Food in Low
Income Countries," American Economic Review. December, 1139-56.
Pollack R. A.,and T. J. Wales, 1979, "Welfare Comparisons and Equivalence
Scales," American Economic Review. 69:2, 216-21.
Prais, and Houthakker, 1971, The Analysis of Family Budgets. Cambridge:
Cambridge University Press.
Price, Edward, and Edwin Mills, 1986, "Race and Residence in Earnings
Determination," Journal of Urban Economics. January, 1-18.
Riley, J., 1979, "Testing the educational screening hypothesis," Journal of
Political Economy. 87, s227-52.
Rosenzweig, Mark R., and Kenneth I. Wolpin, 1988, "Heterogeneity,
Intrafamily Distribution, and Child Health," Journal of Human Resources. 23:4,
437-61.
Schapiro, Morton O., Michael P. O’Malley, and Larry H. Litten, 1991,
"Progression to Graduate School from the "Elite" Colleges and Universities,"
Economics of Education Review. 10:3, 277-244.
Schmidt, Peter, and Robert P. Strauss, 1985b, "The Prediction of Occupation
Using Multiple Logit Models," International Economic Review. 16:2, 471-86.
Slesnick, Daniel T., 1994, "Consumption, Needs, and Inequality," International
Economic Review, 35:3, 677-703.
Smith, James P., 1989, "Career Wage Mobility," in S. Shulman and W. Darity
Jr. eds., The Question of Discrimination. Racial Inequality in the U.S. Labor
Market. Ch. 5., Middletown, Conn.: Wesleyan Univ. Press.
2 0 0
Smith, James P., and Finis Welch, 1978, "Race Differences in Earnings: A
Survey and New Evidence," R-2295-NSF. Santa Monica, CA: Rand Corp.
Spence, Michael, 1973, "Job market signalling," Quarterly Journal of
Economics. 87, 355-79.
Sydenstricker, E., and W. I. King, 1921, "The Measurement of Relative
Economic Status of Families," Quarterly Publication of the American Statistical
Association. 17.
Theil, H., 1969, "A Multinomial Extension of the Linear Logit Model,"
International Economic Review. 10, 251-59.
Thomas, Duncan, 1990, "Intra-Elousehold Resource Allocation:An Inferential
Approach," Journal of Human Resources, 25:4, 635-64.
Timmer, S., J. Eccles, and K. O’Brien, 1985, "How Children Use Time," in
Time. Goods, and Well-being. Juster and Stafford eds., Ann Arbor: ISR,
University of Michigan.
Topel, Robert, 1994, "Regional Labor Markets and the Determinants of Wage
Inequality," American Economic Review, 84:2, 17-22.
Triplett, Jack E., 1973, "Consumer Demand and Characteristics of Consumption
Goods," in Household Production and Consumption. Nestor E. Terleckyj ed.,
New York: Columbia University Press.
U.S. Congressional Budget Office, 1993, "Baby Boomers in Retirement: An
Early Perspective," Congressional Budget Office.
U.S. Bureau of the Census. 1994, "Poverty in the United States: 1987," Current
Population Reports. Series P-60, Consumer Income, No. 185, Washington DC:
US Government Printing Office.
Weiss, Andrew A., 1993, "Some aspects of measurement error in a censored
regression model," Journal of Econometrics. 52, 169-88.
Willis, Robert, and Sherwin Rosen, 1979, "Education and self-selection,"
Journal o f Political Economy. 87, (October), s7-36.
2 0 1
Wohlstetter, Albert, and Sinclair Coleman, 1972, "Race Differences in Income,"
Chapter 1 in Anthony H. Pascal ed., Racial Discrimination in Economic Life.
Lexington: Lexington Books.
Wolpin, Kenneth, 1992, "The Determinants in Black-White Differences in
Early Employment Careers: Search, Layoffs, Quits, and Endogenous Wage
Growth," Journal of Political Economy. 100:3, 535-60.
Abstract (if available)
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
PDF
00001.tif
Asset Metadata
Core Title
00001.tif
Tag
OAI-PMH Harvest
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC11256656
Unique identifier
UC11256656
Legacy Identifier
9617116